• Author / Uploaded
• Sergio Cruz Fernández

Citation preview

Contents Title Page Key to Symbols used & Bibliography 4 Publisher’s Foreword 5 Introduction 7 1 Akiba Showing the Way 13 2 Turning Points 29 3 Passive or Active Defence? 59 4 A Bad Plan is Better than No Plan 77 5 Long Games with Increment 95 6 When is the Right Time to Run? 127 7 Choosing the Right Transformations 159 8 Karjakin 181 9 Stalemate 197 10 Stalemated 235 11 The Relevance of Endgame Studies 265 12 Geometry 279 13 Endings with Opposite Coloured Bishops 299 Name Index 318 Game Index 320

2

Technical Decision Making in Chess By

Boris Gelfand with invaluable help from Jacob Aagaard

Quality Chess http://www.qualitychess.co.uk

First edition 2020 by Quality Chess UK Ltd Copyright © 2020 Boris Gelfand All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, electrostatic, magnetic tape, photocopying, recording or otherwise, without prior permission of the publisher. Paperback ISBN 978-1-78483-064-9 Hardcover ISBN 978-1-78483-065-6 All sales or enquiries should be directed to Quality Chess UK Ltd, Suite 247, Central Chambers, 11 Bothwell Street, Glasgow G2 6LY, United Kingdom Phone +44 141 204 2073 e-mail: [email protected] website: www.qualitychess.co.uk Distributed in North and South America by National Book Network Distributed in Rest of the World by Quality Chess UK Ltd through Sunrise Handicrafts, ul. Szarugi 59, 21-002 Marysin, Poland

3

Typeset by Jacob Aagaard Edited by John Shaw Proofread by Andrew Greet Front cover photo by David Llada Back cover photo by Nikolaos Ntirlis Photos on pages 5, 159, 181 and 279 by Maria Emelianova Photos on pages 13, 59, 95, 124, 197 and 235 by Amruta Mokal Photos on pages 297, 299, 302 and 312 by Vladimir Barsky Photo on page 29 from the Gelfand family album Photo on page 77 by Harald Fietz Photo on page 127 by Vladimir Jagr Photo on page 265 by Boris Bukhman Cover design by Colin McNab

4

Key to symbols used ² ³ ± µ +– –+ = © „ ƒ ÷ ? ?? ! !! !? ?! ™ #

White is slightly better Black is slightly better White is better Black is better White has a decisive advantage Black has a decisive advantage equality with compensation with counterplay with an initiative unclear a weak move a blunder a good move an excellent move a move worth considering a move of doubtful value only move mate

Bibliography Aagaard: Excelling at Technical Chess, Everyman Chess 2004 Aagaard: Grandmaster Preparation – Endgame Play, Quality Chess 2014 Avrukh: Grandmaster Repertoire 2 – 1.d4 Volume Two, Quality Chess 2010 De la Villa: 100 Endgames You Must Know, New in Chess 2015 Dvoretsky: Dvoretsky’s Endgame Manual – 4th Edition, Russell Enterprises 2014 Dvoretsky: Dvoretsky’s Endgame Manual – 5th Edition, Russell Enterprises 2020 Gelfand: My Most Memorable Games, Edition Olms 2005 Gelfand: Positional Decision Making in Chess, Quality Chess 2015 Gelfand: Dynamic Decision Making in Chess, Quality Chess 2016 Gelfand: Decision Making in Major Piece Endings, Quality Chess 2020 Marin: Learn from the Legends 3rd Edition, Quality Chess 2015 Mueller & Konoval: Understanding Rook Endgames, Gambit 2016 5

Navara: Můj šachový svět, Pražská šachová společnost 2015 Shankland: Small Steps to Giant Improvement, Quality Chess 2018 Shaw: Quality Chess Puzzle Book, Quality Chess 2010 Shereshevsky: Endgame Strategy, Everyman Chess 1994

6

Publisher’s Foreword This is the third book Boris Gelfand and I have written together. Like the two previous books, Boris and I would debate the games over Skype. I would record the conversations and continue to analyse the games on my own, to see if I found any additional details. I would then return the games to Boris, fully annotated, and he would make further corrections. We then showed many of the examples to our students, who came up with many interesting suggestions. This is the case even more so with Decision Making in Major Piece Endings, which is published simultaneously with this volume. We would like to thank our students for their contributions.

Alex distinctly unimpressed – St Petersburg 2018

The authors would also like to thank Alexander Huzman for his great contribution and discoveries. Throughout the book, including in the index, you will find him referred to either as Huzman or simply as Alex – he deserves to be mentioned twice! 7

This book also includes a small collection of photos. We would like to thank the photographers who allowed us to use their work. Their kindness and friendship is deeply appreciated; credits are given on page 2. At the beginning of every chapter we have placed up to eight diagrams that will give you the chance to “think along” with Boris and the other players. These are not exercises in the traditional sense, so they will not always have a clear and single path to success. If you feel that the later text about these positions does not answer all your questions, this is an excellent chance to go deeper and analyse the position for yourself to learn even more. These chances should never be missed. Endgame books are usually not big sellers, and there are many club players who find endgames boring. I tend to believe this is because they are struggling to understand what to look for in them. My general thinking is that everything in life is interesting if you dive deep into the subject. Maybe with the exception of cleaning... We hope that this book will spread the fascination we feel for this area of chess, which in our opinion is no less rich than any of the others. It just takes a bit of extra effort to access. This book and Decision Making in Major Piece Endings have been a long time in the making. The first game in this book was originally analysed by us in 2014. There are many reasons for this. First of all, we are both busy with many other commitments. Then these books were particularly challenging to write. The analysis of the endings was at times excruciating. And then this analysis had to be presented in a context where it made sense to the reader. (Hopefully, we did not completely fail on that part). But the main obstacle was this co-author’s entirely. This seems like a good moment to thank Boris for his support, understanding and patience over the last few years. I hope the books were worth the wait. Jacob Aagaard Glasgow, August 2020

8

Introduction I have always liked endgames. When I was young I was especially fond of rook endings and studied them over several periods of my childhood, first with Eduard Zelkind, who was my trainer from age 6 to 11. He was a very strong local player at the time, but did not have opportunities to play in strong tournaments. Thus, he reached his peak rating of 2325 in 1996, at the age of 54, by which time he was living in the US. I have many vivid memories of Zelkind explaining various rook endgame positions to me. For example, with three pawns each on the kingside and an a-pawn. Also, he showed me many pawn endgames. Most of which I cannot remember accurately, of course, but there are still small images in my head, as of the end of the following game: Robert Wade – Viktor Korchnoi Buenos Aires 1960

White won the game after 38.a5!. I remembered it as if Wade had missed the win, but the position I recognized immediately. The inspiration from my favourite player, Akiba Rubinstein’s handling of the endgame is obvious and has stayed strong with me for more than four decades. We will talk a bit about this in Chapter 1 where we shall see one of his less famous games. 9

This is the third volume in this series on decision making, with Positional Decision Making in Chess and Dynamic Decision Making in Chess preceding it and with the fourth volume, Decision Making in Major Piece Endings, accompanying it. The goal of this series is to give the improving player, as well as the average chess fan, a look into the decision process of a grandmaster. In this book we shall cover positions of a technical nature. This is a somewhat obtuse term, so we have decided to bend it according to the material, rather than stick too rigidly to one definition. However, a definition we do have, which is: positions where the main goal is the conversion of a static advantage. (A static or long-term advantage can be anything from weaknesses to better pieces to an actual material advantage.) The flip side is included in this, meaning when it is the opponent who is trying to convert an advantage and we are trying to resist. There is often a focus on the endgame when talking about technical play, but this is not necessarily the case when discussing a technical approach to a position. Although we mainly have focused on the endgame in these books. Where we have allowed ourselves to bend the theme and title a bit are in situations where dynamics is used as a defence against technical play. For example, in the chapters on stalemate. I want to state clearly to any potential and actual readers that this book is not an instructional manual. The goal of it is not to deliver a general theory of technical play. There are plenty of good books published that cover this subject. My own favourites include Dvoretsky’s Endgame Manual by Mark Dvoretsky and Endgame Strategy by Mikhail Shereshevsky, but they are by no means the only good books out there. Other popular books include Excelling at Technical Chess and Grandmaster Preparation – Endgame Play by my coauthor Jacob Aagaard, and a number of books by the always interesting German endgame specialist, Karsten Mueller. I specifically found some positions from Understanding Rook Endgames, co-written with Yakov Konoval, fascinating. I know that a lot of other players have found 100 Endgames You Must Know by Jesus de la Villa helpful. From the kindness people have shown me based on the first two books of this series, I have grown to appreciate the wisdom of the reader and am happy to pass on this recommendation, even though I have not personally looked at this book yet. I strongly recommend that the reader consult these volumes and choose his own selection of positions and ideas to memorize. Relying on a single source will be risky and none of the books eclipse all of the others. While we have allowed the material to dictate the structure of the book, rather than finding material that fits in with chosen themes, it still makes sense to me to go through some of the main themes of this book to ensure that the reader will recognize them when he encounters them later.

Decision making vs analysis There are many ways of improving in chess. The most popular ones are working on openings and solving exercises. Far less popular, but equally important, is analysing games to understand them on a 10

deep level. It is especially important to do this with your own games. How are we supposed to improve our decision making if we do not understand when it was flawed? Still, it is my impression that a lot of young players, many of them now rated higher than me, will maybe just look at the evaluations of the engine briefly after the game, to see if they missed anything of importance. It seems to me that computers have made deep analysis easier – and at the same time less likely to happen. When I was growing up we had to find everything on our own and would analyse our games at length with our coaches, opponents and friends. This book is about decision making, but also about the depth of the game. Some of the games have been analysed to such a depth that we should have avoided most serious mistakes in the analysis. In this connection, I want to say that when we are analysing endgames it makes sense to give final conclusions far more than in the middlegame. A higher level of certainty will naturally exist. This will also mean that the chances of making a wildly wrong assessment are greater. We have decided to speak our mind and not fear making mistakes. If you find out that we have, please tell us, and future editions, should there be any, can include the corrections. At the same time, I do not wish the reader to misunderstand the depth and richness of the analysis in this book. There is a time for calculating long variations, but for most of the game I rely a lot on intuition, with a little bit of calculation. Brute-force calculation only works in positions where you can actually calculate all the way to the end. When we analyse the game later, it is a different process though. We try to reach the absolute truth about a position or a game in order to both understand the position and to speculate on how we could best have reached the same conclusion and made the best decisions during the game. This will naturally include some thinking – which is good. Chess is all about thinking. Thinking about thinking is a good thing.

Some of the themes of this book Although this book is not a manual, a lot of the chapters have a thematic dimension. But they are not the usual themes you will find in other endgame books. While stalemate (Chapters 9-10) is a big theme in this book, in other books it is usually just an extra in the big drama. The same with the discussion of playing long endgames on increment, which is the theme of Chapter 5. The other themes you can find on the Contents page, but there are a few themes that run through this book that I would like to quickly discuss.

What are you playing for? This is a specific theme in Chapter 2, with the discussion of turning points in the games. But it is an underlying theme in chess in general and will be something we will discuss from time to time. In chess language we talk about having an advantage as if it is an objective thing. Yet we all know that if we analyse till the end, the result is a draw or a win for either side, with perfect play. 11

But chess is not about perfect play, but about playing better than the opponent. Jacob told me that one of the common phrases I use is “everyone would choose White here”, which I apparently say quite often when the engines are claiming a position is equal. Most positions are objectively equal, but require one player to solve more practical problems than others. This is a theme you will find going through this book.

Defending and converting are two different skills Some players are famed defenders, while others are known for their ability to outplay their competition in the endgame. Rarely are any players good at both. This is an observation by Lev Psakhis, which I think is accurate. Chess is drawn with quite a wide margin, so winning a game against a strong well-prepared opponent is difficult. You will find some players who have a poor track record of converting their superior positions, while at the same time they are difficult to beat, because defence is their primary skill. No matter how fascinating the games will be from such players, people who don’t care about the games will always find a way to describe these players as “boring”. Hopefully, you will take a second look before you buy into this nonsense. When you have a slightly better position, there are a number of strategies. You can try to play natural moves and see if the opponent makes a mistake that will give you additional chances. This can at times work against even the strongest players in the world, but most often it is a safety-first approach with a high likelihood of drawing the game. Another strategy to try to win is changing the character of the game. If the position is in your favour, your opponent’s problems will also be more challenging if there are complications. On the other hand, if I have a winning position, generally I will be looking for clarity. To me it is always better to win without calculation and complications, if possible. I am sure that most of this is not news for the informed reader. So, let’s talk about something that might be:

The zone of one mistake It is very difficult to outplay a top grandmaster. One of the ways I try to do this is to transform a slight advantage into a situation where my opponent is living on the edge. Every mistake may be his last one. I learned this attitude from Psakhis, who calls it the zone of one mistake. There is not much more to it really, but it is a strategy that can be found throughout the book. If your position is objectively “fine” (0.00 for the benefit of youngsters), but you are in danger of losing the game on every move, your position is not “good”.

Timing 12

Timing is a big theme in this book. When should we try to execute the advantage, when should we build? When should we defend passively and when should we try to break free? I could try to give you some general advice on timing, but I fear it will either be stating the obvious or confusing and misleading. There are positions where action is obviously called for, as disaster is approaching. And there are positions where waiting is the only choice and action will lead to immediate defeat. Then there are the grey areas, where we have to use all our powers to choose the right path. Here general guidance will be so vague that giving it may cost me some of your respect. I could say that first you have to build up the position to the maximum, then strike. When you are in complete control, this is sort of obvious. But we rarely are. Still, it is the one point that is so clear, I should make it. Jacob showed me the following position where a number of his GM-students had chosen an inferior path, as indeed the player did in the game. Semen Khanin – Changren Dai Shanghai 2019

In the game White played 62.Nb4, which was also the choice of most of Jacob’s students. White does manage to win both of the black pawns for the a-pawn and the endgame is indeed winning. However, chess is difficult and later on the evaluation went back and forth before the game ended in a draw on move 119. It would have been far more practical to play 62.a7!, leave the queenside alone and reroute the king to g4 before bringing the knight over to win the pawns. With extra tempos, White’s win is easier.

13

The clear cases are few and far between. My most genuine advice for you is to understand this fact first and then to develop your intuition for timing by considering it in your analysis of games; and when useful, at the board. Being aware of a dilemma is only the first step towards solving it, but it is an all-important first step. There will be some who have not thought about this before. I hope this book will be a good starting point for you.

Time controls I am a great fan of the classical time controls, but also no purist. I have played in rapid tournaments, in blitz tournaments, in blindfold tournaments. I enjoy all of them. Some formats favour a certain approach to the game, while others favour others. Risk takers will be more frequently rewarded in blitz and rapid, which also favours those with a more intuitive approach to the game. Classical chess favours those who can think deeply. Whatever your fancy, there is one point I have to make. Learning to play chess well requires an investment of time. There are moves that I will make in a second in a simul that once took me a long time to work out. Without investing this time when I was young, I would not be able to evaluate the positions as quickly as I can now.

A quick note about learning from computers No writing of a chess book in the 21st century can happen without the assistance of chess engines. This is a theme that will be under the surface throughout the book. At this point I just want to say that any active work with the engine, where you are probing, analysing, asking questions, examining and so on, is useful. Any passive submission to the engine evaluations is likely to make you a worse player. Our working method for this book has included a lot of human thinking. The key question for us has not been which line wins, but why the line is winning. The intention has at all times been to understand chess better. But in this connection it is important to understand that chess is not an abstract exercise, but a concrete game. There are a lot of variations in this book and they are included because it is my opinion that these variations are instructive. A lot of other variations were found and discarded. Because they were not. We have tried to include only the things that are interesting or beautiful. I hope this has been reasonably successful. On top of this, chess is a game. When we analyse a game, we may find that objectively one move is stronger than another, but at the same time we understand the challenges a practical player would be met with over the board when the engine guidance is unavailable. I hope this book and Decision Making in Major Piece Endings will help you with your progress in chess. Boris Gelfand 14

Rishon Le Zion, August 2020

15

Chapter 1 Akiba Showing the Way

Always walking in Akiba’s footsteps...

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

16

Choose your recapture (see page 18)

How to organize the pieces? (see page 18)

17

The hardest position in this chapter. Think deeply. (see page 19)

White is under pressure, but can still hold. How? (see page 21)

18

Be accurate! (see page 22)

What is the thematic move? (see page 27) The previous two books in this series have discussed some of my childhood heroes: specifically Akiba Rubinstein, whom we talked a lot about in Positional Decision Making in Chess; and Tigran Petrosian, whom I had the great fortune to be trained by as a young boy in a few camps – there is more about this in Dynamic Decision Making in Chess. It would of course be possible to come up with a new hero for each volume. In Decision Making in Major Piece Endings, released at the same 19

time as this book, I talk about the influence from Korchnoi, Botvinnik and more of our great predecessors. Any grandmaster will have learned a lot from a lot of people over the years, but usually we only have a very limited set of role models. I have worked a lot with Kramnik and Aronian, and played many times against Anand, Kasparov, Carlsen and other great players of our time, and tried to learn as much as possible from everyone. But all of this interaction was probably too late to determine my underlying style. The early work with my trainers and the influence from Rubinstein is the bedrock upon which my chess personality was built. There have been many extensions to my style and skills over the years, but I have always seen everything through the effect these early influences had on me. Anything else would have been impossible. Akiba’s rook endgame skills are widely celebrated. Both Marin and Dvoretsky have written about them in Learn from the Legends and Dvoretsky’s Endgame Manual. But Rubinstein was strong in all technical positions. Rubinstein’s opponent in the following game was a highly talented Czech grandmaster with a fondness for trying to attack the centre with pieces rather than to occupy it with pawns. It is not a famous game, but it fits perfectly into what I am trying to do with this book: which is to discuss the problems we encounter in technical positions and offer you various strategies to deal with them. Richard Reti – Akiba Rubinstein Gothenburg 1920 1.e4 c5 2.Nf3 Nf6 This provocative move was brought into fashion by Nimzowitsch. Not all his ideas were equally good. 3.e5 Nd5 4.Nc3 e6

20

5.g3 Today we know that this is not challenging at all. White should play 5.Ne4!? or 5.Nxd5 exd5 6.d4 with an advantage. 5...Nc6 6.Bg2 Nxc3 7.bxc3 This is the same pawn structure as Cohn – Rubinstein, which can be found on page 137 in Positional Decision Making in Chess. Now all Black needs to do is to exchange all the pieces and win the pawn ending! 7...d6 8.exd6 Bxd6 9.0-0 0-0

21

10.d3 True to his style, but not true to the needs of the position. After the normal move 10.d4 White has a harmonious position. 10...Be7 This is rather artificial. It would make sense to focus on developing the c8-bishop first. I would prefer 10...b6 and ...Bb7, but also possible is 10...Qc7 or 10...Bd7. Black can always play ...Be7 afterwards. 11.Be3 11.Rb1! was a fraction better. 11...Bd7 12.Qd2 12.Nd2 seems more natural. 12...Qc7 13.Rfe1 Rad8 Black has a very pleasant position. If he had time to play ...e5 on the next move, he might even be better. Reti made a reasonable decision to exchange the bishops to get more space for his pieces. 14.Bf4

22

14...Bd6 The exchange of the bishops is not in Black’s interest. After 14...Qc8!, with the idea ...f7-f6, White would have problems finding a good place for the dark-squared bishop. 15.Bxd6 Qxd6 16.Qe3 b6 The position out of the opening is marginally better for Black. In what follows it is inspirational to see how Rubinstein is able to outplay his opponent without doing “anything special”. One good move after another. 17.Nd2 17.a4! was better, waiting to see what Black was going to do. 17...Ne7 18.Nc4 Qc7

23

19.Qe5? Remarkably, this is a serious mistake. After 19.a4 the position would remain close to equal, though I think everyone would take the black pieces. 19...Qxe5 20.Nxe5 Ba4! Rubinstein immediately puts pressure on White’s main weakness. 21.Re2 Nd5 And then puts pressure on his opponent to let go of his best piece.

24

22.Bxd5 Another concession. Getting the opponent to make concessions is an important strategy for winning games. Often this happen because we have seen things they have not. At this point you could easily imagine that Reti had not fully grasped that the black bishop would be a thorn in his side on a4. Although it looks oddly placed there, it is more effective than the bishop on g2, which shoots into thin air. Once the position has become a little bit unpleasant, White suddenly has to make a lot of decisions where he is choosing between two or more evils. 22.c4 Nb4 23.Rc1 f6 23...Nxa2? 24.Ra1 Nc3 25.Ree1 would give White the advantage. Black’s pieces are in a lot of trouble. 24.Nf3

25

24...Bxc2! The simplest way to ensure a long-term advantage. 24...Kf7!? with an advantage is also good. 25.Rexc2 Nxc2 26.Rxc2 Rxd3 Black has serious winning chances. Bishop and knight are said to be worth the same as a rook and one and a half pawns. If you follow this guideline, Black is half a pawn up. On top of this, the white pawns are split on the queenside and Black is more active. A key strategy for Black will be to exchange the white rook, after which his advantage will be obvious. The best defence was active defence: 22.Re4! Be8! 22...Bxc2? 23.Rc1 Nxc3 24.Ree1 Nxa2 25.Rxc2 Nb4 26.Rc3 would give White the advantage. 23.c4 Nb4 24.c3 24.Re2 Ba4 would transpose to the 22.c4 line. Black has to choose between different types of advantage. 24...Nc2 More human and thus in the style of Rubinstein was 24...Nc6 25.Nxc6 Bxc6 26.Re3 Ba4!, when White is passive and has weaknesses, but also where Black will have to do a lot of manoeuvring to exploit them. 25.Rc1 Ba4

26

26.Re2 Na3 27.Re3 Rd6 Black is better, but there is a lot to play for. Both sides have inconveniently-placed pieces.

22...exd5! An important move. Black is better because of the c2-pawn and because his bishop is superior to the knight. In order to exploit these factors to the maximum, Rubinstein chooses the best pawn structure, where the knight will find no stability and where Black can exchange the rooks in order to increase the advantage. 23.Rae1 Rather automatically played. 27

23.d4 would have given Black a chance to put pressure down the c-line with 23...Rc8. After: 24.Rc1 Bb5 25.Re3 Rfe8 26.Ng4

Black has a large advantage. After 26...Re6! White will not be able to improve his position as much and as fast as Black can improve his. Next the king comes into the game. And after 27.Rxe6 fxe6 28.Ne3 cxd4 29.cxd4 Rc3 White’s position is desperately awful. White should have played 23.Kf1, making use of the king. 23...Rfe8 24.f4 24.Ng4 Rxe2 25.Rxe2 d4! would see the c2-pawn fixed as a weakness and the knight rather poorly placed on g4. 24...f6 25.Nf3 Kf7 26.Kf2 Rxe2† 27.Rxe2 Re8

28

According to plan, Black is exchanging the rooks, clarifying his advantage. This is a common technical idea. By exchanging the rooks, active counterplay becomes far less likely. Traditionally rook and bishop is considered better than rook and knight, but every endgame is slightly different. In this case, the position has a semi-closed nature and White has weaknesses on both flanks. By exchanging the rook, White will find defending c2 more troublesome, while Black will be able to use his king actively. 28.Rxe8 If White had avoided the exchange of rooks with 28.Rd2, then Black would have reacted best with 28...Bd7!, which seems to have been a concept Rubinstein did not grasp. Then 29.Rd1 Ke7 with the idea of transferring the king to the queenside. White would not be able to do the same thing, as after 30.Ke3 Bg4 the king would have to stay on the kingside. 28...Kxe8 29.Ne1 Ke7 30.Ke3 White cannot play 30.g4 at this point because of: 30...d4!

29

31.cxd4 cxd4 32.Nf3 Bxc2 33.Ke2 Ba4 34.Nxd4 Bd7 This should be winning in practice for Black. The bishop is stronger than the knight, and Black has the potential to create a passed pawn on the queenside.

30...Ke6? The only bad move Rubinstein made in the game, allowing White to offer resistance. 30...Bd7! was the strongest move, fixing the white pawns on the dark squares. They are not in danger from the bishop on those squares, but they are also unable to prevent the black king from penetrating the position, which is the greatest problem White is facing. 31.g4! 30

But now this is a great resource. White is managing to block Black’s access to the kingside and make it much harder for him to convert his advantage. As we shall see below, White could hold the draw with accurate defence. 31...Kd6 32.h3 g6 33.Kd2 Bd7

34.Nf3 At this point White could have improved his defensive resources in the following surprising way: 34.Ng2!

The idea is simple. White wants to play d3-d4 and Ne3, when he has managed to plug the holes on the light squares on the kingside to a significant degree. 31

The critical move seems to be: 34...d4 The simplifications after 34...h5

35.Nh4! hxg4 36.hxg4 Bxg4 37.Nxg6 would seriously ease White’s defensive burden. 35.cxd4 cxd4 36.c4

36...dxc3† A logical move, giving a pawn majority on the queenside. But as we shall see, White is not without defensive resources. After 36...Kc5 37.Ne1, it does not appear that Black has a serious way to break through. Both 37...b5 38.cxb5 Bxb5 39.h4 and 37...h5 38.gxh5 gxh5 39.h4 followed by Nc2 appear to just hold.

32

37.Kxc3 g5 38.f5 Taking on g5 also makes sense.

38...Bc6 Black cannot penetrate on the kingside: 38...Ke5 39.Ne3 Bc6 40.Nc4†! Kf4 41.Nd6 and it is time to hit the brakes for Black, as 41...Kg3? loses to 42.Ne4†! Bxe4 43.dxe4 Kf4 44.Kd4 b5 45.e5 fxe5† 46.Kd5 and the f-pawn queens with check. 39.Ne3 Bf3 40.d4 Kc6 41.a3 White draws. His position looks scary, but Black has no way to break through. 34...Ke7 Another quite logical plan for Black would be to place the pawns on the dark squares, in the hope that the pawns on the light squares later would become targets. As in the game, Black has the advantage, but breaking through the defence is not easy. 34...h6 35.Ke3 g5

33

The following variations should not be seen as forced in any way, but as an illustration of White’s resources. 36.Ne1 Kc6 37.Nf3 Kb5 38.f5 Ka4

39.c4! dxc4 40.Nd2 Preparing to attack the f6-pawn. 40...cxd3 40...Bc6 41.Ne4 cxd3 42.Kxd3! c4† 43.Kd4 Bxe4 44.Kxe4 Ka3 45.Kd5 c3 46.Ke6 Kb2 47.Kxf6 Kxc2 48.Kg7 Kb1 49.f6 c2 50.f7 c1=Q 51.f8=Q Qc6 52.a4 and Black has no advantage at all. 41.Kxd3 Bb5† 42.Ke3 Kb4 43.Ne4 Bc4 34

Black has made a lot of progress and would reach a winning position after 44.Nxf6. But White has a fabulous resource at his disposal.

44.a4!! This move is very surprising and speaks to the richness of chess. 44.Nxf6 Kc3 45.Ne4† Kxc2 46.f6 Bxa2 and Black wins. Black also wins after 44.c3† Ka3 45.Nxf6 Kxa2 46.Ke4 Bf7!. 44...Kxa4 Black can also go wrong here. After 44...a5 45.Nxf6 b5? 46.axb5 Kxb5 47.Nd7! a4 48.Kd2 it is White who wins. 45.Nxf6 Ka3 46.Nd7 White has managed to create counterplay, but still has practical problems to solve. 46...Kb2 47.Kd2 Bd5 47...a5? 48.Nxb6 48.f6 b5 49.Nxc5 a5 50.Ne4 a4 51.Nc3 Bf7 52.Nxb5 a3

35

53.Nc3! 53.Nxa3 Kxa3 would allow the black king to get back in time. 53...a2 54.Nxa2 Bxa2 55.Kd3 Ka3 56.c4 Kb4 57.Ke4 Bxc4

58.f7! 58.Kf5?? Bf7 and Black wins. 58...Bxf7 59.Kf5 The black pawns will disappear and a draw will be agreed. 35.Ke3 h5

36

36.Nh2? An awful move. The knight may be dim on the rim, but it is not as bad as being boxed in.

XIIIY 9-+-0 9+n+0 9-+-0 xiiiy Animal Cruelty Reti misses probably the clearest chance to draw the game; and potentially the last. 36.Nh4! Fighting for the light squares is the way to hold. 36...Kf7 36...g5 37.Nf3 will lead to exchanges on the kingside, which are in White’s favour. A draw should ensue quickly. 36...hxg4 37.hxg4 Bxg4 38.Nxg6† also helps White. 37.gxh5! The idea behind this move is rather ingenious. 37.f5 g5 38.Nf3 h4 is uncomfortable for White. As we have seen above, there are many times where he can hold this endgame, but he is certainly under a lot of pressure. 37...gxh5 37

38.f5 A pawn sacrifice is coming. 38...Ke7 39.Ng6† Kd6 40.h4! 40.Nf4? h4 and Black will win a pawn. 40...Bxf5 41.Nf4 Bg4

42.d4! White has created a fortress. Black has no reasonable way to make progress, as anything active would cost him a pawn. 36...Kd6 38

37.Ke2? This allows Black’s most desired positional idea. White had a lot of ways to fight back. 37.gxh5 gxh5 38.h4 This is very logical, but does not appear to hold.

38...Bh3! One of the main themes of the endgame is domination. Here the white knight is prevented from coming to g3. 38...Ba4? would allow White to play 39.Nf1! Bxc2 40.Kd2 Ba4 41.Ng3 Be8 and it is not obvious that Black will be able to break through after 42.d4 Kc6 43.Kc2 Kb5 44.Kb3. 39

39.Nf3 Ke6! 39...b5?! 40.Nd2 would give White more defensive potential. 40...Bg4?! 41.Nf1! for example and White manages to plug the holes in his position. After the text move, it is clear that White suffers from a lack of flexibility and options. Black now threatens to bring the king to g4, forcing the knight to stay on the kingside while Black improves his position on the queenside. 40.Nh2 b5 41.Kf2 White cannot block the position on the dark squares. For example: 41.a3 a5 42.d4 cxd4† 43.cxd4 Kf5 44.c3 Bg2 The knight is totally dominated. 45.Kf2 Bh1 46.Ke3 Be4 47.Nf1 Kg4 Black wins. 41...Kd6 42.Ke3 a5 43.Nf3 Ke6

This shuffling around gives Black time to improve on the queenside. 44.Nh2 a4 45.Kf2 45.a3 Kf5, with the idea ...b4, wins quickly. 45...a3! Fixing the a2-pawn. 46.Ke3 Kf5 47.Kf3

40

47...Kg6 This looks artificial, but when you see the idea behind it, to bring the bishop to e6, while keeping the option of bringing in the black king via f5-g4, you hopefully appreciate it. 48.Ke3 d4† 49.cxd4 Be6 50.Kd2 Bxa2 51.Nf1 51.dxc5 Bd5 52.Kc1 Kf5 53.Nf1 Kxf4 54.Nd2 f5 and Black wins with his two distant passed pawns. 51...Be6 52.Kc1 cxd4 53.Ng3

This blockade is obviously not going to work. 53...Bg4 54.Kb1 b4 55.Ka2 Kf7 56.Kb1 Ke6 57.Ka2 Bf3 58.Kb1 Kd5 59.Ka2 Kc5 60.Kb1

41

60...b3 Black breaks through. 61.cxb3 Kb4 62.Ka2 Bd1 It is all over. 37.Kf3 would allow the thematic 37...d4!, which should be decisive. For example: 38.cxd4 cxd4

39.h4 (39.f5 hxg4† 40.Nxg4 Bxf5 41.Nxf6 Bxh3 42.Ke4 Kc5 43.Ke5 Bf5 and Black will win after something like 44.Ne4† Kb4 45.Kxd4 Ka3 46.Nc3 Kb2, for example.) 39...Bc6† 40.Kg3 hxg4 41.Nxg4 Ke6 42.Nh2

42

42...f5! (42...Ba4 43.Nf3 Kd5 44.f5! and it would be White who was playing for a win) 43.Kf2 Kd6 44.Nf1 Kc5 45.Ng3 a5 Black will win on the queenside. Another idea that looked difficult to overcome was 37.d4 Kc6 38.gxh5 gxh5 39.h4, but it seems that Black will break through after:

39...Bf5 40.Kd2 Bh3! Preventing the knight from going from f1 to g3. 41.Nf3 Kd6 42.Ke3 Ke6 43.Nh2 a5 44.a3 Kd6 45.Nf3 Bf5 46.Kd2

43

46...Ke6 47.Nh2 Bg6 White will have to give up the c2-pawn in order to prevent the black king from penetrating. Then his position will lose slowly. For this reason, White should have tried 37.a3!, when Black has a big advantage, but no clear win seems to be available. But to believe that White would have held the game is a different matter. Chances are low. 37...d4! Fixing the weaknesses on a2 and c2, while opening up the position for the bishop and king. 38.cxd4 It was a bit more practical to try to keep the position closed, but Black should win all the same: 38.c4 Ke7

44

There are two lines here that we should look at. a) 39.Kd2 Bc6 40.Ke2 Bg2!

This forces White to make further concessions, after which there is no way to stop the penetration of the black forces. 41.gxh5 gxh5 42.h4 Ke6 43.Kf2 Bc6 44.Nf3 Ba4 45.Ne1 Kf5 46.Kg3 a6 47.a3 Kg6 followed by ...Bd7-g4 and Black wins. b) 39.Ke1 This looks more flexible. The king waits for the bishop to decide where to go. Instead Black creates a path for the king to penetrate:

45

39...hxg4 40.hxg4 g5!

41.f5 White has no choice. 41...Bc6 42.Ke2 Kd6 43.Nf3 a6 Asking White to decide. He cannot allow ...Ke5. 44.Kf2 Ba4 45.Ne1 Ke5 46.Kg3 b5

47.cxb5 47.Nf3† Kd6 48.Ne1 is the other try. After 48...a5 49.cxb5 (49.Kf3 Ke5 50.Kg3 bxc4 51.dxc4 Ke4 and Black wins) 49...Bxb5 50.Kf3 c4 Black will successfully open up the position and the white weaknesses will count. 51.a3 Ke5 52.Kg3 cxd3 53.cxd3 Kd5 Everything has to be well46

timed. Black cannot allow Nf3-d2-e4, with reasonable drawing chances. 54.Kf3 Ba4 55.Ke2 Ke5! 56.Nf3† Kf4 57.Nxd4 Kxg4 58.Ke3 Kh3 and Black wins. 47...axb5 48.a3 b4

Pawn exchanges should generally help the defender, but the penetration of the bishop is a high enough reward. 49.axb4 cxb4 50.Kf3 b3 51.cxb3 Bxb3 52.Ke2 Kf4 53.Nf3 Bd1† And Black wins. 38...cxd4

39.Kd2 47

White has a few desperate tries left he could have wielded, but none of them will work. For example: 39.f5!? hxg4 40.fxg6 gxh3 and Black is in control. 41.Kf2 Bc6 42.Kg3 Ke7 43.Kxh3 Ba4! and Black will win a pawn and the game. 39...hxg4 40.hxg4 Bc6 41.Ke2 Bd5 42.a3 b5 43.Nf1 a5 44.Nd2

44...a4! Very thematic. The distant passed pawn decides the game. 45.Ne4† Or 45.Nb1 Kc5 46.c3 Ba2 and Black wins. 45...Bxe4 46.dxe4 b4 47.Kd2 bxa3 48.Kc1 g5 0–1 Studying the games of Akiba Rubinstein should be a must for any young players who want to develop their technical play and push it to a higher level.

48

Chapter 2 Turning Points

With Albert Kapengut, my mother Nella Gelfand and Ivanchuk at Tilburg 1990

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

49

White has a deep way to play for an advantage (see page 42)

White is under pressure, but has a nice little move to stay in the game (see page 44)

50

Black to play and win (see page 47)

White can hold the draw only with very accurate play (see page 48)

51

Black has a strong resource here. Can you find it? (see page 51)

This should be easy. Black to win. (see page 52)

52

How does White hold? Do not miss Black’s ideas! (see page 54)

My opponent found a nice way to win the game here (see page 55) It is highly useful to develop a feeling for turning points in chess. These are the moments where a player who was previously pressing loses his advantage. At this point it is important that we accept that things have gone wrong and we have to play an equal position. The game is not over and we can still outplay our opponent if we play according to the needs of the position. However, there are also the days when things are really not working for us. We can see from our play that it is not our day, 53

and if we hit the brakes and pack up now, we may leave the playing hall with our dignity intact, and more importantly, half a point. To illustrate this theme, I have chosen two games from my own practice. In the first game, I am at first on the defending side, then in a quick turnaround, I go from worse to almost winning in five moves. In the second game, I made an early mistake in the opening, squandering an otherwise promising position. I was then unable to decide if I should look for an advantage or if I should play more cautiously. I kept pressing, slowly drifting into difficulties. My evaluation of the position was clouded by ambition. Being slow to accept reality simply means an encounter with its consequences. The resulting bishop ending was very interesting, but please do not allow this to distract you from the main narrative and the lessons that can be learned from it. If you do so, you have my permission to enjoy the amazing finesses in the bishop endgame too. As said, in the first game, I was under a bit of pressure after the opening... Vasyl Ivanchuk – Boris Gelfand Wijk aan Zee 2012 We have spoken enough about Ivanchuk already. All I want to add is that I saw Ivanchuk for the first time in 1983 and played him for the first time in 1985. I have known him for a long time. 1.d4 Nf6 2.c4 e6 3.Nf3 d5 4.g3 The Catalan, which I play a lot myself. As you have to play both colours, you will eventually face your own favourite weapons. 4...Bb4†

54

I have relied quite a lot on this move. To oversimplify the choices for Black here, he can either take on c4, or he can choose to keep the position closed. If he keeps the position closed, then this check makes a lot of sense. All of White’s replies have an obvious drawback. 5.Nc3 allowed 5...dxc4 under favourable circumstances. 5.Nbd2 also allows 5...dxc4. We see this in the game against Caruana on page 161, so let’s focus on the move Ivanchuk played. 5.Bd2 The drawback of this move is that White cannot play Nbd2, Qc2 and e2-e4 quickly, which has turned out to be the main problem with the old closed variation with 4...Be7 and castling. 5...Be7 6.Bg2 0-0 I have played dozens of games from this position with both colours. 7.0-0 c6 8.Qb3 Nbd7 The idea behind this move is that if White plays 9.Nc3 Black has 9...Nb6 and it is not easy to protect the c4-pawn. Black is quite happy after both 10.cxd5 exd5 and 10.c5 Nc4 followed by ...b6. 9.Rc1

55

9...b6 This is something of an admission. After this Black is not pretending to play for equality, but remains solid and just slightly worse. 9...Nb6 would reveal some of White’s ideas. He can play 10.Ba5, or 10.c5 Nc4 11.Rxc4!? (11.Be1 is also reasonable) 11...dxc4 12.Qxc4, with good compensation for the exchange. I think the vast majority of top players would take White here. These days all types of refined moves are played here. Just as 8.Qc2 Nbd7 9.Rc1 has become fashionable. Black is mainly waiting with 9...h6, 9...a5 or something else, reminiscent of the temposaving moves played in the Queen’s Gambit a hundred years ago by the likes of Alekhine, Capablanca and Euwe. Black wants to make minor improvements to his position, while he waits for White to show his intentions. 10.cxd5 cxd5 This is the real starting point for our story. White has achieved a pleasant position and now has to back it up with a game plan. 11.Na3 Bb7

56

12.Bb4 For some it may seem that White’s play is not very ambitious. He is just exchanging his bad bishop for Black’s good one. But actually, Ivanchuk’s idea is very straightforward. He wants to exchange the bad bishop, both pairs of rooks and then the endgame would be better for White, as his remaining bishop is better, his knights are a bit better, and he has a bit more space to use for advancing his position. In some ways this is reminiscent of my game with Magnus Carlsen, which is annotated in Positional Decision Making in Chess on page 106. The position is perhaps not too far from equality, but on the other hand it is not pleasant at all to defend for Black. With the strategy Ivanchuk chose in the game, Black has to acknowledge the need to defend a slightly worse position for a long time, with hardly any chance of playing for a win. Clearly Ivanchuk wanted this kind of game against an opponent in poor form, hoping that I would make a minor concession somewhere that would make it possible to really put on the screws. At the same time White is playing with very little risk. The fact that he lost the game is an outlier and more about his expectations for the game than about the position. 12.Bf4 was also interesting, but would have been an entirely different game. 12...Bxb4 I spent over twenty minutes on this move and realized that I should exchange the bishop and the two sets of rooks and hope to be OK. I was slightly more optimistic than I should have been. But this is the nature of optimism. It is the hope that things turn out better than they should on average. I also considered 12...a6, allowing 13.Bxe7 Qxe7 14.Rc7 Rab8 15.Rac1, when the best I can play here is 15...Rfc8, which is very similar to the game. The reason I spent a lot of time was to try to work out the small finesses. It is possible to lose or win half a tempo easily and regret it for a long time.

57

13.Qxb4 a6 14.e3 14.Ne5 Nxe5 15.dxe5 Nd7 would give me a square for the knight on c5 and Black is OK.

The key thing to understand is that the knight does not belong on c5 immediately. Black should first play ...Rc8, ...a5, ...Qe7, ...Rc5 and ...Rfc8 and he should be very near equality. White is of course slightly better, but the defensive potential is huge. 14.Rc2 was also interesting. After 14...Rc8 15.Rac1 Rxc2 16.Rxc2 Black should be careful.

The idea is that 16...Qb8 can be met with 17.Ne5! Rc8 18.Qe7 with an advantage. This is analogous to the game Kasimdzhanov – Kramnik, Tromso (ol) 2014, where White improved upon my aforementioned game against Carlsen with a big novelty and won a great game.

58

The difference is that the bishop here is on g2 and not on the other diagonal. So, for this reason Black should play more patiently with 16...h6, which makes a lot of sense. It is a good inclusion and the position is not the type where a tempo will matter much. 14...Rc8 15.Bf1 This was of course Ivanchuk’s idea. 15...Rxc1 15...h6 also made sense. This type of move is almost always useful, as explained to me by Levon Aronian. 16.Rxc1 Qb8 17.Nb1 Rc8 18.Rxc8† My idea was to meet 18.Nc3 with 18...Ne4, but instead he exchanged the rooks. 18...Qxc8 19.Nc3 Qf8 20.Qb3 One of the ideas is that White can meet 20...Qd6 with 21.Ne5! with some pressure. So instead I improved my position patiently. 20...h6

This is a good moment to take stock of the position. After the concession with 9...b6, both sides have played reasonable moves. White has achieved his goal with a better bishop and a little bit of extra space. But what is important is that the position is much easier to play for White and with low risk. One of the difficulties for Black is the problem we will speak about in Chapter 3 – to understand when he needs to defend passively and when he needs to be active.

59

21.Ne1?! The first step in the wrong direction. This manoeuvre did not work out well for Ivanchuk in the game. Probably he wanted to bring the knight to b4 and apply some pressure. The best move was also the simplest. White should play 21.Bd3, with the typical plan of 21...Qd6 22.Ne5. After analysing with the engines, it is possible to understand that Black could play 21...Qc8 with the idea of ...a5 and ...Ba6, solving the problem of the bad bishop. White remains a bit better and has many ideas here. We analysed quite a few of them. I do not pretend that the following line by any means is critical, but after: 22.Qa3 Qf8 23.Qa4 Qc8 24.Kg2

24...a5 25.Bb5 Ba6 26.Bxa6 Qxa6 27.Ne5 Qc8 28.Qc6 Even Jacob was able to understand that White is better. We debated if Black should play 24...Ne8 with the idea ...Nef6, when Black is just waiting, which seems reasonable. 21...Qc8 I could also have played 21...Ne4 immediately, but I preferred to improve the placement of my queen first. 22.Nc2 I believe that 22.f3 would be a step in the wrong direction and that White would later regret weakening his dark squares. It prevents ...Ne4 immediately, of course, but it also gives Black the chance to advance with ...g5-g4, or ...h5-h4. Black will play 22...Qc7 and ...Ne8-d6 getting the knight to a great square, from where it will be very effective, similar to the Carlsbad structure. By the way, this sort of simple improvement of an ineffective piece is something we should all learn 60

at an early stage in our chess development. But not everyone does. I have some training positions that 2650-players sometimes get wrong. They look for the active moves, but it is frequently not time to act before all the pieces are ready. 22...Ne4! A nice concrete move. If White manages to play Nb4, he will have achieved something. 23.Nxe4 dxe4

I think Ivanchuk believed that this pawn structure was in his favour, but it is not so. 24.Qc3? Perhaps White is too determined to prove an advantage, leading to the mis-assessment of the position after this exchange. The most natural path would be 24.a4, then 24...Bd5 25.Qc3 Qxc3 26.bxc3. At this point we get into a forced ending, where White is optically better, but where the position is also holdable for Black. In our analysis White was able to create some practical problems for Black, but not many.

61

During the game I wanted to play 26...Bb3. But simpler is 26...Bb7! 27.Nb4 Nb8!, when White cannot make progress easily. Black will bring in the king and be equal. 24...Qxc3 25.bxc3 b5

This move could look a bit unpleasant to some, as most of Black’s pawns are now on the bishop’s colour. But as they are working together, it is not a problem. We have reached the turning point of the game. At this point White should have concluded that things

62

have gone wrong and that it was time to play for a draw. Then comes the next question: how to do it. For example, 26.c4 looks reasonable, but after: 26...bxc4 27.Bxc4 Nb6 28.Be2 a5 29.Kf1 Bd5 30.a3 Bc4!?

White is still a fraction worse. When you were trying to put your opponent under pressure just a few moves earlier, this is not an easy thing to accept. It is not a draw. We do not leave the playing hall at this point, but instead play on and probably for quite a long time. I am sure that Ivanchuk does, but it is my feeling that a lot of players, even very strong ones, do not ask themselves: “Am I better? What should I be looking for in this position?” Often we see in the comments to games that many decisions are made according to how a player feels in a certain moment and not their analysis of the position. “I was in a fighting mood,” would be a typical example of such a comment. Sometimes optimism in your blood helps, but mainly when you are playing against weaker opposition. When you are facing a player of equal or superior strength, this is not the case. In the long run, a faint pessimism will serve you better against stronger opponents. Ideally, we should play with an objective assessment of the position, but this is not always easy. I prefer to follow the assessment of the position rather than emotional factors, but as a human being this is often easier said than done. Thus, my personal attitude is to keep a moderate optimism and avoid extremes; a reflection of my character. 26.a4?

63

A horrible move after which White is probably just lost. We cannot talk about this as a step in the wrong direction, unless we say it is out over the edge of the cliff and include the subsequent long vertical travel to the bottom of it. It is quite easy to see what Ivanchuk’s idea was. The a-pawn will not be easy to push forward and the white king will come to the queenside, claim the pawn and White will be pressing. But in thinking up this long plan for himself, he underestimated Black’s ability to make progress on the kingside. As we shall see, Black arrives much faster than is immediately apparent. 26...bxa4 27.Be2 Kf8 It is possible that 27...f5 28.c4 e5 was a bit more precise, but I see nothing wrong with bringing the king into the game. 28.Kf1 Ke7 29.Ke1 f5 30.Kd2 g5 31.c4 e5

64

At this point it is already apparent that all White has achieved with 26.a4? is to provide Black with an extra passed pawn. 32.Kc3 f4 33.h3?! Stockfish suggests that White should have tried 33.Bg4, when after 33...exd4† 34.Kxd4 Nf6 35.Bf5 I would have had to think and would hopefully have found 35...f3!, when the long-term weakness of the white pawns on the kingside becomes apparent and the f3-pawn is a powerful asset. As the game is very similar, I am optimistic I would have found it.

I do not see a way for White to draw, although equally, I would also not be entirely flabbergasted if it were there. 65

For example: 36.Kc5 36.g4 Kd6 37.c5† Kc7 38.Ke5 Nd7† 39.Kd4 is another try. Essentially, we are still too early in the ending to be able to calculate it all the way to the end. 36...Nd7† 37.Bxd7 Kxd7 Looking at the exchange into knight against bad bishop is quite illustrative, even if White would not necessarily go for it. 38.Na3 38.h4 gxh4 39.gxh4 Ke6 and Black wins. 38...h5

Aimed against g3-g4 ideas. 39.Kb4 Ke6 40.g4!? Trying it anyway, but now both 40...h4!? and the more direct move look promising. Passive play with 40.Kxa4 Kf5 41.h3 Ke5 is also bad for White. Black is preparing for ...Bc8 and after 42.c5 Kd5 43.Kb4 a5† 44.Kb5 Bc6† 45.Kb6 Bd7 46.h4 g4 47.Nb1 Kc4 48.c6 Bxc6 49.Kxc6 a4 and Black wins. 40...hxg4! 41.Kxa4 Kf5 42.Ka5 g3! 43.hxg3 Kg4 44.Kb6 Bc8 45.Nb1 Kh3 46.Nc3 Kg2 47.Nxe4 g4 48.c5

66

48...a5! A smart trick, needed to get access to the c6-square without White’s interference. The tempo won means death to the f2-pawn. 49.Kxa5 Bf5 50.Nd6 Bd7 Black wins. We also looked at 33.gxf4 exf4 34.Bg4, but again Black has a simple plan to attack the pawns on the kingside: 34...Nf6 35.Bf5 Kd6! (35...Kd8 36.d5 would allow White to include the king via d4 and maybe White would have some drawing chances.) 36.Kb2 Kc7 37.Ka3 Bc8 Black will follow up with ...Ng4 with a promising position. Black is close to winning. 33...exd4†! 34.Nxd4 Black would break through after 34.exd4 e3!, and also 34.Kxd4 f3 followed by ...Kd6 and ...Nc5 is hopeless. 34...f3 35.Bf1 35.Bd1 Nc5, with ...Nd3 in the air, is also bad. 35...Nc5

67

36.Kb4 Nd3† 37.Bxd3 exd3 White is unable to control the black passed pawns. 38.Kc3 Be4 39.c5 Kd7 40.g4 Kc7 41.Kd2 a3 42.Nb3 Bd5 0–1 It is of course easy to make this look like a poor game, but it is important to see the game in context. In this tournament I was playing quite poorly. From Ivanchuk’s perspective, this was a game he should try to win and probably this made him a bit reckless, taking risks he would not normally have taken. But the key thing is of course that he failed to realize that the position was turning. And once he did, he was already fighting for his life. Without success. In the second game, as promised, it is my own failure to grasp the realities of how things are developing that is on display. While the Ivanchuk game has a clarity to it, here you can benefit from knowing the thoughts of the losing side. Boris Gelfand – Wang Yue Sochi 2008 I played this game against Wang Yue from China in the 8th round on the 8th day of the 8th month of the year 2008. In China 8 is a lucky number. While we can try to understand the chess reasons why I lost this game and learn from them, we should also accept that it was most likely my destiny to lose this game. My play was below par and probably there was no way I could have avoided this. 68

1.d4 d5 2.c4 c6 3.Nc3 Nf6 4.Nf3 dxc4 5.a4 Bf5 6.Ne5 Nbd7 7.Nxc4 Nb6

This variation of the Slav Defence was popular at the time among the Chinese players. Wang Yue has always had a great fondness for very solid openings. 8.Ne5 a5 White has a wide choice here. 9.h4 is currently popular. I have also played 9.Bg5 a number of times. 9.g3 This is a more classical way. We simply develop the pieces. 9...e6 10.Bg2 Bb4 11.0-0 0-0 12.e3 h6 13.Qe2 Bh7 14.Rd1 Nfd7 15.Nd3 Qe7 16.e4 It makes no sense to debate the opening in a game that will focus on the endgame, so let’s just say that here the normal move was 16...Rfd8, but Wang Yue played a new idea. 16...e5

69

This is where the game starts and the first time for White to make a real decision. It was clear to me that White would want to take twice on e5 and play Be3 – but how to go about it? It occurred to me that if I played 17.dxe5 Black would have 17...Nc5 and that this was his idea. But a closer look will reveal that White has the excellent exchange sacrifice: 18.f4! Nb3 19.Be3 Nxa1 20.Bxb6 Nb3

Now White can play a number of things, most logically to put the queen on e3 or f2, so that Black does not have access to the c5-square with his minor pieces or, as a consequence, the d-file with his rooks. It is rare we see a position like this, where all Black’s pieces are bad. But this is the case. White has an overwhelming advantage. No one cares about the pieces’ list value when their performance is so poor. 70

So, for this reason, Black would perhaps have had to play 17...Nxe5 18.Nxe5 Qxe5 19.Be3, as discussed below. 17.Nxe5?! Nxe5 18.dxe5

18...Bc5! So here it turned out that I had outsmarted myself. In my attempt to deprive my opponent of an extra option, I had given him one. With this move, Black defends his dark squares and thus the stability of his pieces. White now has comparative difficulties in developing his pieces. 18...Qxe5? 19.Be3 would indeed have been very pleasant for White. Black would not like to retreat, but after 19...Bc5:

71

All I would have to do is find 20.Rac1! to claim a significant advantage from the opening. (20.f4? Qe7 only helps Black. 21.Bxc5 Qxc5† 22.Qf2 Qxf2† 23.Kxf2 Rfd8 and he is OK.) 20...Bxe3 21.Qxe3 Nc4 22.Qd4 Qxd4 23.Rxd4 Nb6 24.Rcd1 And White is firmly in control.

So, at this point I started thinking. 19.Be3 I spent 22 minutes on this idea. I still wanted to be ambitious, but my play was not up to a standard that would support this. One idea was to play 19.Bf4, with the idea 19...Nd7? 20.Bxh6. 72

So instead Black would have to play 19...Rad8 20.Rxd8 Rxd8 21.h4, when a strange position arises.

Stockfish says that it is best to simply wait with: 21...Kh8!? The key point is that after 21...Bd4 22.Rd1 Bxc3 23.Rxd8† Qxd8 24.bxc3 Nxa4, White has 25.e6! fxe6 26.Qc4, when the check on e6 gives him a nice tempo. There are many possible lines here, but White is better in all of them. For example: 26...Nb2 27.Qxe6† Kh8 28.Qf7 where the e-pawn is a menace and Black is certainly squeezed. If the computer can hold or not is less important to me. But 21...Kh8 is not a move that anyone would play. I would expect my opponent to have found 21...Nd7 22.Rd1 Bb6, when I believe White is the stronger side. This should be understood in the context that the position is probably objectively equal, but I would choose White. Black certainly has things to play for as well, but he is probably afforded one inaccuracy less than White before trouble is knocking on his door. White can try either 23.e6 with 24.Be3 and f2-f4 coming, or 23.Qg4 with an invasion of the black position via c8. 19...Bxe3 20.Qxe3 Nc4

73

21.Qe2 At this point I am not entirely sure what I missed, but I think I can guess. I wanted to and should have played: 21.Qd4!? Black has only one serious move. 21...Nxb2 21...Nxe5 22.f4 would give White a pleasant position, as the knight has no good squares. For example, 22...Ng4 23.Bf3 Rfd8 24.Qb6 with an excellent game for White. 22.Qd6! 22.Nd5?! is an empty tactic. After 22...cxd5 23.Qxb2 dxe4! followed by ...f5, Black is at least not worse. If White wanted to equalize, this is not the best way. I think that I missed that after 22...Qxd6 23.Rxd6 Nc4, White has 24.Rd4! Nxe5 25.f4 Ng4 26.Rb1! with some pressure. The other lines do not look critical or something I would avoid during the game. The best move is thus: 22...Rfe8 The logical move is still: 23.Rd4

74

When after: 23...Qxe5! 24.Qxe5 Rxe5 25.Rb1 Rc5 26.Rxb2 Rxc3 27.Rxb7 Rc1† 28.Bf1 g5 The position is equal – if you are an engine. Humans would still choose White here. 21...Nxe5 22.f4 Again, I invested a lot of time in this move. 22...Nd7! I was hoping for something along the lines of 22...Qc5† 23.Kh1 Nc4, where I felt 24.Rac1 was dangerous for Black. After 24...Rfd8 the possibilities include:

75

25.Rxd8† Rxd8 26.Nd5? is immediately refuted by 26...Bxe4!. More interesting is 25.Nd5!? cxd5 26.Rxd5 Rxd5 27.exd5, when the ingenious 27...Qe3 28.Qxc4 Bf5! leaves Black with decent compensation for the pawn. The white position is simply too open. But strongest is 25.b3! Na3 (the only square) and now 26.Nd5 Qf8 27.Nb6 does not achieve anything. The knight is not well placed here and after 27...Rxd1† 28.Rxd1 Re8 Black is not worse. Instead White has 26.Rf1!!, when Black is suddenly under a lot of pressure. The key point is that 26...Qb6 27.f5! Qxb3 28.e5 gives White a strong initiative. It is not at all clear what Black is doing on the queenside. It will be a long time before this stolen pawn reaches any type of significance. Black should play 24...Rad8, but still I would choose White every day of the week. But as we saw, instead Wang Yue simply retreated.

23.Rd4?! This was the moment where I had to realize that things were not going well, and that I should secure a draw while it was still easy to do so. The right move was 23.e5 Nc5 24.Ne4, when the exchange of the minor pieces would give White a position with minimal risk. But of course, Wang could also have tried for more with 24...Bxe4!? 25.Bxe4 Rfd8 with chances for both sides. Wang Yue would definitely play on, although I am not sure he would give up bishop for knight. I would also play on. It is in our nature. But neither side would have any claim to an advantage. But psychologically, and this is the main point, if you make a few mistakes, or even miss the opponent’s moves a few times, as I had done, perhaps you should accept that today is not your day. Your head has to convince your heart that you are playing for two results, and maybe it is time to go 76

for the better of these immediately. This is not always easy to do. Often when we see uninteresting draws in games at the top level, this is what happened. One player may have slept poorly or just feel fuzzy in the head, and decided to get half a point and recover. This is not a cowardly thing, but common sense. I have never been good at this, but will learn from this lesson; although Jacob has no faith in me. 23...Rfd8 24.Rad1 Nc5 25.Qe3 Rxd4

The tide has shifted. The position may still be equal, but Black’s side should be preferred. Black is structurally better. The knight on c5 is better than the knight on c3 and the queenside configuration with an extra c-pawn and the weak a-pawn is also not in White’s favour. As he can do nothing on the kingside, it is clear that he should have preferred the mass exchanges with 23.e5. 26.Rxd4 I could have taken control over the d-line with 26.Qxd4, but the reality is that there is nothing there. Black should play 26...f6! after which he is already better. The knight on c5 is stable and the bishop is joining the game via g6-f7. 26...Nb3 26...f6! was also strong here. There is no reason for Black to allow White to play e4-e5 at his leisure. 27.Rd1 Re8! The accurate move. The e4-pawn is the weakness, while there is nothing in the d-file.

77

27...f6 does not work here. White plays 28.Nb5 and after 28...Qc5 29.Nd4 the exchanges will lead to a draw quite quickly. I was also not afraid of: 27...Rd8

28.Bf3 Stockfish recommends 28.Rxd8†?! Qxd8 29.Qa7, but it looks very scary. After 29...Nd2! 30.Qxb7 Qd4† 31.Kh1 Bg6 we have a position where it is easy to believe the engine’s evaluation – that White would save the position with accurate play, but a human player would hopefully also recognize that if he makes a mistake here, he would not be worse, but most likely lose immediately. 28...Rxd1† 29.Bxd1 Qb4

78

It was my intention to take, as I did not believe I would be worse when Black has such a poor bishop. For example: 30.Bxb3 Qxb3 31.Qd2 Kh8! Bringing the bishop into the game. 32.f5 f6 33.Qd8† Bg8 34.e5! fxe5 35.f6 And White has active counterplay, leading to a likely draw. But the engine gives an even more satisfying option: 30.Na2! Qc5 (30...Qd2 31.Qxd2 Nxd2 32.e5 is not scary at all) 31.Kf2 Nd4 32.Nc1! and White is not worse. The key point being: 32...Bxe4!? 33.Qxe4! Nb3† 34.Kg2 Nxc1 35.Qe8† Qf8 36.Qd7 with a lot of activity for the pawn. 28.e5 This is the most natural move, freeing the e4-square for the white pieces. Looking at it now, I am also considering 28.Bf3 with the idea 28...Bg6 29.Kf2 or 29.e5 Qc5 30.Qf2!, when White may be a bit better off than in the game. 28...Qc5 After 28...Bc2 White has a nice reply: 29.Rd6!

79

Suddenly Black is in grave danger. White is about to play 30.Be4 and start a strong attack. The knight on b3 is abandoned. However, Black can escape with a neat tactic. 29...Qxd6! 30.exd6 Rxe3 31.d7 Rd3 with strong counterplay. White should now choose 32.Nd5!, when all Black has is a perpetual check after 32...Rd1† 33.Kf2 Rd2†. If White gets greedy and goes for the full point with 32.Bd5?! cxd5 33.d8=Q† Kh7, he can quickly end up in trouble. For example, 34.Qd7 Rd2 is a bit uncomfortable, even though White can still draw in a few ways.

29.Qxc5 80

This move is probably the clearest indication that I was not having a good day and the power of 8 was against me. White cannot play 29.Kf2, as the exchange would bring the king to the e-file, where it would be exposed. But I simply did not see the option of playing 29.Qf2!, when the position is still basically equal. 29...Nxc5 30.Bf1?! A passive and sad move. There is no pretending that White is not worse after it. Looking at the game now, I am thinking I should have played 30.Kf2.

If Black tries to steal the pawn on b2, I get my rook to d7, which is a lovely trade. If Black plays 30...f6, White has 31.exf6 gxf6 32.Rd2 and no problems. And against 30...Kf8, White can also try 31.Rd2, although I instinctively would prefer to go for an active defence with 31.Ke3!? f6 32.Be4! Nxe4 33.Nxe4 fxe5 34.f5!, with sufficient compensation for the pawn to draw the game. 30...Kf8 31.Kf2 Ke7 32.Ke3?! I played this stupid move to force him to open the position. It was also possible to play 32.Be2 when I do not see that Black has anything better than ...f6, transposing to the game, but giving White two extra tempos. 32...f6 33.exf6† Kxf6† 34.Kf2

81

34...Bc2 It is hard to argue against moves that win games. Wang Yue noticed that 34...Ne4† does not work. White has 35.Nxe4† Rxe4 36.Bd3 Rd4 37.Bc2 with a draw. So, he first fires in the bishop, with the idea of ...Ne4† as a follow-up. But as we shall see, it gives me the chance of exchanging the knight and entering a simplified endgame, which was within drawing distance. Trickier is 34...Bg8!?. After 35.Be2 Ke7! White has to play very accurately not to lose quickly.

82

Black is threatening ...Bb3 and then after Rd4 he will have ...Rd8! and the pawn on a4 is doomed. So proactively, White has to play 36.Rd4! in order to meet ...Bb3 with Bd1. And after 36...Rd8 he has 37.Rxd8 Kxd8 38.Ke1! with Bd1 coming. Black is a little better, but really only a little. The most nuanced move was therefore 34...Bf5!, preparing ...Ke7 and ...Rd8, with ...Bc2 on the way. For example: 35.h3 Ke7 36.Ke3 Bc2 37.Rd4 Bb3 and ...Rd8 wins a pawn. For this reason, White has to find: 35.Rd6†! Ke7 36.Rd2, with the idea: 36...Rd8

37.Ke3!, which would certainly have been a challenge.

83

35.Rc1! I do not know anyone who would allow: 35.Rd4? Ne4†! White can of course allow Black to take on c3 with a big long-term advantage, which is a clear no-no. Or he could go for the forced variations. There the big question is if White is just lost, or if he is in the zone of one mistake, drawing by a prayer and a lucky charm (although not on the 8/8/8). No one would risk it, but I will analyse it, as the reader should know this type of ending. 36.Nxe4† Rxe4 37.Rxe4 Bxe4 38.b3 Bc2 39.Bc4 b5

40.axb5 cxb5 41.Bxb5 Bxb3 42.Ke3 We have known since the time of Reti that the king rarely moves to the corner in a straight line, 84

but should instead take space from the opponent’s king along the way. 42...a4 43.Kd4 a3 44.Kc3 a2 45.Kb2 White’s strongest piece is now permanently occupied in the corner. Still, the break-through on the kingside is not necessarily easy. Remember that White would happily give up all his four remaining men for the two black kingside pawns due to the dark colour of the a1-square. 45...Kf5

The critical moment. White would not have had to prove anything to make it to this position, but from here on, he would have to prove he has a black belt in defending. a) 46.Bd7† Ke4 47.f5 This looks like a serious try, but Black can play: 47...Bd1!! The key point is that White no longer has g3-g4 in many lines. 47...Kf3 looks very natural, but after 48.g4! White draws.

85

48.Kxa2 48.h3!? is a nice trick. White is hoping for 48...Kf3? 49.f6! with a draw. Instead Black wins quickly after 48...Be2! and the white king is far away. 48...Kf3! After 48...Bg4? 49.Kb3 Bxf5 50.Bc6† Ke3 51.Kc3 White would make the draw. The key point is that when Black takes on h2, White has Kf2 defending everything. In the end the bishop will die to remove the last pawn. 49.Kb2 Kg2 50.h3 Kxg3 51.Kc3 Kf4 52.Kd2 Bf3

53.Ke1 Be4 Black wins both pawns. 86

b) But White can hold the draw with: 46.Bc6! Kg4

47.Bg2! Passive defence is the only way. White is aiming for a type of fortress, depending on the ability to play Ka1-b2 in perpetuity. 47.Bd7†? loses to the counterintuitive: 47...Kf3 48.Bc6† Ke3! If the white bishop leaves the long diagonal, Black will play ...Bd5, or ...h5 and ...Bd5. And after 49.f5 Bd1!

50.Kxa2 Kf2 51.Kb2 Bf3! the white king is too far away once again. 47...h5 48.Ka1 Be6 87

48...h4 49.gxh4 Kxf4 looks tempting, but White can save the game with 50.Bh3! Kf3 51.Kb2 Kf2 52.Bf5 Kg2 53.h3 and the g-pawn will never become a queen. The typical set-up is Bg4 and h4-h5. 49.Kb2 h4 50.gxh4

50...Kxf4 After 50...Kxh4 51.f5 it is very hard for Black to win the white f-pawn without giving away all the tempos the white king needs to return to the scene of the battle. 51.h5 Kg4 51...Kg5 52.Bf3 and White draws. The key point being that ...Bf7xh5 is met with Bd5 and Black will be down to one pawn. 52.h6 gxh6

88

53.Bf1! Absolutely the only move. If Black had time to play ...h5, he would win. 53...Kf3 We have a case of mutual zugzwang. After 53...Bf7 54.Bg2 h5, White draws with 55.h4! and Black does not have ...Bg4, so he cannot take on h4 immediately. (And in anything that follows, White will always reply to ...h5 with h2h4.) And after 55...Be6 56.Bc6! the white bishop is activated on other diagonals. 54.Bd3!!

But this is a truly beautiful move. The white king is extremely far away, so the idea that this is a position where counting tempos is important seems counterintuitive. And yet, that is exactly the 89

case. For this reason, the white bishop needs to be transferred quickly to the optimal diagonal. 54...Bf7 A necessary move. If Black plays 54...Kg2 White draws with 55.h4 Kg3 56.h5 Kg4 57.Bg6 Kg5 58.Be8! Bg4 59.Kxa2 Bxh5 60.Bd7 Bg4 61.Be8 Bf5 62.Kb2 and the king is back in time. 55.Bf5 The optimal square, once we accept that the pawn is lost. 55...Kg2 56.h4 Kg3

57.h5! White needs all the time he can win. So diverting the bishop is crucial. 57...Bxh5 58.Kxa2 Bf7† Preventing Bf5-g6. 59.Kb2 h5 60.Kc2 Bd5 The immediate transfer of the bishop is the only idea with any bite. 61.Kd2 Bf3 62.Ke1 Bg4

90

63.Be4! h4 64.Kf1 h3 65.Kg1! h2† 66.Kh1 White draws. But finding a path through this minefield is certainly walking with death as your companion.

35...Bb3! I was not totally sure about the position after 35...Bxa4. After 36.Nd5† cxd5 37.Rxc5 Bc6 38.Rxa5 Re4

91

Black is obviously better. The d-pawn is strong. But on the other hand, both sides will soon have only three pawns (after ...g5) and a draw seems overwhelmingly likely. Stockfish says it is equal, but I would feel unsure. The most human for me is to play 39.Rc5 with Rc2-d2 coming and White is holding. 36.Nd5† Bxd5 37.Rxc5 Re4 At this point I had a tough choice with close to no time left on the clock.

38.Bg2!? In practical terms this was probably a good decision and on another day, it would have saved the game. 92

During the game it was very uncomfortable to look at 38.Rxa5 Rb4 39.Be2 Rxb2 as an option, but after 40.Ke3 Rb3† 41.Kd2 Black is only better. Not winning. But he is a lot better. So instead I decided to go for the bishop endgame and activate the king. From afar I thought that the position should be a draw. You can say that I underestimated his possibilities, but also I had to make a move. It is tough to know for sure if what I did was worse than the above line. 38...Rc4 39.Rxc4 Bxc4 40.Ke3 Bb3 The time control is over. White is about to lose a pawn, but has active counterplay with the king. Here I spent 11 minutes. Not to decide on my move, which was rather obvious, but to get my head straight for the defensive task ahead of me. Sadly, this was not enough, as this day I was my own worst enemy. 41.Kd4 Bxa4 42.Kc5 Bb3

43.Be4 This is the final big moment of the game from a competitive stance. I spent 18 minutes deciding on how to defend and got it wrong. The position we would like to understand arises after 43.Kb6 a4:

93

a) First of all, White has to look for the ideal circumstances before taking on b7. This is not the right time. After 44.Kxb7? c5 45.Kb6 c4 46.Be4 c3 47.bxc3 a3 Black wins trivially. b) To understand how White can defend the game, we need to first understand how he loses. 44.Bh3 Be6 45.Bg2 Ke7

46.Bf3? Below we shall see that 46.Be4! would make a draw. 46...Kd6 47.Bd1 It is too late to play 47.Be4. Black wins after 47...c5 48.Bxb7 Bd5! 49.Bc8 c4 50.Bf5 c3 51.bxc3 a3 and the white king is entirely cut off. 94

47...Bb3 48.Bf3 Bc2 49.Bg2 49.Be2 Kd5 50.Kxb7 c5 51.Kb6 a3 52.bxa3 c4 and ...Bd3 and Black wins.

49...Bf5! 49...Bd3? 50.Bf3 c5 51.Bd1 (51.Ka5!? also looks like it should be holding, but is more passive and less conclusive.) 51...a3 52.bxa3 c4 does not work for Black this time around. After 53.Kxb7 c3 54.a4 Kc5 55.a5 Kb5 56.a6 Be4† 57.Kc7! White is in time to create sufficient counterplay. 57...Kxa6 58.Kd6 Kb5 59.Ke5 Bf3 60.Bc2 Be2 61.f5 Kc4 62.Ke4 with a draw. 50.Bf3 Bd7 As we shall see, White would save the game if he were able to get the bishop to c2 fast enough. But he cannot here. 51.Bd1 51.Be4 c5 52.Kxb7 c4 53.Kb6 c3 and Black wins the white bishop and the game. 51...c5 52.Kxb7 c4 53.Kb6 Kd5 54.Ka5 a3 55.bxa3 Kc5

95

It turns out that Black is winning here, but only by a tempo. The variations are definitely worth going over. b1) 56.Ka6 c3 57.Kb7?! Be6 58.Kc7 Ba2! 59.Kb8 This is the best try in this position. It is just a trick, but why not? 59.Kd7 Kc4 60.Bc2 Bb3 followed by ...Kd4-e3-d2 and a winning method similar to the main line arises. 59...Kc4 60.a4

60...Kb4 60...Bb3? would be falling for the trick. 61.Bxb3† Kxb3 62.a5 c2 63.a6 c1=Q 64.a7 is a surprising draw. Black cannot use his extra tempo to do anything. 96

61.Bc2 Be6 62.Kb7 62.Kc7 Bb3! changes everything, as now the c-pawn would queen with check. And if the white bishop moves, Black will take on a4 and eventually queen the c-pawn. 62...Bd7 63.a5 Kxa5 64.Kc7 Ba4 65.Bd3 Kb4 66.Kd6

66...Kb3! 67.Ke7 Kb2 68.Kf8 Bc2 69.Be2 Bb1 70.Bd1 Ba2 The pawn queens. Black wins. b2) 56.Bc2 Bc6! 56...Bg4 57.Ka6 Be2 58.Kb7 Bd3 59.Bd1 c3 looks very dangerous for White, but after 60.a4 c2 61.Bxc2 Bxc2 62.a5 he has a lot of counterplay. For example, 62...Kd4 63.a6 Ke3 64.a7 Be4† 65.Kc7! and the king will run to the kingside and equalize.

97

57.Ka6! 57.a4 Bb7! makes it impossible for White to improve his position. Black gets a few free tempos. 58.Bb1 c3 59.Bc2 h5 Followed by ...Kd4 and ...Be4, winning first the bishop and then the entire kingside. 57.g4 gives Black an important tempo: 57...Bf3! and ...Be2. 57...Kd4! The right timing. The king races forward. 57...c3 58.a4 Kd4 59.Kb6 Be4 60.Bb3 c2 61.Bxc2 Bxc2 62.a5 Bd3 63.Kc7! would allow White to escape with a draw. 58.Kb6 Be4 59.Bd1 Ke3

98

White cannot play a3-a4 yet, as he needs the square for the bishop as long as the pawn is on c4. 60.Kc5 Bd3 61.Kd5 Kd2 62.Ba4 c3 63.Kd4 Bc2 64.Bb5 h5 A useful waiting move. 65.Be8 White cannot improve his position. 65.a4 Bb1 and 65.h3 Bf5! both make it worse. 65...Bb1 66.Ba4

66...g6! Black needs to improve his position to the maximum before going for concrete action. 66...c2? would lead to a draw. 67.Bxc2 Bxc2 68.Ke5 g6 69.a4! Ke3 (69...Bxa4 70.f5 and the kingside is liquidated) 70.a5 Bd3 71.Kf6 Kf2 72.Kg5 Kg2 73.a6 Kxh2 74.a7 Be4 75.g4 There will be no pawns left in a few moves’ time. 67.Bd1 67.h3 Bc2 68.Bc6 Bf5 and the position improves further for Black. 67...c2 68.Bxc2 Bxc2 69.Ke5 Ke3 70.Kf6 Kf3 71.Kg5 71.a4 Kg2 and wins.

99

71...Bf5! The final accuracy. 71...Kg2? 72.g4 hxg4 73.Kxg4 Kxh2 74.Kg5 Kg3 75.a4 and White makes the draw. 72.a4 Kg2 73.a5 Kxh2 74.a6 Kxg3 75.a7 Be4

Black wins. c) 44.Be4! It was only this year that we found out that White is holding the draw here. The key point is comparison. 44...Ke6 45.Bg6 White is not under any great demands for accuracy here. 100

45...Kd6 45...Kd7 could be met with 46.Bf5† Kd6 47.Be4, for example. 46.Be4

46...Bd1 Black can force a queen endgame, where White will have to show some accuracy to draw: 46...c5 47.Kb5 Bd5 48.Bxd5 Kxd5 49.Kxa4 h5 50.Kb5 c4 51.Kb6 Ke4 52.Kxb7 Kf3 53.Kc6 Kg2 54.Kc5 Kxh2 55.Kxc4 Kxg3 56.b4 h4 57.b5 h3 58.b6 h2 59.b7 h1=Q 60.b8=Q Qe4† 61.Kc5 Qxf4 This is a theoretical draw, but still a practical challenge. You can find more about this in Chapter 9 of Decision Making in Major Piece Endings. 47.Bg2 Bg4 47...Bc2 If Black manoeuvres, White will wait intelligently. 48.Bf3 Bf5 49.Bd1! White holds. 48.Be4! Bd7 This plan is the only way to play for a win we have found. This time White is well prepared. 49.Kxb7 c5 50.Kb6 c4

101

51.Bc2! 51.Ka5? Kc5 and wins. 51...h5 The point of comparison is important. After 51...Kd5 52.Ka5 a3 53.bxa3 Kc5 White saves a tempo as the bishop is already on c2. We saw above how many lines relied on this extra tempo.

54.Ka6! The king races to the kingside. 54...c3 (54...Bc6 55.a4 Bf3 56.a5 Be2 57.Kb7 Bd3 58.Bd1 Kb5 59.a6 Be4† 60.Kc7! Kxa6 61.Kd6 and White arrives at the kingside in good time.) 55.Kb7 Be6 56.Kc7 Ba2 57.Kd7 Kc4 58.Ke7 Bb3 59.Bb1 Kd4 60.Kf8 Bc4 61.Bc2! (61.Kxg7? Bd3) 61...g6

102

62.a4! It is essential that White generates counterplay. He cannot keep the c-pawn from queening forever. (62.Kg7 Bd3 63.Bb3 Ke3 64.Kxh6 Kd2 65.h4 Be2 and Black wins.) 62...Bd3 63.Bb3 Ke3 64.a5 White draws. 52.Bd1 Kd5

53.Bf3†!! 53.Bxh5? c3 and Black queens a pawn. 53...Kd4 54.Bxh5 c3 55.bxc3† Importantly, this is now played with check. 55...Kxc3 56.Ka5 a3 57.Bf7 Kb2 58.Kb4 Bf5 59.h4 Bb1 60.h5 Ba2 61.Bxa2 Kxa2

103

62.f5! White is in time. 62...Kb2 63.f6! White should avoid losing the queen to a check on a3. 63...gxf6 64.h6 a2 65.h7 a1=Q 66.h8=Q White escapes with a draw. 43...a4

44.h4? This move is poor for a number of reasons. First of all, there is how the game went, which in 104

itself is a strong enough case. But also, conceptually, there are positions we have seen above, where Black wins the white bishop and White has an a-pawn and three vs. two on the kingside. Having the kingside blocked by the two black pawns would allow Black to give up the bishop for the a-pawn and win on the kingside with the king. 44.Kb6!, as analysed above, still holds. It is quite possible that there are other ways to hold the position, but as I cannot think of them now, there would be no chance for me to find them in a game, so what use are they to me? A point to Black’s play is that 44.Kd6?! Bd5 would be progress for him. 44...Bd1! 44...Bd5? 45.Bd3 h5 46.Be2 would give White sufficient counterplay. Most obviously with 46...g6 47.g4 and pawns are exchanged in White’s favour. 45.Kd6 I cannot see any hope after 45.Kb6 h5! either. 46.Kxb7 c5 is dead lost and also after 46.Bd3 Ke6 47.Be4 Kd6 48.Bg2 Be2 49.Be4 c5 50.Bc2 a3! 51.bxa3 c4 Black wins. 45...h5!

Surprisingly, I now found myself caught in a mutual zugzwang. If Black were to play, he would have to allow either Bf3 or Kc5 ...Ke6 Bc2, both holding easily for White. But as it is, the white pieces are ideally placed and the fortress collapses after any way they move. 46.Bd3 105

We should also look at the following line, although it is hopeless for White: 46.Kc5 Ke6

47.Kb6 47.Bd3 Bf3 with the idea ...b5, which is similar to the game, would force White to play 48.Kb6. Black wins by brute force. 48...Kd5 49.Bc2 a3 50.bxa3 c5 51.Kxb7 c4

52.a4 White has to hurry up. (52.Kb6 Be4!) 52...Kc5† An unpleasant move for White to face. 53.Kc7 Be2 54.a5 Bd3 55.Bxd3 (55.Bd1 c3 and Black wins easily. The damage of the pawn being on h4 is easy to see here. White lacks those vital h3&g4 ideas.) 55...cxd3 56.a6 d2 57.a7 d1=Q 58.a8=Q Qd6† 59.Kc8 Qf8† 60.Kb7

106

Exchanging the queens also wins, but when faced with a classic sequence, one should conform. 60...Qe7† 61.Kc8 Qe8† 62.Kb7 Qd7† 63.Kb8 Kb6 White is mated. 47...Kd6 48.Bg2 Be2 49.Be4

Eventually we get to this point. 49...c5! 50.Ka5 Black wins quickly after both 50.Kxb7 c4 and 50.Bc2 a3! 51.bxa3 c4 and ...Bd3. 50...b5 Black has achieved the best of all worlds. 51.Bg2 Ke6 52.Be4

107

Black wins with a bit of advanced, but not too difficult, triangulation. 52...Kf6 53.Bc2 Ke7 54.Bg6 Against 54.Bb1 c4! and ...Bd3 decides. 54...Kd6 Threatening ...Kd5. 55.Be4 Ke6!

White is in zugzwang. 56.Bc6 c4 57.Kb4 Kf5 58.Bd7† Ke4 59.Bxb5 Kf3 60.Kxa4 Kxg3 61.Kb4 Kxh4 62.Bxc4 Bf3! The last accurate move, but not a difficult one. Black wins.

108

46...Bf3 47.Kc5 b5 48.Bc2 Ke6

49.Bg6 I was still desperately trying to find an idea to hold this position, but none existed. 49...Kd7 50.Bc2 Kc7 Simple chess. The king is aiming to go to a5. 51.b3

51...a3 109

51...Bd1! would have been a nice shot. 52.Bb1 Be4 53.Ba2 Kd7 54.Kb4 Ke6 55.Kxa3 Kf5 0–1 A good game for my opponent, according to his style. The turning points were quite instructive and show well how we should deal with them, I hope.

110

Chapter 3 Passive or Active Defence?

India 2020

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

111

How did I manage to put my opponent under a bit of pressure? (see page 65)

How can White escape the draw? (see page 66)

112

How can White keep the initiative? (see page 68)

The win is not as obvious as a first glance suggests (see page 73)

113

This one is a bit more straightforward... (see page 74)

White to win! (see page 74) The game in this chapter is one of the oldest games in this book and for this reason my memory of it may not be the best. The tournament was held in honour of the 70th Birthday of Viktor Korchnoi, who true to style won the tournament, beating both Grischuk and myself along the way. It was a very impressive performance overall. I think he rated those games highly, as he included them in his best games collection, released not long after. And Korchnoi played a lot of good games in his life. After 114

all, he probably played the highest number of classical games of any player in history. The following game is far from a masterpiece. In the opening my opponent tricked me with the move order and equalized. Later both players made plenty of mistakes. What it is, is a great representation of the struggle. It shows mistakes we should avoid, but which remain inevitable, due to the complexity of chess. Thematically, the game circles around the concept of active and passive defence. By definition, active defence and passive defence are not good and bad, but better or worse according to the circumstances. It is often easier to defend actively for a number of reasons, including that you would actually be doing something. When you defend passively, there will come a time when you have to change to active defence. Understanding when to do so is very difficult. In the game we will see that Pelletier did not set out to defend passively, but after he got into trouble, he had to choose between active and passive choices continuously, which was very difficult. One exception to passive defence eventually needing to go active, is fortresses. There passive play holds. Often only just. Towards the end, this theme will raise its head and say hi, just for a moment.

Boris Gelfand – Yannick Pelletier Biel 2001 Yannick Pelletier is a Swiss player, and also commentator and ex-organizer of the Biel Festival. He played in Biel often, as he is a local player and often he did better than his rating would suggest. Pelletier has played ten games with Carlsen over the years; nine of them in Biel. The first year they played there together, Pelletier won in 15 moves against the young Grandmaster. The last time they played, in the 2015 European Team Championship, Pelletier exploited a blunder from the now World Champion to win against him a third time. In total his score against Carlsen is three wins, five defeats and two draws. Pelletier has won many good games, for example against Nakamura in the 2015 European Club Cup. He has many achievements to be proud of. And most importantly, he is a very nice person. 1.d4 d5 2.c4 c6 3.Nc3 Nf6 4.e3 e6 5.Nf3 Nbd7 The Meran was popularized by Akiba Rubinstein when he played it in Merano in 1924, but he had already played it in 1923 against Teichmann in Carlsbad. Both Pelletier and I have played it for most of our lives, with good reason. It is an excellent opening. 6.Bd3 dxc4 7.Bxc4 b5 8.Bd3 Bb7 9.0-0 b4 10.Ne4 Be7

115

11.a3 This sort of transposes into the 9.a3 line, which I have played a number of times with both colours. I hope to talk more about the Meran in later volumes. 11.Nxf6† Nxf6 12.e4 0-0 was also possible. This line became popular after the Korchnoi – Polugaevsky, Candidates match in 1977. So, in a tournament dedicated to Korchnoi’s 70th birthday, I should have played it. But the problem is that I did not find anything impressive here. 11...Nxe4 12.Bxe4 Nf6 13.Bd3 bxa3

The challenge for White here is that if Black can castle and play ...c5, no problems would appear. For this reason, I tried a new idea. 116

14.b3 A novelty at the time, but not a particularly impressive one. 14.bxa3 0-0 15.Rb1 is critical. 14...Qb6! Pelletier plays the opening well and successfully equalizes. Sensing the importance of playing ...c5, he carries it out immediately. 14...0-0? 15.Bxa3 would allow White to take control of the c5-square, leaving Black in a desperate state. In a later game I got the chance to show my idea in full: 14...Qa5

15.b4!! Bxb4 16.Bxa3 Bxa3! The only other game in this line went: 16...0-0 17.Bxb4 Qxb4 18.Ne5 Rfc8?! (18...a5²) 19.Qc2 Qe7 20.Ra5!± c5 21.dxc5 Bd5 22.Rfa1 Qb7 23.Ba6 Qc7 24.Bxc8 Qxe5 25.Bb7 Bxb7 26.c6 Qc7 27.cxb7 Qxb7 28.Rxa7 1–0 P.H. Nielsen – Baramidze, Plovdiv 2008. 17.Qb3 0-0 17...Qb4 18.Rxa3!² (18.Qxa3 Qxa3 19.Rxa3 would allow 19...c5!) 18.Rxa3 Qc7 19.Qc2?! 19.Ne5! c5 20.Qc2 with the idea Rc3. 19...Rfc8 20.Rfa1 a5 21.Rc3 g6?! 21...Ba6! would have more or less equalized. 22.Ne5 Nd7?! After 22...Ba6 23.Be4 Nxe4 24.Qxe4 Bb5 25.h4!? White has the slightest of initiatives. 117

23.Nxd7 Qxd7 24.Rc5 Ba6 25.Bxa6 Rxa6 26.Raxa5

26...Rb6? 26...Rca8 27.Rxa6 Rxa6 28.g3² would be uncomfortable for Black, but tenable. 27.g3 Rcb8 28.Ra1 Rb2 29.Qe4 R8b5 30.Rxc6 Rf5 31.Rc2 Rxc2 32.Qxc2 h5 33.Ra8† Kg7 34.e4 Rb5 35.Qc3 Rb1† 36.Kg2 e5 37.dxe5 Qe6 38.Qc5 Kh7 39.Qf8 Qxe5 40.Qxf7† Kh6 41.Qf8† 1–0 Gelfand – Vallejo Pons, Monte Carlo 2004. 15.Qc2 15.Rxa3 looks fancy, but after 15...c5 Black will equalize. This type of position is rarely considered dangerous. 15...c5 16.Bxa3 cxd4 16...Bxf3? 17.Bxc5 would not work out for Black. After 17...Bxc5 18.dxc5 Qc6 19.gxf3 Qxf3 20.Bb5† Ke7 21.Be2 White’s king is not in danger and he has a big strategic advantage. 17.Nxd4 17.Bxe7 Kxe7 18.Nxd4 Rhc8 makes it easier for Black to get his pieces into the game. 17...Bxa3 18.Rxa3 18.Bb5†? would be an unfortunate intermediate move. After 18...Kf8 19.Rxa3 e5 White loses a piece. 18...0-0 19.Rfa1

118

This was all I could come up with. I have managed to create pressure down the a-file. 19...Rfc8 Black only needs to make one accurate move to equalize completely. He will get a few chances to make it, but sooner makes more sense. The first chance was: 19...a5! Black wants to play this; all he has to see is that it works tactically. 20.Qa2 This looks natural, but also invites a natural reply. Instead 20.Qd2 Rfd8! with ideas like ...Rd5 is going in the wrong direction for White. 21.Rxa5 e5! is not a good idea. 20...Nd5! 21.Rxa5 Nb4

119

White has 22.Bxh7†, but 22...Kh8! spoils all the fun. White is about to lose the queen or a piece and should find 23.Rh5 with a perpetual (22...Kxh7 also works out fine for Black, but why find three precise moves when one is enough?). 20.Qd2 Although I had achieved nothing from the opening, I was motivated to play for a win. This game was played in the second round and the day before I had lost to Korchnoi. I was seeking revenge and Yannick crossed my path. My idea here is to put the rook or queen on a5, but it is still not a difficult position for Black.

120

20...e5!? This is the first time we seriously encounter a choice for Black between active and passive defence. Earlier, active moves like 14...Qb6! did not really have an equivalent passive option. At this point it is different. And it is one of the situations where the passive move is simple and the active move complex. In general, if you have two equal solutions to a problem, the simple one is to be preferred. It was practical for Black to simply improve the position with something like 20...g6, or 20...a6, when after 21.Qa5 Qd6! there is nothing to worry about whatsoever. 21.Nf5 Rd8 22.Qc3 g6! This accurate move was no doubt prepared when Pelletier played 20...e5.

23.Ng3 When we analysed the game originally, I wanted to give this move a question mark, because the knight is going to a poor square. It just so happens that all the other moves are no better, if not even worse. I do not remember if I considered 23.Ra6!? Bxa6 24.Rxa6.

121

Wild complications can arise after 24...Nd5, which are not necessarily in White’s favour. I am always looking for a chance to create problems for my opponent and something like 23.Ra6 could be the way to do this in principle, although of course I would check if it really would make my opponent’s life harder or just be taking unnecessary risks. To understand when you do one and when you do the other, is not easy to evaluate and often a part of the discussion between grandmasters after the game. But Black has a simpler solution in 24...Ne4! 25.Qxe5 gxf5 26.Rxb6 axb6 27.f3 Rxd3 28.fxe4 Rxe3 29.Qxf5 Rxb3 when White has nothing other than a perpetual, which means it makes little sense to go in this direction. So I do not really care if 24...Nd5 gives Black chances or not. It does not affect my decision whether or not to sharpen up the game. 23...Qc6?! Forcing the exchange of queens, but also creating the first hint of difficulties for Black. It was possible to force a draw with: 23...Nd5!

122

The main point is 24.Qxe5 Nxe3!, when there are many moves, like 25.Be4, that are just equal. Both players have several decent moves around here that all lead to equality. The important point is that White stays far away from 25.Qxe3? Qxe3 26.fxe3 Rxd3 27.Rxa7 Rxa7 28.Rxa7 Rxb3 when it is Black who is playing for a win. 24.Qxc6 Bxc6

Pelletier was no doubt looking forward to 25.Bc4 Bd5! and White cannot steal a pawn without inviting the black rook in to the first rank, behind the white queenside pawn, creating all types of threats. It often happens when you want to force matters that you miss the quiet moves in favour of active ones. 123

25.Bf1! Now 25...Bd5 26.b4 followed by 27.b5 is uncomfortable for Black. 25...Rdb8?! This is the slightly more passive option and could have led to a few additional problems for Black. At this point the active move would have been easier to play. 25...Rab8! 26.Bc4 Bd5 27.Rxa7 Bxc4 28.bxc4 Rb2 Just by looking at the position, it is hard to believe that White is better here. But he can probably still test Black a little bit. 29.R7a6!

29...Kg7 29...Ng4? 30.Ne4 f5 looks like it is equalizing, but the tactics do not work for Black in the following line: 31.h3! Nxe3? 32.Nf6† Kg7 33.fxe3 Rdd2

124

34.Rd6!! Preventing the perpetual and setting up a mating attack at the same time, and without any material investment! 34...Rxg2† 35.Kf1 Kh6 36.Ra7 Kg5 37.Rxh7! White wins. For example: 37...Rh2 38.h4† Rxh4 39.Rxh4 Kxh4 40.c5 and it is all over. This looks alien to the game, but shows how things can go wrong all the time. 30.Rc1 This gives White a little hope, based on 30...Rdd2? 31.Rxf6! and White has a real advantage. So instead Black will play: 30...Rdb8 31.c5 Rb1 32.Ra1 Rxa1 33.Rxa1

33...Kf8 When the draw would be near despite the extra pawn, but White can still ask a few questions; because of the extra pawn. The key ideas are Rc1 and Nf1-d2-b3. When you analyse a 125

position like this, you can easily get misled and think that everything is equal, but when you have to play it, real-world problems set in. 26.Bc4?! A small mistake that did not make a big difference. 26.Ra6! was stronger. After 26...Rb6 27.Bc4 Bd5 28.Bxd5 Nxd5 29.Rxb6 Nxb6 30.Ra5! White is more active and Black has problems to solve.

26...Bd5?! Black returns the favour. Stronger was 26...Bb5! and I don’t see any way White can apply pressure. Winning a pawn with 27.Rxa7 Rxa7 28.Rxa7 is pointless, as Black has 28...Bxc4 29.bxc4 Rb1† 30.Nf1 Rc1 with an immediate draw. 27.Bxd5 Nxd5 28.Ne4 Activating the last piece. 28...Rb4 A rather pointless move. I think Pelletier wanted to provoke 29.Nc5, but it makes little sense for White to put the knight there. On d2 it is easy to support and can find the kind of stability that knights long for. 29.Nd2! White is better. His pieces are more harmonious and his weakness is well defended.

126

29...Rb7

30.Ra5!? Just played to improve the rook, but very successful as a small trap. Analysing this game now, it becomes apparent that this was the perfect time for White to play 30.g4! in order to separate the e5-pawn from the herd. Black is not about to do anything active, so White should improve his position to the maximum. 30...Rd8? Black has already drifted into minor problems, but this is a poor move that suddenly opens up for all types of pins and forks. I think he felt under pressure and had a strong desire to activate the rook. We also analysed 30...Nc3, when all types of active play work out fine for Black. But again, White should consider 31.g4!? with some pressure. Black’s best move is 31...f6, when he is only worse, but not in serious trouble.

127

31.Ne4! The immediate threat is of course Rxd5. 31...Nc7 A sad move to make. 31...Kg7? would be a further blunder. Black loses a piece after 32.Rd1 Rbd7 33.Nc5 Re7 34.e4 Nb4 35.Rxd8 Nc6 36.Rd6 Nxa5 37.Ra6 Nb7 38.Rxa7 Rc7 39.b4. And White will have achieved all of his strategic desires after 31...Nb4 32.g4! Nc6 33.Ra6 Rb6 34.Rxb6 axb6 35.g5!. 32.Rxe5 It is hard to resist taking this pawn, but other moves were also possible. 32.Nc5 Rb5 33.Rxb5 Nxb5 34.g4! was also strong. 32...Rxb3 At this point I decided to make a classical move. It secures the pawn structure and provides stability to the knight on e4. When you play without depth, it is easy to verbally justify a decentlooking move in this way.

128

33.g4? A natural move but also a serious mistake, squandering some of the advantage. The real problem is that in some variations the pawn will hang with check. My idea was to put the pawn on g5 and secure control of the f6-square. 33.Re7? would be a blunder, giving Black 33...Ra3! for free. The best move is therefore: 33.h4! This has the dual purpose of giving the needed “luft” for the king and supporting the knight going to g5 in many lines. White has many options and ideas. The key point is that you have to find out that the g-pawn is in danger of being captured with check. 33...a6 Otherwise this just falls. 34.Re7 Nd5 35.Ra7 White has a big advantage.

129

With the pawn on g4 instead of h4, Black would have ...Rb4! here. But now Black has to look out for Ng5 ideas too. A forcing line arises that no one could have predicted: 35...Rb4 36.Ng5 Rf8 37.R1xa6! 37.g3 h6 38.Nf3 Nf6 39.R1xa6 Ng4 would also give White good practical chances (see the game with Georgiev on page 179), but the temptation to go for the throat is high. 37...Rxh4

38.Rd6! The knight is trapped in the middle of the board. 38...Rh5! 39.Rxd5 f6 40.Rdd7 We could already stop the line here by proclaiming triumph, but as the next few moves are 130

forced, we may as well include them. 40...fxg5 41.f3! If White gets time to play g4, he will win easily. 41.g4? Rh4! 41...g4 42.fxg4 Rh4 43.g3 Rh6 44.Rg7† Kh8 45.Rge7 g5 46.e4

White is winning. It will take a lot of moves, but there is no likely scenario where Black can hold. 33...a6 34.Rc5

34...Rb7? This is not a good move. After this the black pieces are quite awkwardly placed and White has 131

various ways to exert pressure. The fact that I did not find the best way to continue does not change this fact. Pelletier misses his chance to exploit my mistake by finding the active counterplay. After 34...Ne6 I cannot play 35.Rc6 on account of 35...Rb4!. I would probably have tried something like 35.Rc4 Rb6 36.Ra5 where White is still pressing a bit, but a draw is the most likely result. This does not mean that it would be a draw, but that when making moves, Black would be punished severely for mistakes, while any mistake by White would only make the defence easier for Black. But most likely Black would need to make two or three minor mistakes to be in serious trouble. 35.Kg2 35.Rc6! with the various ideas of Ra4/a5, g5 and Nd6 was a little more accurate. 35...Ra7 36.f4? This is a poor decision. The drawbacks are clear from the options Black had in the game, but I felt it was structurally what I wanted to do. 36...Kg7 37.Kf3

37...Rd5?! I had missed this move and was disappointed when it came on the board. But looking at the game now, I see that 37...Rd3! followed by 38...Ne6 was the right way to defend against the immediate ideas. 38.Rc6 a5 39.Nf6 At this point I apparently had a very dangerous idea: 132

39.Nd6!

Black would have had to defend with great accuracy to hold the game. I seriously doubt that my opponent would have succeeded. 39...Ne6 This looks very natural but once you go deeper, you see that Black would already be in serious trouble after this move. 39...Kf8? 40.Rb1! would already spell disaster for Black. 39...Nb5 40.Ne8† Kf8 41.Nf6 is very unpleasant. White has many tempting options at this point. 40.f5! Re7!? I believe this tactic to be essential for Black to hold in a practical game. 40...gxf5 41.gxf5 Nd8 42.Rb6 is treacherous territory for Black. Analysis shows a few ways for Black to hold, but many more for him to lose. If I told you that the best move is 42...Rc5 I am sure you would already accept that in practical terms, this would have been extremely dangerous for Black.

133

41.h4! The most dangerous, preventing the check on g5. 41...Rd3 42.Ra4 Preventing the check on d4, preparing for Ne4. Right? 42...Nd4†! Not quite... 43.Rxd4 Rexe3† 44.Kf4 Rxd4† 45.Kxe3 Rxg4 46.Ne8† There are other tries as well, but this is the dangerous one. 46...Kf8 47.Rc8 Ke7 48.f6† Ke6 49.Rc7

This holds for Black, surprisingly. 134

49...Rb4! Why the rook has to go here, and only here, is apparent from the main line. 50.Re7† Kf5 51.Kd3!? Rb8! Black will hold the draw. 39...Rd8 40.Ra4

40...Ne8!? Move 40 is one of the most dreaded moments in a defender’s life. I believe it is a matter of luck whether you have to make the important decisions, which you have to live with for the next hour or two, on move 40 or move 41. At this moment Pelletier decides to aim for further simplifications and a rook ending. But although it is a likely draw, Black will still have problems to solve. Alternatively, Black could remain passive and play 40...Ne6, when White would strengthen his position after 41.g5 h6 42.h4 with some pressure. If Black had played like this and lost the game, anyone with a computer would criticize him for not exchanging the knights when he had the chance. Sadly, we do not have parallel universes that can show chess commentators what “plus equal” really means. 41.g5 h6 Again, this is a decision that is easy to debate. 41...Nxf6 42.gxf6† Kh6 with the idea ...g5 was another option.

135

This was arguably simpler, although the king may also feel exposed on h6. The life of a defender is never easy. 42.h4 Nxf6 42...hxg5 43.hxg5 Nxf6 44.gxf6† Kh6 is however a very bad idea. Although Black would like to exchange pawns, he does not want to get mated. And after 45.Ra1 Rd5 46.Rc8 Kh7 47.e4 Rh5 48.Rd1 g5 49.f5 this is what will happen. 43.gxf6† Kh7

We now see that it was possible to put this game in Decision Making in Major Piece Endings as 136

well, but although this rook endgame is very interesting, it is the narrative of the game that is most important. 44.Ke2 Preparing to advance the e-pawn. 44...Rd5?! At this point Pelletier decides to go for the least influential white pawn, the h-pawn. I think such a poor decision can only arise after relying too much on calculation, without working out which problems he really needed to solve. Winning the h-pawn would not come high on such a list, but activating the king should. I find it most logical to play: 44...g5!? 45.hxg5 hxg5

Now after 46.fxg5 Kg6 47.Rc5 Rh8 48.Raxa5 Rxa5 49.Rxa5 Rh5 we have a quick draw. So White would have to try 46.f5. And now Black can activate the pieces. After 46...Rb7! 47.Rxa5 g4! White has an extra pawn, but Black will draw easily with obvious moves. 45.e4 Rh5 Here I found a way to create problems for him. This is not surprising after the last two moves. 46.Rac4! Rxh4 47.e5! The idea was not 47.Rc7 Rxc7 48.Rxc7 Rxf4 49.Rxf7† Kg8 50.Rg7† Kf8 with an immediate draw.

137

White’s threats look menacing and it is possible to get scared, which seemingly is what happened to my opponent. We are of course returning to the question: passive or active. When you are under pressure, the psychological pressure to be active is very real, with passive defence given an aura of helplessness that we all want to avoid. At the very moment when he had to be passive, he chose to be active and weakened his position irreparably. 47...g5? The curious thing is that passive defence in this position means establishing a fortress. Black has to wait for White to execute his plan. For example: 47...Rb7!? would have held the draw. (Black can also hold with 47...a4 and similar.) The point is that after: 48.Rc7 Rxc7 49.Rxc7 Rxf4 50.Rxf7† Kg8 51.Rg7† Kf8 52.Ke3 g5

138

White has no way to make progress. It is a simple fortress. In some positions Black can play ...h5 and in others his king comes to e6. Pawn endings are only lost if White gets his king to f5, but this will never ever happen. My original analysis of this game included lines where Black is pushing all the pawns and proving counterplay, narrowly making a draw, but looking at it now, I can see the futility of this. The white king is cut off and the black rook solidly anchored on f4, as the pawn constellation is strongly supporting the rook. From this it logically follows that Black is also lost after 47...h5? which to great satisfaction turns out to be true. 48.Rc7 Rxc7 49.Rxc7 Rxf4 50.Rxf7† Kg8 51.Rg7† Kf8

52.Ke3 White wins after both 52...Rf5 53.Ke4 and 52...Rg4 53.e6. Black simply cannot achieve the double defensive need of preventing the white king from advancing and being ready to meet e5-e6 139

with ...Rxf6. 48.f5 Rxc4 Black could have offered more resistance with: 48...Rb7!

49.e6? fails surprisingly to: 49...Rxc4 50.Rxc4 fxe6 51.fxe6 Kg6 52.f7 Rb8 with a positional draw. We had to spend some time analysing this position before we came to understand it. White has two winning methods. The more prosaic, but not necessarily easy, solution is 49.Rxh4 gxh4 50.e6 fxe6 51.fxe6 Rb8, where White has a few ways to do it.

140

Jacob was looking with the computer and he asked me to solve this position. I needed to concentrate to be able to find the win. My first intention 52.e7? Kg6 53.Rd6 Re8 54.Rd8 Kf7 55.Kf3 fails to 55...h3! and Black wins a vital tempo, meaning that the king will not be able to come up and deliver mate in time. The key idea is that the rook belongs on f4, so the clearest win comes after 52.Rc4! h3 53.Kf2 Kg6 54.Rf4 and the pawns cannot be stopped. Going back a bit, the win you want to find is this one: 49.Kd3! a4! This loses only to a very enjoyable variation. 49...h5 50.e6 Rxc4 51.Kxc4 fxe6 52.fxe6 Kg6 53.f7 Rb8 54.f8=Q! Rxf8 55.e7† is a nice touch. 50.Rxh4 gxh4 51.e6 fxe6 52.fxe6 Rb8 This is an unfavourable version of the game for White. But he is still in time. 53.e7 Kg6

141

54.Rd6! Re8 55.Rd8 Kf7 56.Ke4 h3 57.Ke5 h2 58.Rxe8 Kxe8 59.Ke6 h1=Q 60.f7# 49.Rxc4 a4

50.e6 fxe6 It is of course possible that Pelletier simply blundered and missed that after 50...a3 51.e7 a2 52.Rc1! there is nothing Black can do. White collects the a-pawn in slow motion. 51.fxe6 Kg6 52.e7 Ra8 53.Rxa4! Rb8

142

However, it is also possible he believed that this position was a fortress. But this is far from the case. 54.Ra6! h5 No better is: 54...Re8 55.Rd6 Kf7 56.Rd8 g4 57.Ke3 h5

White has two winning methods. He can put Black in perpetual zugzwang and collect the pawns, or he can go directly for mate with: 58.Ke4 g3 59.Ke5 g2 60.Rxe8 Kxe8 61.Ke6 g1=Q 62.f7# 55.Rd6 Rb2† Sometimes we make an extra move or two, simply because resigning was not part of the plan. 143

1–0 The theme of active and passive defence is worthy of a massive manual all on its own. This little chapter should not be seen as anything more than highlighting the theme and some of the dimensions of it. But if it has made a few of the readers think, it will have been a good start. And if one decides to write about this topic, please send me a signed copy of the book! But hurry up before we do it ourselves...

144

Chapter 4 A Bad Plan is Better than No Plan

Pentala Harikrishna

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

145

How should Black mobilize? (see page 81)

Black to play. Go as deep as you can. (see page 82)

146

Did you calculate this far? What next? (see page 83)

Black must retreat. But where to? (see page 85)

147

How should White continue? See all the options! (see page 87)

Time to convert the advantage! (see page 90)

148

Things can still go wrong. What is the best move? (see page 91) In the game in this chapter – a not very important or impressive game played against Indian star Pentala Harikrishna – we shall see how a harmless handling of the opening, accompanied by not very impressive plans, was able to outcompete a player without a serious plan. Harikrishna is first of all a very nice person. He has good technique and is excellent at the calculation of clear lines. Since he moved to Europe and played in more leagues and other tournaments, he has improved a lot. In November 2016 he joined the Top 10 for a brief period. At the time of writing he is in the Top 30. Wijk aan Zee 2014 was a horrible tournament for me until I made a slight comeback towards the end. Wijk is a long tournament and I had suffered a lot before this game, which was played in the last round. Luckily, I had won the day before, but still I wanted to forget about this tournament as quickly as possible. I have no definite insight into the state of mind of my opponent during this game, but I can share my impressions. From the start of the game he decided to play a solid system. Once he equalized, he avoided making tough decisions and instead waited for the “inevitable” draw to happen. The problem is that although Black is at least equal, White can still improve his position. In a few moves I managed to regroup my pieces and suddenly Black was a little bit worse. He did not react well to that situation either and got into difficulties. It looked to me as if he did not realize he was getting into trouble before the problems became serious. This is not unusual. It is very difficult to choose between various unpleasant scenarios. Some will be losing and some not. But all of them look bad. But the key lesson to take home is that it was his poor play when the position was good for him that led to the problems he faced and needed to solve. 149

The game was lost in a long string of planless decisions from moves 21-30. Boris Gelfand – Pentala Harikrishna Wijk aan Zee 2014 1.c4 Going through Harikrishna’s repertoire, I had noticed that he was less well prepared for the English. Obviously, this would not entirely nullify his repertoire, but it would narrow his options. Of course, there are many options after 1.c4. What is taken from Harikrishna is replaced with other ideas. But he did not play 1...c5 or 1...e5 often. 1...e6 2.Nc3 d5 3.d4 Nf6

The Carlsbad structure has always been considered pleasant for White – and solid for Black. In modern times the engines have changed the assessment of a number of key positions and breathed new life into the variation. 4.cxd5 exd5 5.Bg5 c6 Often the Carlsbad structure arises after 1.d4 d5 2.c4 c6 3.Nf3 Nf6 4.Nc3 e6 5.Bg5 Nbd7 6.cxd5 exd5, where White has already committed the knight to f3 and Black the knight to d7, so both sides have narrowed their options. 6.Qc2 6.e3 gives the possibility of playing 6...Bf5, which after 7.Qf3 Bg6 8.Bxf6 Qxf6 9.Qxf6 gxf6 is 150

very solid for Black, but dreadfully passive. 6...Be7 Recently 6...h6 7.Bh4 Be6 has been popular. There are always new ideas in the opening. 7.e3 Nbd7 8.Bd3

8...h6 In the last few years there have been a lot of different attempts around here. This is one of the main ones. Players are less worried about general considerations like not advancing pawns prematurely. Essentially, people like to experiment and work out what can be played and what should be avoided. 8...0-0 is a standard move, but here we can see the advantage of not having developed the g1-knight; people consider 9.Nge2 to be an achievement for White. There are a number of attractive plans, often connected with f2-f3, taking control of the e4-square. So, instead people either play like my opponent or the immediate 8...Nh5 9.Bxe7 Qxe7. It is a constant discussion if the additional move ...h6 is to Black’s advantage. If he does not include it, Black has a fairly straightforward plan in ...g6, ...Ng7 and ...Bf5, aiming to exchange the bad bishop. 9.Bh4 Nh5 10.Bxe7 Qxe7 11.0-0-0 Other options were 11.Nge2 Nb6 12.0-0 (as Leko has played a few times) and 11.Nf3 Nb6 12.00. The positions is open to multiple interpretations. Whichever path you choose to follow, you will be walking in the footsteps of a great player. 11...Nb6 12.h3 Be6 13.Nf3 0-0-0 14.Kb1 Kb8 151

15.Rc1 Initiating a fairly straightforward plan. The way I see it, all types of exchanges are fine for Black, but of course the most desired exchange is of the light-squared (bad) bishop. Black is very solid, so the best option for White should be to go for an attack. 15...Nf6 16.Na4 This is a bit simplistic. The simplifications seems to be pleasant for Black, but it is also hard to avoid them. I played this because I needed to act on the queenside and it is very hard to do without this move. Despite this, it was a novelty at the time of the game. Later 16.Ka1 was played. It seems likely that White will need this type of subtlety in order to put Black under pressure in this line. The key point of the move being that there is unlikely to be an unpleasant pin from f5 to b1. 16...Ne4 Harikrishna plays the position in a very straightforward way. It seemed natural for Black to take on a4 first and then play ...Ne4, transposing to the game. Later on, Black players have also played 16...Nxa4!? 17.Qxa4 Nd7!, when the critical position arises after 18.Rc3 Nb6 19.Qc2!? (19.Qa5 Nc4! is fine for Black).

152

17.Bxe4?! I did not think Black had lost anything by delaying the exchange on a4. White is certainly not about to take on b6; Black’s structure would only be improved. But it appears I could have played slightly better with 17.Nc5!? Bf5 18.Ne5, with the idea 18...Ka8 19.Nb3!, when White has actually managed to make things rather awkward for Black; f2-f3 and e3e4 may be on the way. There are also ideas with a4-a5.

And after 19...f6?! White has 20.Nxc6! bxc6 21.f3 Ng3 22.Bxf5 Nxh1 23.Qxc6† Qb7 24.Qxb7† Kxb7 25.Rxh1 with two pawns and a very pleasant position for the exchange. White is simply better.

153

17...dxe4 18.Ne5 The knight has nowhere to go to from d2. 18...Nxa4 18...Bd5 19.Nc5 Rhe8 was also fine for Black. 19.Qxa4

19...Rd5! This is a natural way to play. Black anticipates White’s attacking ideas and prevents them, while making space to include the rook from h8. During the game I was considering if I was actually threatening anything. Say that Black plays 19...Bd5 to control the c6-square. After 20.Rc3 Qg5!? 21.Rhc1 Qxg2 22.Ra3 a6 23.Qb4 you want to believe that there is a strong attack.

154

But after 23...Ka8, no way forward can be found. For example: 24.Qb6 Rd6! and White is lacking a path for the knight to join in the attack. Without this, he has lost a pawn and should quickly secure a draw by taking the perpetual check. But it would not make a lot of sense to rely on such calculations, when you have a reasonable and simple way to play, as Harikrishna did in the game. Without 24...Rd6, Black is in serious danger of getting mated. 20.Rc3 Rb5 21.Ra3 I understood well that Black’s position was fine. But it made a lot of sense to provoke a weakening.

155

21...a5?! Black is okay after this move, but still it feels as if White achieved something, however slight; it is a slightly superficial move. At the highest level, you should be careful about weakening your pawn structure in this way. During the game I became entangled in a web of complications, but failed to come to the right conclusion. 21...a6!! Instead 21...Qb4 would transpose if White takes on a7, but it is not clear to me that Black wants to exchange the queens in this position. Although the resulting endgame is certainly fine for him, it is quite likely that that middlegame holds some hope for optimism. 22.Qxa6 This is clearly the critical move. 22...Qb4 23.Qa7† Kc7 24.b3 This was the critical position. I was uneasy about it, but did not find anything concrete for Black. But the key point is that it is already clear that White is not the stronger side, and it is Black looking for a way forward now. 24...f6 25.Ng6

156

Here I rushed with the obvious 25...Bc4 in our analysis. Had I taken a moment to think, I would have seen that Black has no need for the rook on h8 at all. 26.Nf4 (26.Rd1 Be2! and Black wins) 26...Bd3† (26...g5? 27.Ra4! and White wins) 27.Nxd3 exd3 28.Rd1 Rd8 And Black has enough compensation for the pawn. Slowing down, we noticed the much stronger 25...Bf7!! 26.Nxh8 (26.Nf4 g5 27.Ne2 Qd2 and White’s position collapses. The queen will prove to be a most effective wrecking ball.) 26...Bc4 27.Rd1 Qc3 and White is simply lost. Black will play ...Bd3† next. This is incredibly difficult to find unless you suspect that Black is looking for a win. So, for this reason, White would instead have to play something like 22.Rc1, accepting that the advantage has already passed to Black. There is no attack on the queenside and coming up, Black will start probing White on the kingside, where the pawns will need a lot of defending. The knight will also find a lack of stability on e5. After the prophylactic 22...Rc8, Black is ready for ...f6, although he will not rush with it. The queen is likely to want to go to g5 first. Essentially, Black has options and the white defenders are offside. 22.Rc3 Sort of forced. Black’s last move creates the threat of ...Qb4, which would win against 22.Nc4? and lead to a highly promising position after 22.Rc1?!. 22...Rd8 This is entirely logical. Black brings the rook into the game and clearly has no problems. I did not even consider 22...Qg5, as suggested by the engines.

157

To me it was a rather unnatural way to play. Looking a bit deeper, it turns out that most lines end with a natural perpetual. White’s attack is not strong enough to lead to mate and Black achieves something on the kingside. 23.Rhc1 leads to a perpetual in several ways. 23...Qxg2 is met with 24.Rxc6 Rxe5 25.dxe5 bxc6 26.Qxc6 Rc8 and only the checks remain. While after 23...f6 the following line is pretty, but not too difficult. 24.Rxc6 bxc6 25.Nxc6† Kb7 26.Nxa5†! Rxa5 27.Qc6†, also with a draw. 23.g4 Qd8, where Black is planning to follow up with ...h5, shows a structure where White is already in danger of being worse, technically, and thus has to go for dynamic counterplay at all costs. This understanding inevitably leads to the following variation: 24.Rhc1 h5 25.Rc5 Qb6 26.b3 hxg4 27.hxg4 Rxc5 28.Rxc5 Qb4 and it is time for White to bail out with a draw: 29.Nxc6†! bxc6 30.Qxc6 Qe1† 31.Rc1 Qb4 32.Rc5 With a repetition, as long as Black stays away from temptation: 32...Rh1†?? 33.Kc2 and White wins. My key point here is that the position is already equal. Black does not need to equalize. But if he plays actively, probably White must. 23.Rhc1 Qb4?! This is a step in the wrong direction. All it achieves is forcing White to play 24.Qc2, which he was going to do anyway. And later we shall see that the queen is poorly placed on b4. I felt that 23...f6! was a serious option.

158

If White plays 24.Ng6, which he should, Black has absolutely nothing to worry about. He has 24...Qb4 for example. His bishop is not “bad” and the knight is not finding any stability. Black has completely equalized. And also, after 24.Nc4 Rdd5! 25.Qd1 Rf5, it is clear that White has been pushed into a defensive position. The key point is that the sacrifices are rather dodgy: 24.Rxc6 Rdd5! 25.Rxe6 Qxe6 26.Nc4 Qe7 and again White has to prove compensation, but most likely it is not really there. 24.Nxc6† bxc6 25.Rxc6 Rdd5 26.Qc4 Rb7 Now after 27.Qa6 Qb4! 28.b3 Bd7 Black wins. So White has to play 27.b3, which is a bad sign. After 27...Rd6, White is maybe still equal, but he is the one who has to prove compensation. 24.Qc2 Bd5 25.Qe2 This is a reasonable play. White is preparing Nc4 with the idea of taking back and putting a rook on c5, preventing ...Rd5 on account of the tactic Qxb5. This was successful in the game, but is really nothing special. At this point I could also have tried: 25.Nc4!?

159

The idea is that after 25...Qe7, White has 26.a4, winning a pawn. And the direct threat is 26.a3 Qe7 27.a4. So, for this reason Black could be compelled to exchange the bishop here: 25...Bxc4 26.Rxc4 Qe7 27.a4 Without this move, White has nothing much. (27.Qe2 Rdd5 and Black is fine) 27...Rg5 28.g3 Rdd5 29.Qc3 White will regroup on the queenside before deciding how to continue. There are ideas with Rc5 available in some positions, with the intention of taking the a5-pawn. In others White can even contemplate b2-b4. I would not say that White is better, but Black’s rooks are rather artificially placed on the 5th rank, and he was forced to part with his bishop, so there is an argument for some progress having been made.

160

25...f6?! An unnecessary concession, which is closely linked with the next move. I think the right attitude in this position would be to change direction, play 25...Qe7! and keep a solid position. I cannot see that White can do too much. The key thing to understand about these minor concessions is that sometimes we make them and it seems fine, but ten to fifteen moves later, we have to show just a bit more accuracy to avoid having to drift into a position with real problems. The computer will claim that the mistake was later, because accurate moves still defend, but to a grandmaster it already feels unpleasant. Once Harikrishna equalized, it was difficult for him to find a path forward. The exchanges seem harmless to Black superficially, but as we will see in the game, eventually microscopic concessions led to real problems. 26.Nc4 Bxc4 As said, I don’t like this much. The knight has no good squares, so why exchange it? 26...Qe7! was still better, but every small concession creates problems for Black down the line. 27.Rxc4

27...Qe7? Here Black should have played: 27...Qd6 Toying with the idea of ...Qh2 followed by ...Rd5 in various lines. White has a lot of reasonable moves, all giving him some pressure. 28.a4, 28.h4 and: 161

28.Qc2 Qe6 29.Qd2 Black is struggling to defend the a5-pawn. White’s next move could be Rc5 and against 29...Rdd5 comes 30.a4 when the rook has no great squares available. Active counterplay on the kingside is not without risk. The following variation is just an illustration of what could happen and should by no means be seen as conclusive, but it shows the issues Black is facing. 29...Rg5

30.h4! Rxg2!? 31.Qxa5 Rd5 32.Qb6 Black has to play a series of forced moves. 32...Rb5 33.Qd8† Qc8 34.Qd6† Qc7 35.Qf8† Qc8 36.Qa3 Rxf2 If you ask Stockfish, the position is still equal. But to a human it is clear that Black is facing much more challenging problems. Stockfish is still optimistic about its chances after: 37.b3 Kc7 38.Ra4 Qd7 39.Ra8 g5 But after: 40.Qf8 The machine soon realizes the amount of trouble Black is in. 40...gxh4

162

41.Rc4! White is threatening 42.Qb8† followed by Rca4 and Ra6† with a deadly attack. 41...Kb6 42.Rd8 Qe6 43.Rb4 Kc7 44.Ra8 The black king is obviously going to be mated. 28.Rc5 Rxc5 I believe that Harikrishna had lost his sense of danger and had missed that 28...Rd5? (with the intention of exchanging rooks and keeping control of the d5-square) would be met with 29.Qxb5!, when White wins. Obviously, Harikrishna understands everything said above, but missing this small tactic suddenly exposes the concessions made for what they are. 29.Rxc5 Qc7? The computer advises 29...Rd5!, but it does not look reassuring after: 30.Rxd5 cxd5 31.Qb5

163

31...Qd8 32.g3 looks unpleasant. Probably it is a draw objectively, but this has reduced influence on how the game would proceed. Black would have to find a fine move in 31...Qd6!, where 32.Qxa5 would be met with 32...Qh2, ...Qg1† and ...Qxf2 with counterplay. I would have had to try hard to find some way to make Black’s life difficult. The best I can come up with is 32.a4!?, where there is no 32...Qh2 on account of 33.Qxd5. Instead Black would most likely play 32...b6, when he is very close to equal, but White can still probe. A continuous theme in this book is the difficulty of defending; especially situations where we have to choose between passive and active defence. A longer description can be found in Chapter 3. In this game we can see how Harikrishna tried to force a draw with exchanges, at a time when there was nothing wrong with his position. But because of an oversight, he now found himself faced with a situation where he had to find active moves to hold. The adjustment is not easy. But like with a counterattack in football, you get little chance to readjust in chess. This is a clear example of a turning point; from safely equal and then after a few unnecessary concessions, made with the intention of forcing a draw, Harikrishna will have seen his position going from totally fine to difficult. And at the same time, he would lose a lot of his confidence, second-guessing his conclusions. And we are already in the fourth hour of play in the last round of a long tournament. The narrative we have attached to the previous eight moves is not the experience Harikrishna had in the game, nor was it mine. It is a description of the reality that emerged after deep analysis over several sessions.

164

30.Qg4! b6? This is a really poor move and shows that my opponent was shaken. The weakening of the pawn structure is catastrophic. Very quickly Black’s position becomes untenable. If Black had played: 30...Re8

I would have gone immediately for a better rook endgame with: 31.Qg3 31.d5? would be a blunder, due to 31...Re5. 31...Qxg3 32.fxg3 a4 33.Kc2! 33.Rc4? is greedy and punished by 33...f5! with the idea ...Re6-g6. 33...Rc8 Necessary because after 33...Kc7 34.Ra5 White wins a pawn. 34.Rc4 b5 35.Rc5 Kc7 36.Kc3

165

Although Black has defensive resources, I am optimistic that I would have been able to win this endgame. But even clearer is 31.b3!, as prescribed by the engines.

White is threatening 32.Qg3!, winning a pawn, and Black has no pleasant way to deal with this. After 31...b6 32.Rc4 Ka7, I would be tempted to play 33.d5, with similar lines to the game. Although even stronger is the strategically pleasing 33.Kc2! Qb7 34.Kd2 Re7 35.Ke1 to defend the f2-pawn before playing d4-d5. White is winning. 31.a4 is less effective, as after 31...Re7 32.Qg3 Qxg3 33.fxg3, Black has the elaborate 33...g6! – preventing Rxa5, after which Black’s position is not so bad. The idea would be ...Rc7 and ...b6. 166

For this reason, Black had to find: 30...Rd5! 31.Qxe4 Qh2 32.Rxd5 Qg1† 33.Kc2

33...cxd5! This is the difficult move to imagine. 33...Qxf2†? looks so tempting, but the pawn ending after 34.Kb3 cxd5 35.Qf4†! Qxf4 36.exf4 is hopeless. During the game I believed that White would have good chances. Actually, it requires no great discovery to convert the advantage. 36...Ka7 (36...f5 37.Ka4 b6 38.Kb5 Kb7 39.h4 would see Black run out of tempos very soon. For example: 39...h5 40.a3 Kc7 41.Ka6 Kc6 42.b4 and the b6-pawn will fall very quickly.) 37.Ka4 Ka6 38.f5! b5† 39.Kb3 The king goes to h5 and penetrates the black position. Black can make all types of concessions, but he will obviously be in zugzwang. This is not a position for calculation or analysis, but for logic. White wins easily. 34.Qf3 a4 35.g4 35.Kd2!? Qf1 36.g4, with something to play for, also makes sense. White is a pawn up, but Black’s queen is well placed, so the defensive potential is great. 35...Qe1

167

We analysed this ending for a bit. Stockfish supports Black with a 0.00 evaluation, but if you make human moves, you can see that White can make a lot of progress and offer Black a lot of difficulties. If Stockfish can solve them or not does not matter a lot. 36.Qxd5 Qxf2† 37.Kd3 Qf1† 38.Kc3 Qxh3 39.Qe4 Qf1 40.Kb4 Qe2 41.Kc5! A theme in queen endings is that passed pawns are very powerful, while extra pawns close to their starting squares (see Black’s queenside) do not influence the result that often. 41...Qxb2 Or 41...Qa6 42.Qe8† Ka7 43.b3 with an advantage for White. 42.Qe8† Kc7 43.Qf7† Kb8 44.Kd6! a3 45.Ke7 Qb4† 46.Ke8 Qb2 47.Kf8

It is clear that White has made great progress. But deep analysis will show that it is not enough.

168

47...g6 48.d5 Qxa2 49.Qe8† Ka7 50.Qa4† Kb8 51.d6 Qe6 52.Qxa3 f5 53.Qc5 Qf6† 54.Kg8 Qd8† 55.Kg7 Qd7† 56.Kxh6 fxg4 57.Qe5 g5 58.e4 Kc8 59.Kxg5 g3! Accuracy is needed. 60.Qxg3 Qg7† 61.Kf4 Qf6† And Black makes a draw. 31.Rc4 Re8

32.d5 c5 Black’s position is so horrible that he should have considered 32...Qf7 33.Rxc6 Qxd5 34.Rxb6† Ka7 35.Rb3, when White has an extra pawn without any compensation. But in practical terms, maybe this would give Black better chances than the game. 33.Rxe4 Rd8 34.Qh5 I am happy with this move. It seems very convincing to me. During the game, I was thinking that he was hoping I would play: 34.Qe6 Qb7

169

35.d6, when after 35...Qc6 Black wins back the pawn, although his position is still unpleasant. But even stronger was 35.Kc2! Rxd5 36.Qe8† Qc8 37.Re7 Qxe8 38.Rxe8† Kc7 39.Re7† transposing to the game, making this a more convincing way to play. 34...Qd7 This was Black’s idea and it is quite logical, although as it turns out, hopeless. Black should have tried something along the lines of 34...Qh2 35.Qf3 Qh1† 36.Kc2 Qf1, which is not very convincing, but where moves still have to be made and mistakes can be made in the conversion.

170

35.Re6! An important move. 35...Qxd5 36.Qxd5 Rxd5 37.Kc2! A good display of technique. Black is not allowed to get counterplay with ...Rd2. 37.Rxb6†? Ka7 38.Rb5 Ka6 39.a4 Rd2 40.Rxc5 Rxf2 41.Rc2 gives White an extra pawn, but the resulting rook endgame looks more drawn than won to me. 37...Kc7 After 37...Rg5 38.g4 Kc7, White can force the pawn ending immediately with 39.f4 or prepare for it with 39.a4. He is winning either way. 38.Re7† It was possible to play 38.a4 and prepare for the pawn ending, but Black can play 38...Rd7 and although the position is lost for Black, we should never miss the chance to play a winning pawn ending. 38...Rd7 39.Rxd7† Kxd7

171

40.a4 This is a move you should play instantly if you were awoken in the middle of the night and asked about this position. Luckily, this is move 40 and requires no thinking. I am unable to imagine there could be a scenario where a move like this, where two pawns will control the three pawns in this way, would be the wrong direction. Study composers should see this as a challenge. 40...Kc6 After exiting the time control, we can see that White has a very promising pawn ending. I should point out that when I entered this, I had not calculated it till the end. White has a healthy pawn majority on the kingside and Black has an unhealthy majority on the queenside. Thus, White has good winning chances. If I had had major alternatives, I would perhaps have gone deeper, but since none existed, saving the energy to solve the real problems later was more rational. 41.Kc3 b5 41...Kd5 would lose quickly to 42.e4† Kxe4 43.Kc4. 42.axb5† After the game Alex berated me for not playing 42.e4?, which he thought won more simply, but it was fortunate that I did not do so. There is no win for White after 42...bxa4 43.Kc4 g5.

172

For example: 44.g3 h5 45.f3 Kd6 46.Kb5 Ke5 47.Kxc5 h4 48.gxh4 gxh4 49.Kb5 Kf4 50.Kxa4 Kxf3 51.Kxa5 Kg3 52.b4 Kxh3 53.b5 Kg2 54.b6 h3 55.b7 h2 56.b8=Q h1=Q with an evident draw. Usually Alex is right, but not this time. 42...Kxb5 43.e4 g5 44.g3 h5

45.f4? But here I definitely messed up a little bit. This move allows Black to create counterplay. 45.f3! was a subtle alternative. 45...a4 (45...c4 46.h4! is very similar and 45...h4 46.gxh4 gxh4 47.b3 173

does not offer Black any chances either) 46.h4!

This is the key move. It is not very obvious to me, but it works. For a start, it threatens to play e4e5-e6 with the idea of giving a check on e8, in lines such as 46...c4 47.e5 gxh4 48.e6!. And after 46...gxh4 47.gxh4 c4 48.f4 Kc5 49.f5! White wins because of mutual zugzwang. 45...gxf4?! We analysed the game afterwards, but my memory of this is somewhat hazy. So, I cannot tell if my opponent had already given up all hope and was just making moves. Harikrishna should definitely have grabbed the chance to play: 45...h4! 46.gxh4 gxf4

174

I would certainly have thought for a while in this position. Let’s look at the options. 47.b3? is not something you would really believe in, but at the board, I would check it anyway. It turns out that Black even wins, and in an attractive way: 47...f3 48.Kd2 a4 49.bxa4† Kb4!! The key point is that ...c3† comes with a tempo and no matter where the white king goes, one of the bishop pawns will promote with a check again, winning another tempo. Here there is a curious thing with Stockfish. It insists that White should play 47.e5, with the idea that 47...fxe5 48.h5 f3 49.Kd2 a4 50.h6 a3 51.bxa3 c4 52.h7 c3† 53.Kxc3 f2 54.h8=Q f1=Q 55.Qxe5† is a theoretically winning ending.

Actually, it is mate in 103 moves. This tells us a lot about a different perspective. I would 175

definitely not feel convinced that I would win this position. The natural way to play would be: 47.h5 f3 48.Kd2 a4 49.h6 a3 50.bxa3 c4 51.h7 c3† 52.Kxc3 f2 53.h8=Q f1=Q 54.a4†

Harikrishna would not fall for 54...Kxa4 55.Qa8† Kb5 56.Qb7† Ka5 57.Qb4† Ka6 58.Qc4† with an exchange of queens and a win for White. Rather, after 54...Kc6 a lot of moves would have to be played. In a practical game I would not call this winning, but a massive advantage. This is not because I do not understand that the two extra pawns should convert, but because we inevitably align our play to our evaluation. And calling this a big advantage would help me to keep my focus and not underestimate Black’s chances. Black is lost, but chess without chess players is not what we are talking about. 46.gxf4 c4 46...a4 would have given me a chance to mess up.

176

White is not winning after 47.e5? fxe5 48.fxe5 Kc6 49.Kc4 h4. However, it was my intention to play 47.h4!, which is totally winning. For example: 47...c4 48.e5 fxe5 49.fxe5 Kc5 50.e6 Kd6 51.Kxc4 Kxe6 52.Kb4 Kf5 53.Kxa4 and the black h-pawn only arrives to the third rank, which is not enough to make the draw. 46...h4 47.b3 is similar; White wins easily. 47.e5 Black resigned. 1–0 Petrosian told me to never make a move without an idea – even in blitz. It is important to have an intention with your moves. And ideally, it should be with the intention of reaching a goal. This is called a plan and is a useful weapon to include in your play.

177

Chapter 5 Long Games with Increment

Aeroflot Open 2016, with Bartel and Najer

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

178

A nice prophylaxis exercise (see page 100)

Time to regroup (see page 103)

179

In time trouble, how should I have played? (see page 103)

How should White organize the pieces? (see page 106)

180

How to preserve the advantage? (see page 113)

How should Grachev look for counterplay? (see page 115)

181

At this point I missed a big shortcut (see page 120) In this chapter we shall look at some longer games with long increment finishes. This is a different experience from when I was young and we had adjournments and a new time control every twenty moves. I know that many, if not most, active grandmasters today have never had this experience and have always considered games with additional moves after move 40 as long games. The time control matters. A lot. And for this reason, we will in this chapter look at some very long games played with increment in open tournaments, where a victory is far more necessary than in a closed tournament, where ‘plus two’ is often good enough to win the event. The games in the Aeroflot Open 2016 were all played with a super-classical time control. 100 minutes from the start, another 50 at move 40 and finally an additional 15 at move 60. All seasoned with an increment of 30 seconds per move. The organizer of Aeroflot was Alexander Bakh, who at age 77 (at the time) would not be tempted to introduce modern nonsense. On top of this, many of the arbiters were past retirement age (maybe around 70) as well. I recognized a lot of them from my junior tournaments, and everybody knows they know what they are doing. Boris Gelfand – Rinat Jumabayev Moscow 2016 Jumabayev is a young player from Kazakhstan. He is a very polite guy. I had previously played two horrible blitz games with him with the score 1–1. In this tournament he had beaten Korobov and Vallejo convincingly to reach the top of the tournament. This game was played on Board 2, so a lot 182

was riding on the result. Since we played, I have had more contact with him and I think highly of him both as a player and a person. While we were working on this book, I had the chance to play him in the inaugural Grand Swiss on the Isle of Man 2019, where we drew in 81 moves in a batshit crazy game. We also drew in the 2018 Olympiad in Batumi. 1.d4 Nf6 2.c4 g6 The Grünfeld is his main opening, but he varies a lot, so I had decided not to focus too much on opening preparation, but instead played my main anti-Grünfeld weapon, focusing on a long game. 3.g3 c6 4.Bg2 d5 5.Nf3 Bg7 6.0-0 I used to think that if White wants to play the lines with 6.cxd5, it is better to do it before castling. But time and engines have killed everything. 6...0-0 7.Nbd2 This move was played against me by Lev Polugaevsky in the famous tournament in Reggio Emilia 1991. I won a really good game. Another possibility is 7.b3, with the idea to decide later between putting the knight on c3 or d2. These lines became popular after Boris Avrukh wrote about them in Grandmaster Repertoire 2 almost a decade ago. But it allows 7...dxc4 8.bxc4 c5, which is considered fine for Black. I played it a few times with both colours.

7...Na6 A rather rare line. Usual are 7...a5 or 7...Bf5. I was surprised by this move, as the classics have taught us that the knight should not be put carelessly on the edge of the board.

183

8.b3 c5 9.Bb2 My feeling was that White should be better here, as Black has lost a tempo with the c-pawn and his knight is weirdly placed on a6. 9...b6 10.dxc5 This seemed pretty logical. 10...Nxc5 10...bxc5 11.cxd5 Nxd5 12.Bxg7 Kxg7 13.Rc1 would give White a persistent advantage. 11.b4! This is a bit annoying for Black. 11...Ne6 11...Na6 12.Qb3 and White slightly better.

At this point I spent 14 minutes to check the forcing lines. One of the things that makes chess a great game is that you cannot be certain exactly when you have to act and when you have to build up. Certainly, guidelines exist (improve your positions slowly, when you have more scope for improvement than your opponent does – for example), but chess is too complex to be nailed down by one-liners. 12.cxd5 Nxd5 13.Bxg7 One of the things I had been deliberating about was if I had to take on g7 or play 13.Qb3!? Bb7 14.Rfd1, which also looked attractive. In the game I decided that it would be more flexible to decide if I wanted to have the queen on b2 later, by giving a check, than to allow Black to make the decision 184

for me. 13...Kxg7 13...Nxg7 was Leko’s choice. 14.Qb3 Nf6 15.Ne5 Be6 16.Qb2 Bd5 17.b5?! But this last move was very peaceful. 17...Bxg2 18.Kxg2 Ne6 19.Nc6 Qc7 20.Nf3 ½–½ Ki. Georgiev – Leko, Cacak 1996. Instead 17.e4! Bb7 18.Rfd1 would have given White a serious advantage. 14.Qb3 Bb7

15.Rfd1! Because Black has problems finding a space for his queen, it is easier for White to activate the rooks. So it makes sense to do this before taking any further actions. 15.Bh3 was possible, but it is not clear what it achieves after 15...Ndc7, where Black could think about solving the problems with his queen with ...Qd5-h5. Also, there is the option of 15...Nef4!? 16.gxf4 Nxf4 17.Bg2 Nxe2† 18.Kh1 Nf4, where the computer might tell us that White is a bit better, but to the human eye, it looks like compensation. From what I have seen of Jumabayev, he thrives far better in dynamic positions, so I think allowing this would be a poor strategy. I want to make sure this nuance is properly understood. From the position I have, I do not want this position with White. White already has a very pleasant position, so allowing something like this makes no sense to me. 15...Nf6! Getting the knight out of trouble, forcing White to make choices immediately. 16.Nc4 185

I thought for a long time, being totally unsure if I should play this or 16.Qb2!?. Both moves are promising, so it was hard to choose then and is hard to choose now. In both cases White is better. ‘Plus equal’, meaning that the position is comfortable, but nothing dramatic. 16...Qc7

But at this moment I am not satisfied with my play! This is clearly an important moment. If Black is able to bring his rooks to the centre, he will be fine. So White should do something. I did consider some interesting things, but instead I blundered. 17.Rac1?! I tried hard to make the following line work: 17.Nce5! Rad8 If he played 17...Bxf3, I would be happy. The knight will eventually get to c6 and dominate Black. This is what a Catalan player dreams of.

186

18.Nxf7!! Bd5 The only move. 19.Rxd5 Rxd5 20.e4 Rb5! I did not consider this move, but apparently it is the best one. The idea is ...Nc5 and it sort of holds Black’s position together. I expected 20...Rh5 to be the only move. 21.Qxe6 Rxf7

I thought that Black was OK. But when I asked the computer it gives 22.Nd4! and White is much better. I am not sure if this was something I could have found. It is a bit surprising that things are going so well for White. Intuitively it does not seem to me that it should be overwhelming. 21.a4 Nc5 22.Qc4 Rxb4 23.Qxb4 Rxf7 187

White is of course a bit better here. The two pawn islands against three and the weaknesses in the black position. I tried to convince Jacob that this position should go into his files, but he seems to have the opinion that whatever goes in our books, should not go to his files. At times when I am sitting at the board, contemplating a difficult position, I can get distracted by the thought: “I need to tell Jacob about this position so he can put it in his files!” Over the last decade I have solved maybe 10,000 positions from Jacob’s files (a small part of these were published in his series Grandmaster Preparation and the coauthored Quality Chess Puzzle Book, published under the other author’s name, John Shaw). I should also give you this variation, which is by no means obvious: 17.Bh3!? Bd5! 18.Rxd5 Nxd5 19.e4 Nf6 20.Bxe6

20...b5!! 20...fxe6 21.Ng5 would win a pawn for White. 21.Nb2 The sharp line 21.e5 Nh5 22.Na3 fxe6 23.Nxb5 Qb7 24.Nbd4 Kh8 offers White nothing. 21...fxe6 22.Ng5 Qb6 23.Nxe6† Kh8 24.Nxf8 Rxf8 Black has enough counterplay. After: 25.Nd3 Nxe4 26.Qb2† Qf6 Black is not worse. 17...Rac8 18.Qb2

188

At this point I was very happy and it was not until the next morning that I realized my mistake. 18...Qb8?? If my mistake on move 17 was a blunder, then what do you call this? My opponent told me the next day that he had seen almost everything, but believed he had all the time in the world and did not need to rely on tactics. He totally lost his sense of danger, which is perhaps a mistake of the young, who have yet to make as many horrible mistakes as an “experienced” player like me. I want to commend my opponent though. Once he made a mistake, he defended stubbornly. It is one of the characteristics of modern chess. Gone are the days when an opponent would have the decency to roll over and die the moment he made a mistake. The attitude has totally changed and you have to overcome lots of problems in order to win a game. It does not matter if you are playing against a Grandmaster or an accountant. No one gives up without a fight anymore. This is partly the fault of computers, who have shown us so many defensive possibilities in positions we would previously have believed to be totally lost. Black should play 18...Rfd8!, with the important point that 19.Nd6?? drops a piece to 19...Rxd6!, something we both missed. This underlines the necessity of finding 17.Nce5! earlier and that this position belongs rightfully in Jacob’s files. It is such a rich position. 19.Bh3!± Jumabayev had completely missed this move.

189

19...b5 19...Bxf3 20.exf3 does not make things better for Black. 20.Bxe6 fxe6

21.Nce5 I think I found quite a good plan. I wanted to find something more conclusive at first, but at some point I realized it was not there and instead focused on how to optimally organize my pieces. 21.Na5 was also attractive. 21...Rxc1 22.Rxc1 I would not allow 22.Qxc1 Rd8 when Black is getting a tiny bit of freedom. 22...Qd6 This is forced to defend against Nd7. 23.Qd4 Bd5 Any simplification will be unpleasant for Black. 24.a3

190

My intention was to follow up with Ne1-d3 and f3&e4. This urged my opponent to look for activity. 24...a5! There is no downside to this move. Either the game becomes sharper or Black gets to exchange his weak a-pawn. 25.Ne1 Of course I could take, but I was not convinced that it was better. My position is so pleasant and easy to play that it makes no sense to sharpen the position without a big reward. If you are a computer, you would definitely be tempted to go for the complications: 25.bxa5 Qxa3 26.Ra1 Qd6 27.a6 Ra8

191

28.a7 b4 Despite the computer’s enthusiasm, it is not clear to the human eye that White is going to win here. It is just as possible that the queenside pawns will be exchanged, which will bring the draw dramatically closer. Based on this, I am happy with my decision. What I played was easy to evaluate and play, and I knew what I was going to do. 25...axb4 26.axb4 Bc4 I thought he would play: 26...Ra8 27.N1d3 Ra2 My plan was to play: 28.Qe3 Here a nice idea is: 28...Be4 White has to play accurately to keep his advantage.

192

I hope I would have found: 29.Kf1!± A simple move, protecting the e-pawn. White is in no rush; the weaknesses will be there in a few moves as well. And a few moves after that too. 29.f3 is a bit too rushed. 29...Bxd3 30.Nxd3 e5! and Black manages to create counterplay. The trick is 31.Nxe5 Nd5. So White should play 31.Qxe5, when the endgame is closer to a draw than anything else. 29...Bxd3 30.Nxd3 Ng4 This does not lead anywhere. White could play: 31.Qe4! Nf6 32.Qb7 When it is time to collect a pawn and head for a more technical phase of the game. 27.Qb2 Qd5 Against 27...Bd5 I wanted to play 28.f3, when things are only getting worse for Black. 28.Ng2 The knight will be strong on both f4 or e3. 28...Rc8 28...Ra8 29.Ne3 and White wins a pawn without making any concessions.

193

29.Nf4 I decided to play this, as I liked the forcing line in the game. I felt that White’s pieces were displaying great harmony. I did not see any threats, but was also not sure I needed any. Black’s problems are permanent. But I was short of time and mainly focused on not allowing any random counterplay. 29.Ne3 Qe4 30.Nf3 was also interesting. White has so many options here. 29...Qe4 30.f3! Qe3† 31.Kg2 Kg8 31...Rd8 would lose to 32.Rxc4!.

194

32.Nfd3! Planning to place the knight on f2, where it can even consider going to g4 later. 32...g5 Looking for counterplay. 33.Nf2 It is very difficult to find a move for Black here. I am apparently threatening 34.Nfg4 Qa7 35.Qd2! and Black cannot defend the g5-pawn. 33...Rd8! Jumabayev realizes that he has to give up a pawn to continue the game. 34.Nxc4 bxc4 35.Rxc4 At this point my remaining task was to pass the time control without messing up too badly. 35...Kf7 36.b5 h5 37.Rc3 Rd2 38.Rc2 So far so good. 38...Rd8 Also 38...Rd7 39.Nd3 and 38...Rxc2 39.Qxc2 Qb6 40.Qb2 are hopeless.

195

39.Nd3? I had two moves to make. Of course I should have repeated the position with 39.Rc3! Rd2 40.Rc2 Rd8 and then I would have had time to think. The objectively winning move would be: 39.b6! Rb8 40.b7 Qa7

41.Rc7 The natural move. 196

Whether this had been move 41 or 43 would not have made a difference. I would have had the time to see ideas such as 41.Nh3! Qxb7 42.Nxg5† which wins easily. What required a bit of time was to see that Black cannot immediately force the rook away from c7. 41...Ne8 41...Nd5 loses to the same moves. 42.Rc8! Rxb7 43.Qh8 Nf6 White has so many tempting options here. I guess I would have gone for: 44.Qh6 e5 45.Ne4 White wins. This is very easy to follow with a calm state of mind, lots of time on your hands and when the principle is available to you. In contrast I had seconds, really wanted to win the game and was nervous I was losing control, with no awareness of other possibilities. I wanted to queen the b-pawn! Had this been move 41, I would have found this win in five minutes or less. It is a horizon problem. When you have no time, you have no horizon. 39...g4?! This gives White a chance to get back in control. Black could have made the game much more complicated with 39...h4!, giving me little time to decide how to deal with all the threats.

The computer says that the best move is 40.gxh4, which looks entirely counterintuitive. I would rather imagine that I would have chosen something cautious like 40.Nf2, where 40...hxg3 41.hxg3 g4! would confront White with serious challenges. The advantage has diminished. 197

Importantly, White should not get tricked: 40.b6? h3†! 41.Kf1 (41.Kxh3? Qg1! and White is mated)

41...Rxd3! 42.exd3 Qxf3† 43.Ke1 Qh1† 44.Kd2 Qxh2† 45.Kc3 Qxg3 and it is Black who is playing for a win in this complex ending. 40.Ne5†? is also a blunder. Mainly because 40...Kg7 41.Nc4? h3†! leads to a quick checkmate. Instead White can play 41.Qa1, but the advantage is more or less gone. 40.Ne5† Kg7 41.Nc4 Qc5

42.Qa3 198

When I played this, I thought I would be winning easily and the game would be over soon. But it turns out that I am one move short and the knight is a bit out of play on a3. If it could stay on c4, my judgement would have been right. Or if I had time for Rc5 or Rc6, consolidating before attacking his other weakness. The computer gives 42.Rc1!? to prevent ...Rd1. 42...Rb8 43.b6 and it would be time to swing the queen to the kingside to put pressure on the weaknesses there. Again, this is easy to understand when underpinned by such a strong authority as Stockfish, but to decide to do this in a game is not elementary. 42...Qxa3 43.Nxa3 Ra8! Black cannot waste any time. During the game I was not sure this was the right way to play, but now it is hard to argue against it. Black chooses to activate his strongest piece and leave it for the knight to take care of the b-pawn. Very classical thinking. 43...Nd5?! 44.Rc6 and White is in control. The following line is curious, but maybe not too important for understanding the position well. 43...gxf3†?! 44.exf3 Nd5 45.Rc5!? Ne3† 46.Kf2 Rd3

47.b6! Rxa3 48.Rb5 Nd1† 49.Ke1! Nc3 50.b7 Nxb5 51.b8=Q Ra1† 52.Kf2 Ra2† 53.Kg1

199

There is no chance for Black to establish a fortress here. All his pawns are weak and there is no credible counterplay against the king. 44.Rc3 This feels like the most natural move. 44.Nb1!? is the computer’s choice. 44...Nd5 45.Rb3 This is where the rook obviously belongs. Now my opponent spent a long time. This endgame is very complicated and both players found it hard to work out which were the right choices.

200

45...gxf3† This is not that obvious a move. But I believe that White is winning, even though the conversion has not been clear sailing. 45...Nb6 was the alternative. I was planning to play: 46.fxg4 (46.h4!? to encourage Black to take on f3 is also interesting) 46...hxg4 47.Nc2 Ra2 48.Ne3 Ra4 (48...Rxe2† 49.Kg1 should be winning for White, although there is a lot of play remaining. But three passed pawn should count for something.) 49.Rc3 Rb4 50.Rc5 White has excellent chances of winning. But due to my mistakes earlier, the win is neither easy nor guaranteed. 46.exf3 Nb6 47.Kh3! This move has several advantages. The king becomes active and Black loses the check on a2, which can be a useful tempo in many positions. Jumabayev however continued to fight hard against the inevitable darkness. 47...Rc8 48.Nb1 I thought the knight would come into the game now. 48...Rd8! A nice little touch. Unfortunately for Black, I had a good way to deal with this.

201

49.Re3! Winning an important tempo. 49...Kf7 Here I spent most of my time. I found a very practical solution. 50.f4!? This is a cautious move that certainly ruins nothing. I wanted to play 50.Nc3, but after 50...Nd5 51.Nxd5 exd5, I was not sure at all that White is winning. 52.b6 d4 53.Rb3 This leads to a forcing sequence. 53...d3 54.b7 d2 55.Rb1 Rb8 56.Rd1 Rxb7 57.Rxd2

202

This endgame is of course promising, but it is not easy to see in advance if White is definitely winning here. Nor, if we assumed that he is, how easy it would be to win the game. Afterwards it is easy to say that 57...Rb3 58.Kh4 Rxf3 59.Kxh5 is mate in 50 moves, just by looking at your mobile phone. But during the game this certainty is only available to the everjudgemental spectators. And after 57...e5 58.Kh4 Rb3 59.Kxh5 Rxf3 60.Kg4 White mates in 35 moves. The computer also gives 50.Kh4 as a winning move. But during the game I was afraid that it would involve walking into an obscure mating net. When you are thinking about converting your positional and material advantage, you do not want to be distracted by such things as getting mated! Remember that we are now in the sixth hour of play. 50...Nc4 During the game I was most concerned about 50...h4, but the computer can find harmony in any position. After 51.g4! Rd4 52.Kxh4 Rxf4 53.Nd2! the knight is coming to f3/e4-g5, winning the e6pawn. 51.Re4 Rc8 52.Re2! At this point I am getting close to finding harmony with my pieces. 52...Nd6 I was thinking that he should try something active at one point, like 52...e5 here.

203

But after 53.Rc2! I was very confident in my chances in the knight ending. White just wins after: 53...exf4 54.gxf4 (The computer of course finds a direct win: 54.Nd2! Nd6 55.Rxc8 Nxc8 56.Nc4! and it is time to resign.) 54...Nd6 55.Rxc8 Nxc8 56.Nc3 Kg6 57.Nd5 e6 58.b6 Nd6 59.Nc7 Kf5 60.Kh4 White is completely winning. A very human way to play; the wing pawn is unstoppable. 53.Rb2 Rb8 54.Nc3 Nc4 55.Rb4 Nb6 It was only after the game I started to realize that I should have been more careful about these ...e5 options.

56.Ne4?! 56.Ne2! would have been stronger, as prophylaxis against ...e5. Rather than going to g5, the 204

knight is heading for d4/e5-c6. It is a Karpov/Carlsen move. Carlsen is definitely Karpov 2.0. He feels the harmony of the pieces as well as Karpov ever did and just like Karpov, he has a strong determination to win in any way possible. I am very impressed with how Carlsen is able to avoid chaos on the board, although contrary to Karpov, he is also very good when it does arrive. Another difference is that Karpov wanted to win every tournament, while Carlsen’s idea is to try to win every game. He once said: “I want every game to be an event.” Another thing that is not often paid attention to is how much Carlsen is a leader for the young talents in Norwegian chess. Among others things, he invited me to give a training session with them a few years ago. 56...Nd5?! The last chance to defend was: 56...e5! 57.fxe5 Ke6 58.Ng5† Kxe5 59.Kh4 Nd5

Now 60.Rb2 could be a mistake at the time control. Soon you would have time to realize that Black has counterplay: 60...Kf5! 61.Nf3 Ke4! (61...e5? would lose quickly to 62.Re2 Re8 63.Rxe5†! and the pawns beat the knight.) 62.Rb3 e5! The single source of counterplay. 63.Kxh5 Rh8† 64.Kg6 Rg8† 65.Kf7 Rb8 The win is far from certain. 66.h4? Kf5! would already be a draw. But more natural would be 60.Nf3†! Kf5 (60...Kd6 61.Rb2 Kc5 62.Kxh5 is rather simple to win.) 61.Nd4† Kg6 62.Rb2, when White will win. But it is quite possible to make a mistake on move 60! 57.Ng5† Kf6 58.Re4 White is in control. The rest is just death rattling.

205

58...Nc7 Against 58...Rb6 I had planned 59.Re5!, when both 59...Nc7 60.Ne4† Kf7 61.Nc3 and 59...Nc3 60.Kh4 Nxb5 61.Nxe6 leave White totally in control. 59.Rc4 Nxb5 60.Rb4 e5 A panic move. We should not forget that this is the last time control. 60...Rb7 61.Ne4† Kg6 62.Nc5 Rb6 63.Kh4 also wins. White will take on e6 next. 61.Ne4† Kf5 62.Nc3

206

62...Nxc3 63.Rxb8 exf4 64.Rf8† Kg6 65.Rxf4 Nd5 Black is of course hoping that there would be some sort of fortress around here, but none of them come close to working. 66.Ra4 Ne3 67.Re4 Nf5 68.Kg2 Kf6 69.Kf3 Nd6 70.Rf4† Kg5 71.Ra4 Kf5 72.Rd4 Ne8 73.Rf4† Kg6 74.Ke4 Nf6† 75.Ke5

75...Kf7 75...Ng4† 76.Rxg4† and wins. 76.h3 Kg6 77.Ke6 Kg5 78.h4† Kg6 79.Kxe7 1–0 This was a game with a technical slant. I think I played really well until the wobbly parts of the technical phase. The missed combination at move 17 is not something I regret. The next game almost starts in the conversion phase. Boris Gelfand – Boris Grachev Moscow 2016 Grachev and I have played for the same team in Russia. He is a solid player, who was rated in the Top 100 consistently. He is very stable, and always puts up a strong defence if he gets into a worse position. He also has a reputation as a strong blitz player. At the time this game was played, he was 29 years old.

207

1.d4 d5 2.c4 e6 3.Nf3 Nf6 4.Nc3 Nbd7

This was one of the openings I expected; in 2016 it was very popular. You might remember the game against Magnus Carlsen on page 106 in Positional Decision Making in Chess or against Leinier Dominguez on page 173 in Dynamic Decision Making in Chess. 5.Bf4 I have good results here and Grachev less so. I do not normally pay a lot of attention to such things, but I like this position. 5...dxc4 6.e3 b5 This had become popular at the time of the game. Grachev had yet to play it, but still I expected it. 7.Nxb5 Bb4†

208

8.Nd2 Nd5 9.Bg3 N7b6 10.Qc2 Against Kramnik I got the chance to repeat this line just as the book went to print. 10.a3!? Ba5 11.Rc1 0-0 12.Nc3 Bb7 13.Nxc4 Nxc4 14.Bxc4 Nxc3 15.bxc3 Bxg2 16.Rg1 Be4 17.Qh5 Bf5 18.Ke2 Bb6? (maybe 18...c5) 19.Be5 Bg6 20.Qg4! I was winning and won the game on move 41 in Gelfand – Kramnik, Internet 2020. 10...0-0 11.Be2 Ba6 12.Nc3 c5 13.dxc5

This is the first important moment in the game. I hoped that he would play as he did. 13...Bxc5 209

I thought that 13...Nxc3 14.bxc3 Bxc5 was stronger, and indeed a few days later it was played in Aronian – Anand, Moscow Candidates 2016. This is the starting position of a critical variation, where I am sure more games will follow in the future. 14.Nxd5!N This is a natural move. It is an improvement over 14.a3, as had been played in the Bundesliga 2014, between Lenic and Stern. Technically my move is a novelty, but it cannot be said to be a new idea, as it is the top suggestion of the engine. A definition from the ever-original Alexander Morozevich. Black is faced with an unpleasant choice. 14...Nxd5 I kind of expected this. The computer suggested it and it was happy with it. Black cannot take with the queen because of 15.Bf3, so the alternative is 14...exd5, where I guess I would play 15.0-0.

I think White is simply better already. Usually Black is keeping control of the centre with the pawn on c5 in these positions, but here I have access to the d4-square and Black has lost a lot of the flexibility that hanging pawns are supposed to give him. White has ideas such as Nf3, Be5-d4, a4-a5 and of course Rfd1. 15.Bxc4 Accepting the pawn sacrifice is the only thing that makes sense.

210

15...Bxc4 16.Nxc4

16...Rc8 Black is playing for long-term compensation. His pieces are not yet in the game, so he cannot successfully force things at this point. 16...Nb4 17.Qb3 Nd3† 18.Ke2 makes no sense and only helps White to get his rooks into the game faster. 17.Qb3 Black’s first plan is revealed in the line: 17.0-0 Bxe3! 18.fxe3 Nxe3 19.Nxe3 Rxc2 20.Nxc2 Qb6† 21.Bf2 Qxb2

211

Black is not in any danger. Anyway, it makes no sense for White to allow it. My feeling in this position is that Black should prove that he has enough compensation for the pawn. My opponent seemingly felt the same way, as at this moment he started to think a lot. This made me happy. As said, Grachev is a player with a technical style, who does not usually sacrifice pawns. I do not think it is entirely easy for Black to find compensation and Grachev struggled to do so as well. It is not the topic of this book, so here we shall just vaguely point in the directions where Black could have been looking. 17...Bb4† Also possible was: 17...h5!? The computer’s suggestion, which leaves us with a number of critical lines: 18.h4 This may look more cautious, but things can easily catch fire! 18.0-0 h4 19.Be5 f6 20.Bc3 h3!? looks quite messy, but White is material up and far from getting mated. 18...Be7

212

19.e4!? Very critical (yet not the only reasonable move). Black is forced into wild complications. 19...Nb6 20.Rd1 Rxc4! Maybe not an earned exclam. There is nothing else. 21.Rxd8 Rxe4† 22.Kf1 Rxd8 23.Kg1

White continues to have a slight material advantage, but also struggles to activate his rook. I will allow the reader to form his own opinion on the quality of this queen sacrifice. 18.Ke2 Around here the engine is close to offering Black equality, but it does not mean a lot, as it always 213

changes its mind when you make moves.

18...Qg5?! This is certainly not the way to prove compensation. 18...Rc6 with the idea ...Qc8-a6 was one thing we looked at, but 19.Ne5! disrupts Black’s coordination. For example: 19...Rb6 20.Qc2 and White is definitely somewhat better. 18...h5!? is probably the best try. But I think White is better all the same. If you analyse anything deeply enough, many positions will turn out to be defensible, as we can see a lot of places in this book. But this time around it is too early to analyse everything to the end. In reality our game starts now. With a few accurate moves, White is able to prove an advantage. At this point I thought for 27 minutes. The position is new and quite concrete and I wanted to develop a strategy for the game, just as much as I wanted to work out what to play on the next move. 19.Rhc1! It is important that White moves the right rook, so that ...Qh5† does not annoy. Of course I would prefer to put the rooks on c1 and d1, but the luxury of arranging the pieces exactly as I wanted was not available, so at least I get the rook out and the king to safety, keeping the extra pawn. The rest can be achieved later. 19...h5 I did not see any other ways to disturb the white pieces. 20.Ne5! 214

I spent a good deal of my time making sure this works. 20...Rxc1 A key line was 20...h4 21.Nf3 Qg4 22.h3! and White wins a second pawn. 21.Rxc1 Bd6 This phase of the game has come to its conclusion. It is obvious that Black does not have compensation for the pawn. We can move onto the next phase of the game where White will consolidate his advantage.

22.h4!? 215

Again, I decided to spend a lot of time calculating before playing a sharp move. I saw a few ideas of limited counterplay for Black after 22.Nf3 Qe7 23.Qc4 and wanted to diminish it. Now I am wondering if this was not a poor decision. White is clearly a pawn up here, and going for this position quickly would not have shown any obvious downsides. On the other hand, in the game Black is forced back to the 8th rank. 22...Qd8 If not here, then White will force the exchange of queens. 22...Qf5 23.Qc2! and the queen is “trapped”. And against 22...Qe7 I had planned23.e4 Nb6 24.Nc6 Qc7 25.Bxd6 Qxd6

26.Qg3! and Black would have to either make sad concessions or exchange the queens, which in itself is a rather sad concession. 23.e4!? Of course I saw the option of 23.Nc6! Qd7 24.Bxd6 Qxd6 25.Qb7 during the game.

216

White wins a second pawn, but I was unsure about whether or not I would allow additional counterplay in return for winning a probably superfluous additional pawn. Looking at it with the computer now, it is clear that White is simply winning and Black has no threats, but I did not want to take unnecessary risks. Of course if I had had the time and believed that this was a reasonable way to play, I might have invested the time I needed to play like this. 23...Bxe5 Black has no choice but to make this concessionary exchange, as after 23...Nf6 24.Nc6! White wins a piece. And after 23...Ne7 24.Rd1 Qc7:

217

25.Ng6 Bxg3 26.Nxf8 White wins the exchange, as 26...Bd6 does not trap the knight, but loses even faster after 27.Nxe6!. So 23.e4 was not without its attractive attributes. 24.Bxe5 Ne7 I might have overestimated how easy it would be to win here.

25.Rc5?! Again I spent 12 minutes, but this time I did not come up with the best move. 25.g3!, securing the h4-pawn, was strong. After 25...Ng6 26.Bc7! Qd7 27.Qd3! White is in control 218

and ready to advance the queenside pawns in the near future. 27...Qa4 28.Qc4 Qd7 29.Rc3 and White will consolidate. I would argue that if I had played this, there would have been little reason to criticize any of my previous decisions. 25...Ng6 26.Bg3 26.g3 was still decent, but I did not want to exchange the good bishop, so I did not consider this for a moment. 26...Qd4 27.Rc4 Somehow I underestimated Black’s counterplay in this position. 27...Qd7 Black is now ready to play ...Rd8. 28.Rc7 Qd4 29.Qc2 Rd8

30.Rc8!? This is an entirely practical decision. I did not enjoy the counterplay and although I would have preferred to exchange queens, I felt that one exchange would at least remove some of his counterplay. 30.Rc4 Qd7 31.Kf1!? was the alternative. The king goes into safety and White wins in the long run. 30...Rxc8 31.Qxc8† Kh7 32.Qc2 e5 Let’s assess this position. White is a pawn up, but his bishop is cut off and a lot of the possible endgames, such as the queen ending, are not that easy to win.

219

After a few imprecise moves, it is now time for White to show some accuracy in order to win. Probably it was quite a poor thing that I allowed the bishop to get buried on g3. 33.Kf1 Qd8! This could have been a great trick, if it was intended as such. 33...Nf4 would lead to a queen ending, if White wants it. 34.Qe2? This was simply a blunder. 34.Kg1 was the right move. 34...Nxh4 is met with 35.Bxe5 and White keeps his advantage, as in the game.

34...Kg8? Grachev missed the chance to play 34...Nf4!, when I cannot take the knight on account of 35...Qxh4!. And after 35.Qf3 Kg8 36.Kg1 Qd2 Black should have enough counterplay to make a draw. 35.Kg1 Nxh4! In general it is a good thing for the defender in an endgame to exchange pawns, and not surprisingly Grachev decided that he should do so while he has the chance. After slower play like 35...Qd4 36.Kh2 Nf4 37.Qc2 Ne6 38.f3 followed by Bf2, White seems to stay in control.

220

Here I had a difficult choice. 36.Bxe5 I decided to take on e5, as it made my plan very simple. 36.Qxh5 Ng6 37.Qe2 was of course also possible, but the exchange of pawns has clearly helped Black. (Note that 37.Bxe5? is a blunder. After 37...Qa5 38.f4 Nxf4 39.Qf5 Qe1† Black has a perpetual.) 36...Qa5 I was hoping for 36...Nxg2

37.Qxh5! Nh4 38.Bg3 Ng6 39.Qd5, where the bishop on g3 controls everything and the queen on d5 the rest. Of course some technique would be required, but things are moving in the right direction. 36...Qg5 37.f4 Qg3 38.Qf2 followed by Bb8 and Kh2 also allows White to consolidate. 37.Bg3 Ng6

221

38.a3 I felt this was more ambitious than putting the pawn on b3. White wants to play b4 and put the bishop on c5, and I cannot see a way for Black to exploit the weakness of the b3-square. 38...h4 39.Bd6 Qb6 40.Qd2 Qb3 41.Kh2 h3! I believe this is a must. Black has to weaken the white king and try to create some sort of counterplay. If Black was waiting, White would get more time to consolidate, starting with 42.f3. I would play Bb4-c3 followed by Qd5 and then White is ready to advance on the queenside, in a scenario where the bishop dominates the knight and Black has to run with the queen every time its exchange is suggested. In short, it is much easier for White to improve his position than it is for Black. So Black has to change the character of the game now, or he will have to do it later under worse circumstances.

222

If possible, I should take this pawn. The weakening of the king’s position cannot be avoided. 42.gxh3 Nh4 43.Qe3 This felt very natural. 43.Qc3! turns out to be strong, but only due to some concrete details which are not easy to detect. 43...Qd1

44.Be7! Nf3† 45.Kg3 Nd2 46.Qc8† Kh7 47.Qf5† Kg8 48.Bb4 Qg1† 49.Kh4 Followed by 50.Bc3 with total control. If you think this looks rather hair-raising, I can assure you that it is not only you. When you look with the engine, everything is easy. But during the game where you do not have such luxuries, you will look at a variation like this and wonder what you have missed and where the 223

unexpected knight check will come from, dragging you straight to hell... 43...Qd1!? A good idea, looking for counterplay. My key idea was: 43...Qxb2 44.Bg3 Qf6 45.Qxa7! Nf3† 46.Kg2 Ne1† 47.Kf1 Qa1 (47...Nc2!? offers more resistance, but White remains two pawns up.) 48.Qb8† Kh7

49.Qb3!! White is in total control. For example: 49...Nf3† 50.Kg2 Ne1† 51.Kh2 and the checks have finished. White will win. 44.Bb4! I was very happy about finding this move, removing the check on e1. 44...f6 45.Qg3?! I rushed things here. 45.Bc3! was much stronger.

224

White consolidates and is on the way to winning the game. In these long games where it is very hard to spot the critical moments, because every moment is a mini-version of it, inaccuracies are bound to happen. This is why it is important to analyse the games and improve our feeling for how to spend our time, how to organize our pieces, how to organize our thinking and how to control the opponent’s counterplay. A lot of it happens subconsciously. We analyse the games, find out what actually happened, compared to our experience during the game, and our feeling for the details will be slightly better the next time around. 45...Nf3† 46.Kg2 Ng5 47.Qe3 Here I had realized my mistake, but there was no longer any way to prevent him from getting a pawn back. These are the sort of random chances Black was hoping for with 41...h3!, and which he got. Not inevitably so, but in practice he did this time around. 47...Qc2

225

48.h4! My thinking here is that I do not want to risk ending up in an endgame with two pawns against one on the queenside. I much rather want two passed pawns and a lot of randomness. Therefore I provoke the knight to take on e4. Something like 48.e5 Qxb2 49.exf6 Qxf6 is far less convincing. 48...Nxe4 49.Qxa7 Qc8! Unfortunately I had missed this move. I had seen 49...Qc6 50.Qe3! and White is just winning.

226

I think I found a reasonable plan, to give up the h-pawn in order to get my king to the queenside. 50.Qe3!? From a practical viewpoint there is no reason to criticize this move. It is a good steppingstone in my dream to come to a2 with the king. The computer suggests a surprising idea: 50.Kf3! Qf5† 51.Ke2 Qe5 The human move here is 52.Kd3 and White has a big advantage. But the machine finds something even more profound and, I may add, not entirely human. 52.Qe7 Qxb2† Normal moves like 52...Qf5 allow White to consolidate with 53.Qd8† Kh7 54.Qd4. 53.Ke3!

The knight is trapped in the middle of the board! 53...Nxf2 54.Qe6†! And after a check on f5, the knight is won. Obviously this is the type of thing the engine does much better than a human. Finding a tactic in a position where there are a large number of possibilities. No human can go 2-3 moves deep in all lines and see these types of options. So, all in all, I pay attention to this kind of information, but I do not regret not seeing it. I should mention that the computer will find small improvement everywhere in this type of position. If this has your interest, please look at it. Here, I will describe the way I approached it during the game and what I was trying to achieve.

227

50...Qg4† 51.Kf1 Qxh4 52.Be1 Ng5 In practical terms I believe that Black had very limited chances of holding the queen endgame arising in the long forcing line: 52...Qh1† 53.Ke2 Qh5† 54.Kd3 Qd5† 55.Kc2 Qc4†

56.Bc3!? (56.Kb1 was my idea and also very strong, but allowing the exchange of the minor pieces is interesting.) 56...Nxc3 57.Qxc3 Qe2† 58.Qd2 Qc4† 59.Kb1 g5 60.Qc2 Qe6 61.b3 The two passed pawns are very powerful. 53.Ke2 Qg4† 54.Kd3 Qf5† 55.Kc4

I have achieved my plan. The king is partly ready to help the passed pawns and partly hoping to shield itself from checks with their assistance. 228

55...Qc2† 55...Ne4 56.Qd4! is a nice piece of micro-domination. Also after 55...Ne6 56.Bc3 Nf4 57.Qd4 Qe6† 58.Kb5 White is making progress, although many more moves will be played. 56.Bc3 Qa4† 57.Bb4 Qa6† 57...Qc2† 58.Kb5 is the kind of thing I was hoping for. Now 58...Qxb2? loses on the spot to 59.f4! and the knight is trapped. And after 58...Qf5† 59.Bc5 the checks are about to come to an end. 58.Kb3 Ne6

59.Qe4! A fine centralizing move, taking control of the light squares. My plan has been a complete success. The king has managed to escape the knight’s reach. 59...Qb6 60.Qd5 A further improvement. 60...Qa6

229

At this point we received a final 15 minutes on top of the 30-second increment. But at this moment I am very angry with my play. I played too quickly. I was thinking: “This move is winning, that move is winning, why should I spend time here?” 61.a4 There is nothing objectively wrong with this move, but it misses the chance to make things much easier. After 61.Qc4! Qb6 62.a4 Kf7 63.Qd5! White is in total control and the advance of the pawns can commence. This would have saved a lot of trouble. The a-pawn will quickly advance. 61...Qe2 This is the only move to keep the game going. 62.a5 Kh8

230

63.Bc3 At this point I missed a really beautiful move: 63.Bd2!! The idea is to play Be3, when the bishop will dominate the knight. Maybe the knight is even lost (or forced to go to g5, when it might as well be lost). The move is immediately winning, on account of 63...Qxf2 64.Qxe6 Qxd2 65.Qc8† Kh7 66.Qc2† and White wins. 63...Nf4 Black manages to create some counterplay; White is still winning. 64.Qc4 Qd1† 65.Kb4 Despite missing the immediate win, I was able to continue in a decent way. My plan was 65.Ka2, but I was unsure about 65...Ne2 66.a6, where White is winning, but I was reluctant to go for it, as it seemed out of control. But looking at it now, I can see it is totally in control. 65...Qd6† 66.Ka4 Nd5 67.Bd4 Nc7 68.Bb6 Qd1† 69.Ka3 Qa1† 70.Kb3 Qd1† 71.Ka2 The king finally arrives on a2. 71...Na6! The only chance.

231

72.b3! Sort of trapping the knight. 72...Qd2† 73.Ka3 Qd6† 74.b4 Qd2! A good practical try. I had missed this.

75.b5 I wanted to play: 75.Be3 Qd6 76.Bf4 But suddenly I became scared that I would be trapped in a perpetual check. With only 30 seconds on the clock, it is not so easy to go for something like this. It could just win, but also, it could 232

allow a draw in a totally winning position. To show caution was not wrong, but my fears of a perpetual were exaggerated. 76...Qd1

77.Qxa6 We should also not forget that although this is the most obvious idea, it is not obligatory. 77.Be3! would have been a very practical way to deal with these fears. This is the set-up I missed 15 moves earlier. 77...Qa1† 78.Kb3 Qd1† 79.Kc3 Qa1†

I only had time enough to realize that I should not drop my queen to ...Qf1†, which is the reason I decided not to try to win the piece. 233

80.Kc2! Qa2† 81.Kd3 Qb1† 82.Ke2 Qe4† 83.Be3 Qc2† 84.Ke1 Qb1† 85.Kd2 Qb2†

86.Kd3 Qb1† 87.Kc3 Qe1† 88.Kb3 Qd1† 89.Kb2 And White wins. For the few good dancers among the readers, this win will appear even more demanding than to those who are just confused by the many options White had on every move. 75...Qd6† Once again White is required to show some technique to win the game. 76.Kb3 Qd1† 77.Qc2 Qd5† 78.Ka4 Qd7

Black is managing to keep the game going. But his set-up is more fragile than White’s. 234

79.f4 This felt very natural during the game. White is of course winning, but while being pinned it is not so easy. I anticipated the queen ending that occurs in the game. In the game I made a few repetitions to gain time on the clock. Actually, I was seriously concerned that I might overdo it and allow a three-fold repetition. White has many winning tries, but has to settle on one. A nice one would have been: 79.Qc1!? Kh7

80.Qh1†! Kg8 81.Qe4 Kh8 82.Qh4† Kg8 83.Qd4 Qe8 84.Qc4† Kh8

235

85.Kb3 and the king is safe from the harassment. 79...f5 80.Qh2† Kg8 81.Qe2 81.Qb2! Kh8 82.Qd4 was very strong. 81...Kh8 82.Kb3 Qd5† 83.Ka4 Qd7

84.Qh2† Kg8 85.Qa2† Kh7 86.Qb3 Kh8 87.Be3 Nc7 88.Bc5 Na6 89.Bb6 Kh7 90.Qc2 Kh8 91.Qe2 Kh7 92.Kb3 Qd5† 93.Kc3

This is where I was always aiming. I had faith in White’s chances in this queen ending.

236

93...Nc5 94.Bxc5 Qxc5† 95.Kb3 Qd5† 96.Kb4 Qd6† At this point I was afraid of: 96...Qd4† 97.Qc4 Qd6† 98.Kb3 Qd1† 99.Kc3

237

99...Qc1† After 99...Qe1† 100.Kd4 Qxa5 101.Qc6 White wins with accurate play. Black does not really have anything better than what happened in the game. 100.Kb4 Qd2† 101.Kc5! Qxa5 102.Kc6 Qa8† 103.Kd7 White is winning; the passed pawn is very strong. Still, a lot of moves would have to be played. 103...Qe4 104.Qc6 Qxf4 105.b6 Qd4†

106.Kc8 Qb4 107.b7 Qf8† 108.Kd7 Qf7† 109.Kd6 Qf6† 110.Kd5 Qf7† 111.Kd4 Qb3 112.Qh1† Kg6 113.Qg2† Kf6 114.Qc6† Kg5 115.Ke5 Qb2† 116.Kd5 Qb3†

238

117.Kd6 White will soon win. 97.Kc4 Qc7† 97...Qxf4† 98.Kd5 would make things easy. 98.Kb3 Qxa5 99.Qh5† Kg8 100.Qxf5 Qe1 101.Qe5

At this point I was confident that I would win. I knew what I should do. The queen controls everything and the white king can go forward. White is also winning in this position if there were no f- or g-pawns on the board. With the exception of ...Qxe5 that is! 239

From this point on, we are just making moves. The death rattle of the black player takes a bit of time to end, but there is no way for him to escape my grip. White can always play f5-f6, creating a situation where the black king is exposed, limiting the number of checks Black is able to give. The general idea is basic: White will aim to put the queen inbetween on one of these checks, spiking the black king, forcing the exchange of queens and Black’s resignation. 101...Qd1† 102.Kb4 Qd2† 103.Kc5 Qc1† 104.Kb6 Kh8 105.Kb7 Qh1† 106.Kb8 Qb1 107.Qc5 Qe4 108.Qh5† Kg8 109.Qe5 Qb4 110.Kb7 Kh7 111.b6 g6 112.Kc7 Qc4† 113.Kd8 Qd3† 114.Kc8 Qh3† 115.Kb8 Qd3 116.b7 g5 117.Kc7 Qc4† 118.Kd8 Qd3†

119.Ke8 1–0

240

Chapter 6 When is the Right Time to Run?

Never run away from a challenge...

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

241

White is in trouble, but not lost (see page 137)

White is under pressure. How should he react? (see page 138)

242

The white queen is off-track. What next? (see page 140)

How can Black break through? (see page 142)

243

How should Black convert his advantage? (see page 146)

Black needs to show accuracy (see page 150)

244

How can White defend? (see page 151)

Black should look for a way to win the game... (see page 156) One of the most difficult things in chess is, as in everything else, timing. In this chapter we shall look at situations in both the middlegame and the endgame, where both players had to make decisions about when to improve their position and target the opponent’s resources and when to roll the dice and attack the king or let the passed pawns roll. Although there are some general observations we can make about these concepts, this is perhaps the most essential theme in top-level chess, as a good 245

feeling for what kind of action is needed in various positions, when to calculate and when to simply improve the position, is worth many points. The game in this chapter was the first in my match against David Navara in Prague. It was a wonderful match in a lovely city, organized by Pavel Matocha. It was played in a beautiful castle. I am sure that all the players who have played there – and by now we are many – enjoyed it a lot and are very grateful that we had the chance to play against David under such splendid conditions. I should say that the format of this yearly match has changed over time. Sometimes it is a rapid match, sometimes with classical time control and yet again, sometimes a mixture of the two. In my case it was played with a classical time control after four blitz games at the opening ceremony. My opponent was David Navara, a very talented Czech grandmaster who is maybe a bit unstable in his performance, but on a good day, he can play truly exceptionally, which many top grandmasters have acutely felt! On top of this he has a wonderful friendly personality. That my opponent found this game to be a creative experience as well is shown by him including it in his book in Czech on his own games. David Navara – Boris Gelfand Prague (1) 2006 1.d4 Nf6 2.c4 e6 3.Nf3 b6 4.g3 The classical Queen’s Indian set-up, which I love to play as White as well. 4...Bb7 5.Bg2 Be7 6.Nc3 In the third game of the match David tried out the Polugaevsky-system: 6.0-0 0-0 7.d5!? exd5 8.Nh4 c6 9.cxd5 Nxd5 10.Nf5 Nc7 11.e4 d5 12.Nc3 Bf6 13.Bf4 Bc8 14.g4 Nba6 Theoretically this is fine for Black and I got a very promising position after the opening. But things went wrong after I missed a mate on move 41, after which I had to defend for a long time, before making a draw on move 104. Later, this 7.d5 sacrifice became fashionable again. 6...0-0

246

7.Qc2!? A small sideline. White wants to avoid the main line 7.0-0 Ne4. At the time of the game it was a somewhat forgotten line, but as we shall see, it was played in the years following this game. 7...c5 7...d5 8.cxd5 exd5 is quite a good version for White of this typical hanging pawn structure. 8.d5 Nothing else makes much sense. 8...exd5

247

9.Ng5 White is also able to make this a real gambit with: 9.cxd5 The critical line goes: 9...Bxd5 9...Nxd5 10.Qe4 f5 11.Qc4 b5 12.Qb3 c4 13.Qxb5 is favourable for White it seems. Topalov played 9...Na6 against me in a rapid game. After 10.Nh4 d6 11.Nf5 I was slightly better in Gelfand – Topalov, Nice (rapid) 2008. 10.Nxd5 Nxd5 11.Nh4 Nb4 12.Qd2 N8c6 13.Nf5

13...Bf6 248

This was a not very difficult improvement at the time, since then played in a few games. In the stem game Black played 13...Nd4!? and after 14.Nxd4 cxd4 15.Bxa8 Qxa8 16.0-0 Qe4

17.a3! Nc2 18.Ra2 Rc8 19.b3² Black did not have full compensation for the exchange, even though the game ended in a draw not long after in Aronian – Carlsen, Morelia/Linares 2008. 14.0-0 Na6 15.Rd1 Nc7 Black has more or less equalized, Sargissian – Adams, Merida 2008.

9...h6 One of the most famous miniatures in history was played in this line. At a time when Korchnoi was trailing 3–1 in decisive games, with four games remaining of the match, he managed to deliver an early knock-out blow to his opponent. 249

9...Nc6 10.Nxd5 g6 If you look at this 7.Qc2-variation with a computer you will quickly find that everything is equal; but in reality, some things are more equal than others. 11.Qd2

Here Karpov made a big blunder. 11...Nxd5?! 11...Rb8 was one of a few better options here. 12.Bxd5 Rb8?

13.Nxh7! Re8 After this it is already over, but Karpov made a few more moves to avoid the embarrassment of 250

losing in 13 moves, causing only more embarrassment... 14.Qh6 Ne5 15.Ng5 Bxg5 16.Bxg5 Qxg5 17.Qxg5 Bxd5 18.0-0 Bxc4

19.f4 1–0 Korchnoi – Karpov, Moscow (21) 1974. Korchnoi failed to win any of the three final games, losing the match 12½–11½. The next year Karpov was anointed World Champion when Fischer refused to play. 10.Nxd5 A year later Grischuk came up with a nice idea. After 10.h4! Na6

251

11.Nxd5 Bxd5 12.cxd5 he won a fine game in Grischuk – Adams, Calvia 2007. 10...Bxd5

11.Bxd5?! Better is: 11.cxd5 Still, it is also not really challenging, for concrete reasons. 11...hxg5 Black should take, as after 11...d6 White will return with the knight, when he clearly has achieved something. 12.d6 Nc6 13.dxe7 Qxe7 14.Bxg5

252

If White had completed his development and was to move, he might have a chance to fight for an advantage. This is chess: Black can use the moment to solve his problems. 14...Qe5! I like this move the best, although 14...d5 is also OK. But exiting the pin and clarifying the situation makes perfect sense. Obviously, the engine does not care about such issues, but then the computer does not feel under pressure and has never experienced how it feels to blunder while under pressure... The key point is that after: 15.Bxf6 15.Bf4 Qe6 is OK for Black too. 15...Qxf6 16.0-0 d5! Black has equalized. 11...Nc6 12.Bxc6 12.Nf3 Nb4 is even worse for White. The d-pawn can easily become a target. 12...dxc6 13.Nf3

253

The opening has sort of almost ended. White will castle on the next move and the manoeuvring of the early middlegame can begin. This makes it a reasonable moment to stop and think about this slightly unusual pawn structure. When I was growing up in Minsk, I was taught that a doubled pawn was almost the same as a pawn deficit. But we see now, for example with the Berlin Wall in the Ruy Lopez, that there are many things that affect the evaluation of the position, with the pawn structure being only one of these. White has an extra pawn on the kingside and Black has doubled pawns on the queenside. This would certainly be relevant in the endgame, but for the moment this is not so important. Black can put pressure down the e-line and exploit the weakening of the light squares on the kingside, where the white king will (and should) most likely go. Another important point is that Black will be able to exploit that he has an extra open file in the centre by placing his knight on e4 quickly. If White was able to advance his pawns to e5 and f4, for example, he would be in control. But it is not going to happen. 13...Qc8 14.0-0 White must complete his development. After 14.Nh4 the natural idea is 14...Re8 15.Nf5 Bf8 with a nice position. But even stronger is: 14...Qh3!

254

This keeps the white king trapped in the centre. The key tactical justification for this is 15.Qf5 Ng4 16.f3 (16.Nf3 would not improve things for White) 16...Bxh4! and White does not have a move. This is of course not a flashy line, but it is the sort of line you have to see in order to choose to go down this path, as you cannot rely on hope that “something will turn up” and consistently make good moves. 14...Qe6 This is the natural move. Mate is unlikely, so 14...Qh3 makes less sense now. Taking control of the central squares appears more prescient.

I do not claim I am better at this point, but I was happy with the result of the opening. What is 255

important to me is that Black can play natural chess from here on. Instead, sometimes you get a good position where everything is hanging by one move. One inaccurate move would be able to change the evaluation of the position entirely. 15.b3 This makes sense. It does weaken the dark squares a little bit, as we shall see later. But on the other hand, the bishop would be useless on f4. It is worth comparing this game with the game against Caruana from Amsterdam (see page 161), where I also managed to get a very comfortable position after the opening, laying the responsibility of maintaining the balance on White’s shoulders. It is very rare you get an advantage in the opening with Black, but just putting the opponent under some pressure is already a big achievement. What is important here is that it can be difficult to be psychologically ready to play accurately to hold the balance this early in the game when you are White. Even if a player realizes that things are going the wrong way, the adjustment is rarely effortless and takes some time. Both Navara and Caruana have a good sense of danger, but it was not enough. Even though we cannot offer a draw early on anymore, we can still try to equalize. But to do this with White early on can be difficult, psychologically. 15...Ne4 16.Bb2 Bf6 Of course, the luxury option would be to play 16...Rfe8 and only then ...Bf6 and then take back with the queen. But the problem was 17.Ne5 when Black does not have any problems, but it does not look like an improvement on the game. 17.Bxf6 Nxf6 18.Rad1 Rfe8 18...Rae8!? was suggested by Navara as another reasonable option.

256

19.Rfe1 During the game I was certain that 19.e3! was better. The rook is too passive on e1; instead White should double the rooks on the d-file. The knight does become a bit loose on f3, but I cannot see how Black can exploit it. Actually, it would be rather easy to get into big problems with Black: 19...Ne4 20.Rd3 Qf5?! 21.Nh4 and 22.f3 and White even has an advantage. After 19.e3 I think White is entirely out of trouble. 19...Ne4 20.Rd3 This does not harmonize very well with the last move. Doubling on the d-file makes a lot of sense, but then why is the rook on e1? White could try 20.Nd2. The exchange of the last minor piece would make a draw very likely. And after 20...Ng5 21.h4! Black can retreat with the knight, but then White really has no problems, or he can play 21...Nh3† and ...Nf4†, which leads to an immediate draw. When you play according to the position, you will have to accept that sometimes a draw is the most natural result. 20...Qf6 A natural move. I want to put pressure on f2 and advance the g- and h-pawns. It was also possible to put the queen on f5, with the idea of putting pressure on the diagonal from f5-c2 as well. But the downsides seem quite important as well. White can gain a tempo on the queen with both Nh4 and f3&e4 in various lines. We analysed it for a while, and it does not seem that this is radically different from the game.

257

21.Kg2 21.Rf1!? was also possible and White is still fine. But as in the game, he will have to find a few accurate moves in the near future. As you will know from Positional Decision Making in Chess, I was very impressed by Glauser – Polugaevsky, Havana (ol) 1966, where Polugaevsky admitted his mistake and returned the rook to f8 the move after moving it to e8. Not a proud moment, but sometimes you have to do whatever you can to survive. So, to conclude, I think that 21.Rf1 is an improvement of the rook’s placement. I like it very much, although there is nothing really wrong with what David played in the game. 21...g5 Black has an extra pawn on the queenside, but it is obvious that a real push for an advantage will have to come on the kingside, where White’s king position is ever so slightly weakened and where the knight on e4 is within striking distance. 22.Rf1 Finally, White is forced to put the rook where it belonged all along. 22...Rad8

258

23.Rxd8 23.Nd2 is no longer an obvious move. The following line is sort of forced: 23...Rxd3 24.Nxe4 Rxe4 25.Qxd3 Rd4 26.Qc3 (26.Qe3 is met with 26...Qe6! eliminating White’s best piece. 27.Qxe6 fxe6 28.Kf3 Rd2 and the endgame is unpleasant.) 26...Qd6

Stockfish helpfully tells us that the position is still equal. But is it really? Black is in full control of the d-file and can penetrate to the second rank. Maybe White can keep the balance with a4-a5 or something like this, but he is not equal. Black is much more comfortable, even if it can be hard to make progress. But it is obvious for those looking at the board rather than elsewhere, that Black is the stronger side.

259

23...Qxd8 24.Rd1 I think this is a bit superficial again. There is not much happening on the d-file and the rook seems to belong on f1. But having said that, I should volunteer that there is no obvious path to full equality as far as I can see. White will feel some pressure in all lines. And the moment you must choose between two unpleasant lines, things have suddenly become much more perilous. 24...Qf6 25.h3 I spent a good deal of time here as there are several interesting possibilities, but none of them are obviously preferable.

25...h5 This is the most natural move. The following line makes no sense at all: 25...Nc3?! 26.Rd3! Nxe2? 27.Re3 Rxe3 28.fxe3 Nc3

260

29.e4! and the knight is trapped. White is much better in this line but this is not something I would concern myself with during the game. It is obvious that Black is not playing for a win, and as I start out by calculating in a position where I feel very comfortable, there is no reason to use energy on something that feels uncomfortable. 25...b5 But this was an interesting move. What I want to play is ...b4 and ...Nc3. 26.a4 b4 27.Rd7 h5

28.e3 The following dynamic approach is also relevant: 28.Qd3!? g4 29.hxg4 hxg4 30.Qe3! gxf3† 31.exf3 Qf5 32.Rxa7 Nd6 33.Re7 and White does not look worse, despite being a piece down. 261

The a-pawn is strong, the black king is all open and the pawns on the queenside in need of a bit of protection. 28...g4 29.hxg4 hxg4

30.Nh2 The natural place to put the knight, to create counterplay against g4. Although the position is very dangerous for White, accurate defence will keep him going. 30.Nh4! is a better way to keep the balance, keeping the h-file closed. 30...Qf5 31.Rxa7 Kg7! The most dangerous. 32.a5 Qh5 White is in big danger.

262

33.Ra8!! The only defensive resource. White survives barely. 33.Qe2? Nf6 and White cannot defend himself on the h-file anymore. 33...Rxa8 34.Qxe4 Rxa5 The endgame after 34...Rh8 35.Qxg4† Qxg4 36.Nxg4 Ra8 37.Ne5 Rxa5 38.Nd3! is surprisingly a direct draw. The counterplay against the c5-pawn keeps everything going. 35.Nxg4 Ra1 36.Nh2

This looks scary for White, but the moment we are looking at actual progress plans for Black, they are elusive. We are in a phase in the game where things are difficult for White, but with accurate play, he can still 263

neutralize the pressure effectively, meaning that he will have to do something accurate, not to have problems lingering down the line. This is one of the most difficult things in chess, as what type of defence you choose is so reliant on style. For example, Karpov would try to anticipate the opponent’s ideas and make the wins elusive. While his most frequent opponent, Kasparov, would rather give up material than be doomed to passive defence. In the age that followed the hegemony of these two World Champions, the computer has been teaching us a lot about defence and the champions of the 21st century try to use active defence and passive defence interchangeably. To those who want to look deeply into defence, a good starting point could be to look at Karpov and Kasparov’s games with a focus on that angle, learning the techniques these players applied. Another path would be to get all your friends to buy this book and have them write letters about the need for us to do a book on defence as well... But you will soon exhaust the study material from these two players, as they mainly had to defend against each other. But many players continue in the same style: Shirov defends in the style of Kasparov, while Karjakin defends in the style of Karpov, while Anand likes to go for positional sacrifices in the style of Petrosian.

26.Qd3?! One of the ways White could have defended at this point would have been 26.e3!, where the core plan is Ne1-d3-f4 under the right circumstances (if ...g4 can be followed by h4), but also to defend f2 with the queen. Direct play with 26...g4 27.hxg4 hxg4 28.Nh4 does not give Black any ways to play for an advantage either. White can target the g4-pawn and activate the rook down the d-line. With 26.Qd3, Navara is looking for a counterattack. 26...Kg7 Taking the king off the 8th rank is a good move. Black is in no hurry. 264

There was also a cheapo to look out for. 26...g4?! 27.hxg4 hxg4

28.Qd7! and Black must find 28...Qe6 not to be worse, which was certainly not the idea.

27.Qd7?! This tempo-winning move is the first move that leads White into real trouble. It was quite late in the process of working on this book before I realized that this is not a good move and White should have defended f2 right away. I sort of saw it like “this is what he intended, so of course he played it” and in practice, this is how it works when you play. Having said that, this is objectively the first real mistake.

265

27.Rf1! would have kept White at the balancing point of not having long-term problems, with accuracy now. One of the key points is that 27...Nc3? would not work due to 28.Nxg5!. 27...Re6 28.Rf1 The only move. 28.Qc8? This would be a kind of trap-based move. The key idea is that 28...g4? 29.hxg4 hxg4? would lose to 30.Rh1!. But there is a big downside.

28...Qb2! 29.Re1 g4 30.hxg4 hxg4 31.Nh4 31.Nh2 is best refuted with 31...Nxf2! 32.Kxf2 Qd4† 33.Kg2 Qd2, but there are many other dangerous options for Black. 31...Qd2 32.Kf1

266

32...Kh7! Black is winning, as given by Navara.

28...Nc3! My idea during the game was to set up a mating attack. I should say that at the time I only saw this as one of several possibilities, but my intuition was good to me. It is important for Black to exploit the weakening of the light squares that happens after e2-e3, in order to increase the power of the queen check on e4. This is the type of thing you rarely find out by calculation and comes from a feeling you cannot explain the origin of as anything but decades of experience distilled into intuition.

267

28...Qb2?! 29.Qd8 gives White unnecessary counterplay. Clearly the queen is lost in the wrong spot on the board, rather than creating threats. 29...Rf6 30.Qd3! and Black has nothing better than to repeat the position with 30...Re6. 28...g4 This is premature. 29.hxg4 hxg4 30.Nh4! 30.Nh2? Qf5 with lots of ideas like ...Qh5, ...Ng5 and ...Rh6 leaves White’s position untenable. 30.Nd2? Qf5 is also dead lost for White. 30...Qe5

31.Qd3! White is holding on, not falling for: 31...Rf6 With the threat of ...Rxf2†. 32.f4 Although it may look scary, White is fine. 29.e3 Navara gave the following variation in his book: 29.h4 gxh4! 30.Nxh4 Nxe2 31.Qxa7 Qe5! And Black is much better:

268

After 32.Qd7 Nd4 Black is in control, and this is what the engine suggests as best, as after: 32.Qxb6 White will be met with brute force: 32...Rf6! 33.Qd8 Qe4† 34.Kh2 Nxg3 And in order to avoid immediate resignation, White would have to try building a fortress with: 35.Qxf6†! Kxf6 36.fxg3† Ke7 37.Rf2

But if David had looked at the line all the way to here, he would have realized that this fortress is strong, and the endgame should be very easy to hold for White. Zugzwang is impossible, as the knight can go between h4 and f5 without consequences. The pawn endings are all better for White and the only breakthrough would be with ...f5-f4, which I cannot see how to engineer.

269

29...Rd6 30.Qxa7? A strange mistake, which is based on a miscalculation. Somehow, I must have believed David to some extent, allowing him to escape what should have been an immediate end. 30.Qc8 with some advantage to Black was better.

30...Ne4? This is a very careless mistake, played “by hand” rather than with the head. I lost the sense of timing. The knight is not poorly placed on c3, as the check on e2 is also important in some lines. And on top of this the queen check on e4 is very effective. Black had an immediately winning line. 30...g4! 31.hxg4 hxg4 32.Nh2 32.Nh4 does not work here. Navara gives: 32...Rd2 33.Qc7

270

33...Nd1!, as winning, with the idea 34.Qf4 Qxf4 35.exf4 Ne3†. 32...Qe6! The check on e4 is going to be very unpleasant. At the same time the queen controls the e7square, so the white queen cannot return to the game easily. 32...Qf5 33.Qe7 Rh6 is also very strong, but has more the air of being a computer line. As in, yes it looks strong, but no one would choose this if he was trying to find a win in the position.

33.Qb8 There is no such thing as the toughest defence at this stage, but we can eliminate a few of the alternatives all the same. 33.f3 is most easily refuted by 33...Rd2† 34.Rf2 Rxf2† 35.Kxf2 Nd1† and Black is entirely 271

winning. 33.Re1 Qe4† 34.Kg1 is not something I would analyse during a game and it does not surprise me that everything wins.

Still 34...Ne2† 35.Kf1 Nxg3†! and 35.Rxe2 Rd1† 36.Nf1 Rxf1† are both immediately mate, so why not include them? 33...Qe4† 34.f3 Rd2† 35.Rf2

35...gxf3†! Black wins a piece. 36.Nxf3 Rxf2† 37.Kxf2 Qc2† 38.Kg1 Qd1† 39.Kg2 Qe2† 40.Kh3 Qxf3 41.Qe5†

272

41...Kg6! Not strictly necessary, as 41...Qf6 keeps the extra piece, but we will get a study-like mate after: 42.Qxc3 Qh1† 43.Kg4 Qh5† 44.Kf4 Qf5#

As said, I did not pay enough attention to this being the moment to calculate and find a direct win. Partly because I did not see anything better for White than what happened in the game, where I have to add, to avoid letting myself off the hook too easily, the win also was by no means trivial.

273

31.Qc7?! David understood that the queen had to come back, but this is not the way. I did not look at 31.h4? during the game and I doubt Navara did either. But the refutation is a nice little tactic: 31...gxh4 32.Nxh4 Rd1!!

The white position collapses. Navara gave the following variation: 31.Qxb6 g4 32.hxg4 hxg4 33.Nh2

274

33...Rd2! 34.Nxg4 Qg5 and White’s position collapses. But the thing we both missed, and which required the assistance of a computer to see, was that White could have equalized with: 31.Qb8!! The optimal square for the queen, preparing things like Qe5 as well as hitting the rook on d6 and, importantly, the h8-square. To see the possibility of counterplay down the h-file here requires a special kind of imagination. 31...g4 Black has no meaningful way to improve the attack any further. As we already know, he is a move too late with executing the attack. 32.hxg4 hxg4 33.Nh2

275

33...Qg6 There are many good moves, but there are also some innocent-looking moves that do not work. 33...Qe6? would be a blunder. Suddenly things are difficult for Black. 34.Nxg4! Qxg4 transposes, and 34...Nd2 loses spectacularly to:

35.Nh6!! 34.Nxg4! Nd2! Not only a nice intermediate move, but also the only move! 34...Qxg4? 35.Rh1 Rh6 36.Rxh6 Kxh6 37.Qh8† Kg6 38.Qg8† Kh5 39.Qxg4† Kxg4 40.f3† and the pawn ending wins for White. 35.Nh6!! 276

White has a plethora of options, but this is the move that causes by far the most problems to Black. 35...Rf6!! The only move. White was threatening Qxd6 and 35...Kxh6 loses to 36.Qh8† Kg5 37.f4† Kf5 38.Qe5† Kg4 39.f5 Qg5 40.Rf4† Kh5 41.Qxd6. 36.Qg8† Kxh6 37.Qh8† Kg5 White has nothing better than a perpetual here. 38.f4† Kf5

39.Rf2! Ne4 This sets up a crazy perpetual. 277

40.Qh3† Qg4 41.Qh7† Rg6 42.Qxf7† Rf6 43.Qh7†

43...Rg6 With a draw. I should mention that variations like this are unlikely to happen in a game. There are too many deep ideas to find. But of course, it would have been very unpleasant to face counterplay like this. The whole game you are improving your position and have an advantage and suddenly you miss a tactic and you are facing dangerous counterplay. You will not know if you should continue looking for an advantage or if you should already settle for something less. It is very likely if something like this was to happen, I would have been entirely thrown and lost the game. We have seen this happen to many. But we should also remember that this is an excellent tactic, but I don’t play against the computer; I play against human beings, so these things don’t happen too often. A strong player can find these things, but he would also need to know to look for it. Of course, there is nothing else satisfying, but Navara had a big challenge, which was to decide if it was worth investing his final minutes here or if he should accept that the position is bad and use the time to defend down the line. He chose the latter. This really is the normal trajectory of a chess game. One player gets problems to solve and if he does not solve the minor problems, he will have to solve big problems soon enough. 31...g4 32.hxg4 hxg4 33.Nh4! The only move. 33.Nh2? loses to the typical 33...Qe6! with the idea ...Ng5. 34.Kg1 is an attempt to avoid the check, but after: 278

34...Nd2! 35.Rc1 Nf3† 36.Kg2 Ng5! Black wins. As I found after the game and as given by Navara in his book. 33...Rd2 34.Qf4 Qxf4 35.exf4 Kf6

Black is much better. It is now difficult for White to find counterplay and Navara did not manage either. 36.Kg1 It is a natural plan to try to include the knight in the game again, but this cannot be done while his position is falling apart. 279

A general principle in the endgame is to try to activate your strongest piece first (here the rook). In this case, this means giving up a pawn. 36.Re1! It can be difficult to choose to give up a pawn in a bad position for counterplay. Will it yield compensation, or is it just a panic reaction? With only a few minutes left on the clock, it is hard to make such a decision. But analysis shows us that this probably was the best decision, as the passivity of the white position is a great problem for him. 36...Nxf2 37.Kg1! This is the difficult move to foresee. It is unpleasant to allow the king to be cut off on the first rank, but the activity of the rook is more important.

The material is equal, but of course Black will cash in soon enough. But can White create counterplay? The task for Black is to prevent this. 37...Nd3! The beginning of a great knight manoeuvre without which Black would lose his advantage. It is not the active rook (of which White has one of his own), but the difference in the placement of the knights that gives Black the advantage. 37...Rxa2? is premature and allows White to get back in the game. 38.Re8 Nh3† 39.Kh1! (39.Kf1? loses to the cute tactical idea: 39...Rf2† 40.Ke1

280

40...Rh2! with ideas such as: 41.Rc8 Nxf4 42.Rxc6† Kg5 43.Rxb6 Nh5) 39...Rd2

40.Rc8! gives White a lot of counterplay. If Black has to play passively with the rook to avoid losing this c6-pawn, White will play Rc7-b7, keeping an eye on all of Black’s weaknesses, giving him enough counterplay to make a draw. 38.Re8 Nc1! 39.Kf1 Otherwise the knight comes to d4 and it is game over. 39...Rxa2 40.Rc8 Re2 41.Rb8!? Re6 42.Rxb6 Ne2 43.Kf2

281

43...Nd4 With very real winning chances for Black, but also some drawing chances for White. 36...Rxa2 37.Ng2 The computer suggests this as a moment for active counterplay again: 37.Re1 There are some nice dynamic ideas for White after 37...Nd2 38.Re3, but Black can close them down with: 38...Rb2! 39.Rd3

39...Ne4!, when he is in full control. For example: 40.f3 Nxg3! and the knight comes to d4. 37...Nd2 38.Rd1 282

38...Ke6?! There is nothing really wrong with this, but Navara gives 38...Nf3†! 39.Kf1 Ke6 40.Ne3 f5 with domination as an improvement. Black is a pawn up and in control. I think this should be easy to win. But OK, there are so many subtleties here that it is impossible to get them right during the game. 39.Re1† Kd7 40.Rd1 Kc7 41.Ne3 Nxb3 42.Nxg4

But here we have a real critical moment. We have passed the time control and there is once again time to take in the various subtleties that were beyond reach a few moves earlier. 42...Nd4? 283

This looks very natural. The knight goes to a centrally dominating square. Yet it is unnecessarily time-consuming. The knight was already well placed on b3, where it was constricting the white rook significantly and controlling one of the likely promotion squares of the black pawns. I should have reacted quickly with: 42...b5! The pawns are coming along way too fast. 43.cxb5 An important point is that on 43.Ne5 Black can play 43...f6! 44.Rd7† Kc8 45.Rd3 Nd4 46.Nf7 Kd7 and win. 43...cxb5 44.Rd5 b4 45.Rf5 Kb6 46.Rxf7 46.Ne5 gives Black many options. The simplest is probably 46...Ra7, when the pawns will win any race.

Black is totally winning here. Interestingly we see that the knight does not belong on d4, but that 46...Nd2 and 46...Na5 are the most effective moves. There is an important lesson to be learned here about flexibility, especially when dealing with passed pawns. 43.Nh6!

284

David does not waste a moment looking for counterplay. If White does not hurry up, the black pawns will arrive first for sure. It is the same mistake as on move 28. I make one move by hand and suddenly everything is much harder and there is counterplay to deal with. 43...f5 This was a misjudgement. I wanted to prevent things getting out of control. I was thinking that most likely it was good to go into a total race with: 43...b5 No certainty is offered. When you have the advantage in an endgame, one of the most difficult things is to accurately evaluate when you must allow the counterplay, because you cannot contain it fully, and when you must prevent it. This endgame is an excellent example, where the win is jeopardized because I allowed way too much counterplay. In this case I did not realize just how powerful it is for White to create a passed pawn by brute force. But the complexity in the variations we are giving here should tell their own story. Whether the position is ultimately winning or not is only one dimension. Reality is that I missed the chance to control things on move 38, and now I had to put my material advantage to work. And urgently! The following variation is also by no means trivial. 44.Kg2 The following variation was found by Jacob. Together we only analysed 44.cxb5 cxb5 45.Nxf7 b4 46.Kg2 b3 47.Ne5, where Black’s pawns are faster:

285

47...Rc2! The following is just for an example. What matters is that White will not manage to give up his knight for both the pawns. Together they are unstoppable. 48.g4 c4 49.f5 b2 50.f6 Ne6 51.Rb1 c3 52.Nd3 Kd7 53.Kg3 Ke8 54.f4 Nc5! 55.Nxc5 Rc1 Black wins. 44...f6!

45.g4 There are other moves here; you can always keep on analysing. It is not totally impossible a miracle draw can be found for White here, but we did not find it. 45...Ne2 46.Kf3 Strategically it makes little sense for White to untangle the black pawns. 46.cxb5 cxb5 47.Kf3 Nd4† 48.Ke4 Re2† 49.Kd5 Ne6 wins for Black only because 50.f5 can be

286

met with 50...Re5#. Already this is closer than it should have been. 50.Nf5 c4 51.Ne3 Nxf4† 52.Kc5 Rb2 Black is in control and will win. 46...bxc4 47.g5

47...c3! 48.g6 c2 49.Re1 Rb2! 50.g7

50...Nd4† 50...Rb1 51.Rxe2 c1=Q 52.g8=Q Qh1† 53.Ke3 Rb4

287

This is supposedly winning according to the computer, but it is both unnatural and totally insane at the same time. 51.Kg2 Rb1 52.Re7†

52...Kd6! 53.g8=Q Kxe7 Not that this is much better. White cannot prevent Black from getting a queen, but he can create some counterattack. 54.Qh7† Kd6 55.Nf7† Kd5 56.Qg8

288

56...Rg1†! 56...c1=Q? 57.Ng5† is just wrong. Black is almost lucky not to be mated. 57.Kxg1 c1=Q† 58.Kg2 Qxf4 59.Ng5† Ke5

Black will win in the long run. But again, look at the position after the time control. This got far too close! Let me underline why 42...Nd4? is the mistake. Even if we are able to prove that Black is still winning and the move made in the game does not change the objective evaluation of the game, we should notice the length of the variations and just how many options there are along the way for Black to go wrong. I am not like one of those players that run to the computer after the game, look what evaluation it gives and then go on Twitter and say I made a mistake on move 25 and if I had 289

made this or that move... A real mistake is determined when we look at how the game did and could have proceeded and compare the complexity. We must remember that an advantage is rarely blundered away, but slowly squandered by a series of imprecise moves. Something the engine, which has no way of telling how difficult a move is to find, can tell us little about. To illustrate how difficult it is to assess the nuances once I played 42...Nd4?, I can tell you that Jacob and I have a recording where we both argue, after analysing the game together, that 43...b5 would have been a mistake and 43...f5 was actually a good move. Now that our opinion has turned 180 degrees, we should not pretend that this is done with absolute certainty. Rather, we have gone deeper and this is now our opinion. Maybe in a few years we will come to another conclusion if we revisit the analysis with more powerful engines and hopefully even more experience... But for now, this is what we believe. Black could still have won by force at this point; it was just really difficult. The way I saw it during the game is that to choose between 43...b5 and 43...f5 is difficult and to me the latter just seemed more natural. But as this allows White to get a passed pawn much faster, this was a misevaluation.

44.Kg2! We analysed extremely deeply in order to determine the objective truth of the position and, more importantly, to understand the logic of the position that distinguishes it from other positions – but at the same time will add to our overall understanding of the game, and help form our intuition and potentially verbalize it in a (by definition) oversimplified way. The general impression is that White needs to create a passed pawn to have drawing chances and Black allowed it by not pushing the pawns on the queenside, but manoeuvring and moving the pawns on the kingside. Another conclusion we reached is that almost no matter how winning a position looks, you cannot 290

beat the computer. The following variations are rather dense. I apologise for this. But to understand the position, we have to look at the complexity of refuting this line. 44.g4!? fxg4 45.f5?! 45.Kg2! would transpose to 45.g4 below.

45...Ra3! The only winning move. 45...Ra4 46.f6 Rxc4 47.Kg2! (47.f7? Nf3† 48.Kg2 Rf4 is not going in the right direction.) 47...Ne6 48.Re1 Nf4† 49.Kh2 Ng6 50.f7 Rf4 51.Kg3 Rf6 52.Kxg4 and White will not lose. 45...b5 46.f6 bxc4 47.f7 Ra8 48.Kg2 c3 49.Rc1 c2 50.Nxg4 and we are transposing to the rook endings we will see after 44...b5 below. 46.f6 Rf3 47.f7 47.Nxg4 Rf4! wins. 47...b5

291

48.Rd3 48.Kg2 Nf5! is a useful trick. 48...Rf6 49.Re3 I think this is the best chance. 49.Kg2 Ne6 50.Re3 Nf8 51.Nxg4 Rxf7 52.Ne5 Rf5 53.Nd3 Nd7 Black is winning. 49...bxc4 50.Kf1 Kb6 51.Nxg4 Rxf7 52.Ne5

52...Ra7! 53.Nxc4† Kb5 54.Ne5 c4 Black appears to be winning here, but a lot of play remains. And with that, a lot of chances to make mistakes.

292

44...Ra8 I like this move, although Black is no longer objectively winning. The white pawns are now much more difficult to push forward and the route to a draw is more difficult than in the following line: 44...b5 45.g4 fxg4

46.f5! This is the major difference. Here I want to have ...Rf8 to just grab the pawn. 46...bxc4 This is the only critical move. 46...b4 47.f6 b3 48.f7 Ra8 49.Re1 Kd7 50.Nxg4 Ne6 51.Rb1 Ra3 52.Ne5† and Black even loses his extra pawn. The draw is imminent. 47.f6 c3 48.f7 Ra8

293

49.Rc1 c2 Black’s extra pawn is one of a trio of pawns on the c-line. It is not a miracle that White is able to hold this position. 50.Nxg4 Kd6 51.Ne3 These rather easy moves lead to a drawn rook ending. 51...c4 51...Rf8 52.Nxc2 Nxc2 53.Rxc2 Rxf7 54.Rd2† Ke5 55.f3 is also drawn. 52.Nxc2 Nxc2 53.Rxc2 Kc5 54.Kf1 Rf8 55.Ke1 Rxf7

56.Kd1 White can of course still make mistakes here, but it should be rather simple to hold the draw. 294

I had a version of this endgame with doubled pawns on the f-line against Yermolinsky back in 1987 and got nowhere. It was under less favourable circumstances, but probably it would have affected my judgement during this game as well.

45.Rc1? David falls for the same prophylactic-thinking attitude that I fell for, failing to see that his chances to save the game lay entirely in the immediate search for counterplay. 45.g4! fxg4 46.Kg3!

We have analysed three critical lines here:

295

a) 46...Rf8 47.Kxg4 b5 48.f5 b4 49.f4 Nc2 50.Kh5 b3 51.Kg6 and White is surviving. b) 46...Rh8 47.Nf7 Rf8 48.Ng5 48...b5 49.Kxg4 Kb6 50.Rb1 White has a number of other tries here, but I think this works, so he does not need them. 50...b4 51.Nf3 Rg8† 52.Ng5 b3 53.Ra1 Rb8 54.Ne4 b2 55.Rb1 Ka5 56.Nxc5 Rg8† 57.Kh5 Kb4 58.Nd3† Kc3 59.Nxb2 Kc2 60.Rd1 c5 61.Re1 Kxb2 62.Re5 Rc8 63.f5 Rc6 64.f6 White makes a draw. A nice point is that 48...Ne2† is met with:

49.Kh4!! This amazing move preserves the coordination of the pieces and obviously prevents ...Rxf4. 49...Nxf4 50.Kxg4 White has enough counterplay. For example: 50...b5 51.Nh7 Rh8 52.Ng5 Ne2 53.Ne6† Kb6

296

54.cxb5! Rh6 55.Kf5 cxb5 56.Rd6† Ka5 57.Kg5 The absence of pawns will as a minimum give White the not very dangerous endgame of rook and knight vs. rook. c) 46...Ra4! This is the most dangerous move as far as I am concerned. White has two reasonable tries here. c1) 47.Kxg4 Rxc4 48.f5!? This is the first try.

But Black has a strong option in: 48...Ne6† 49.Kh5 Nd8! 297

This is the key move. 50.Rf1 50.Kg5 b5 51.f6 Rd4 52.Rc1 c4

53.f7 Ne6† 54.Kf6 Nf8 and the black pawns are too fast. The white king and knight may look active on the kingside, but in reality they are cut off from the real action on the queenside. 50...b5 51.f4 b4 52.f6 b3 53.f7 Nxf7 54.Nxf7 Rb4

Who in their right mind would go into this with any sort of certainty about what is going on. Especially from afar? 55.Ng5 b2 56.Rb1 Kd6 57.f5 Ke5 58.Kg6 c4 59.f6 c3 60.Nf3† Kd5 61.f7 c2 62.Rxb2 Rxb2 63.f8=Q c1=Q 298

Black wins in 37 moves according to the tablebases. “Obviously,” a sarcastic soul may feel tempted to say. c2) 47.f5 Rxc4 48.Rd2! 48.f6 Ne2† would decide the game rather quickly.

48...Rc1! We analysed several other options and finally concluded that this is the best try. 48...Ra4 is another line. 49.f6 Ra8 50.f7 Kd7 51.Kxg4 Ke7 52.Rd1 Rh8 53.Kg5 and Black is not winning. 49.Kxg4 b5 50.f6 Kd7

299

50...Rg1† 51.Kf4 Rh1 52.Ng4 Rh5 53.Ke4 Kd6 54.f4 Ke6

55.Ne5 Kxf6 56.Nxc6 Nf5 57.Kd5 c4 58.Nb4 Rh8 59.Ra2 and the white pieces have surrounded the black pawns. The best chance Black would ever have is to win the knight for both the pawns, leading to a not very difficult pawnless endgame.

51.f7! 51.Kg5 looks logical but fails by the tiniest of margins. 51...Rg1† 52.Ng4 b4 53.f3 b3 54.Kg6 Rg3 55.Kg7 Kc7

300

56.Rxd4!? cxd4 57.f7 Rxf3 58.Nf6 Rxf6! 59.Kxf6 b2 60.f8=Q b1=Q 61.Qe7† Kb6 62.Qd8† Kc5 63.Qe7† Kc4 64.Qe6† Kb5 65.Qe5† Ka4!

66.Qxd4† Qb4 Black is objectively winning. Mate in 79 moves says the tablebases. That is a lot of good moves. In practical terms we can say that the game would have been incredibly difficult to win as well as to draw, and we can say with certainty that multiple mistakes would have been made by both sides. And are plausible on each and every move. 51...Ke7

301

52.Ra2! Rg1† 53.Kf4 Ne6† 54.Ke5 Rg5† 55.Ke4 Nd8 56.Kf4 Rg6 57.Ra7† Kf8

58.Nf5 White will not lose. Those wanting to know the truth have been informed, but those wanting to understand chess better have also something to take from this. Black wasted time in pushing his own pawns and as a result the win was in total doubt if White had acted with great energy and pushed his own pawns. If the computer had told us it could find a win in literally 101 moves with best play (if 51.Kg5 was the best move) is irrelevant to the greater lesson of not wasting the moment where the pawns can be pushed, when they are structurally set up as a race.

302

Whether this ending is a draw or winning for Black has nothing to do with decision making. The most we have managed to prove is that chess is difficult. Still, I believe that a good feeling for chess comes from understanding its subtleties. And for this reason, we have analysed some of the endgames in this book to such a great length. 45...Rf8 The counterplay with g4 and Kg3 is now impossible. At the same time White ironically has no other moves than a few rook moves, due to the threat of ...Rf6 trapping the knight eventually on e7, meaning that White has not prevented ...b5 at all. 46.Ra1 46.Re1 b5 47.g4 fxg4 48.Kg3 bxc4 49.Nxg4 c3 50.Ne3 Kb6 51.Kg4 Kb5 52.f5 Kb4 and Black wins. 46...b5

47.cxb5?! This makes Black’s path a bit easier. We also analysed: 47.Nf7!? bxc4! 47...Rxf7? 48.Ra7† Kb6 49.Rxf7 bxc4 does not work.

303

White plays 50.Kf1! c3 51.Ke1 Kb5 52.Kd1 Kc4 53.Re7! and the rook returns to e3 to save the day. 48.Ne5 c3 49.Ra5!?

49...Kd6 Later on we found a simpler way for Black to win. But again, would I have done so during the game? We have already seen my reluctance on the day to part with extra pawns... 49...c2! 50.Rxc5 Kd6 51.Rc3 Ra8 52.Nc4† Kc5 53.Ne3† Kb4 54.Rc4† Kb5 and it is time to give up the exchange and perhaps also sign the scoresheet. 50.Nc4† Kd5 51.Ne3† Ke4 52.Rxc5 Kd3 53.Nd1 c2 54.Ne3

304

54...Rf6! 54...Re8? has the idea to meet 55.g4? with 55...Rxe3!, when Black wins in a beautiful long line – which we shall skip, as White can instead play 55.Nxc2! Nxc2 56.Rxc6 and afterwards exchange the f5-pawn with f3&g4. 55.Kh3 Kd2 56.Nf1† Ke2 57.Kg2 Re6 58.Ne3 Rxe3 59.fxe3 Kxe3 60.g4!?

60...Kd2! Black wins. The last trap is for Black to be careless and play 60...fxg4?, when after 61.f5 Kd2 62.f6 c1=Q 63.Rxc1 Kxc1 White can play 64.Kg3! Kd2 65.Kxg4 Ke3 66.f7 Ne6 67.Kf5 Nf8 68.Ke5! and 305

eliminate the last pawn, with a draw as a result. 47...cxb5 48.Nf7

48...b4 A flashy win would be: 48...Rxf7 49.Ra7† Kd6 50.Rxf7 c4 and the pawns are unstoppable. 49.Ne5 b3 50.f3 Kd6 51.Ra6† Kd5 52.g4 Rb8 53.Nd7 b2 54.Nb6† Kc6 55.Na4† Kb5 56.Ra7

56...b1=Q 57.Nc3† Kb6 0–1 306

307

Chapter 7 Choosing the Right Transformations

Playing Caruana at the Batumi Olympiad in 2018

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

308

What is Black’s best move? (see page 165)

White is under pressure. How should he defend? (see page 166)

309

Black has the advantage, but how to use it? (see page 168)

Things have improved, but what now? (see page 171)

310

Black still has the edge in the knight ending. But what next? (see page 173)

What is strongest for Black? (see page 174)

311

A last-minute chance to save the game (see page 178) Fabiano Caruana – Boris Gelfand Amsterdam 2010 This game was played in one of the tournaments played in the “Rising Stars against Experience”, a tournament which existed for a few years and where the winner among the youngsters qualified for the Melody Amber tournament. It was one of Joop van Oosterom’s tournaments. As the level of the youngsters was very high, the organizers found it necessary to invite stronger rather than older players. This included Peter Svidler, who was only 34 years old. All the juniors who played in 2010 are in the Top 50 today. Hikaru Nakamura, Anish Giri, David Howell, Wesley So and of course Fabiano Caruana. Four of them crossed 2800. I played well in the tournament and was the highest scorer with 7/10 – four wins and six draws. Svidler was second with 5½ points. In the Rising Stars section, Nakamura and Giri shared first with 6/10. Caruana was one of the big talents at the time when the game was played. He had just turned 18. He famously left the US for Europe with his family to focus on improving his game and for years played for Italy, although I don’t think he ever lived there. Now of course he is the No. 2 rated player in the world and came close to dethroning the World Champion in the 2018 match. Almost as close as I came in 2012...

312

Although he was not the youngest, he impressed me in the tournament. His style was wellfounded. Although he was very concrete, as young players usually are, you could see that his play was developed while working with experienced players. He played normal classical openings, trying to get an advantage with White and equalize with Black. He combines that attitude of being ready to “just play” any position without serious preparation in a way that offers a significant advantage over many of his competitors. There are of course many ways to play chess, but this mature approach from a young player made an impression. In 2009 he visited me for one week. He is the only human who has ever managed to “out-work” me. Of course super-human Korchnoi managed to do so easily. Into the night he was still coming up with interesting suggestions and had the energy to go on and on. Recently I checked some of my older analysis, including the ones done with Fabiano. To my surprise they are largely accurate even today, despite the enormous rise of the engines. 1.d4 I had expected 1.e4, which was his primary choice at the time. 1...Nf6 2.c4 e6 3.g3 The Catalan, which is my favourite opening. At the time Caruana was working with Boris Avrukh, who is a big specialist in that opening, having written two seminal books on it in the Grandmaster Repertoire series. Caruana also worked with Razuvaev, who was another great expert in this opening. 3...d5 4.Bg2 Bb4†

There are many playable systems against the Catalan. I like this system. It gives interesting 313

games with chances for both sides. I had played this before, in the Dresden 2008 Olympiad against Kveinys, in a really interesting game we might get to see in later volumes. 5.Nd2 One of the three main check-blockers (the last one is less attractive). I have played it myself a number of times, for example against Almasi. Later in Zurich 2014, we played the same variation again. At that time we played 5.Bd2 Be7. Fabiano won the game after a long fight. See also the game with Ivanchuk on page 31. 5...0-0 6.Ngf3 dxc4 This is the sharpest way to play the position and the more ambitious way to react to 5.Nbd2, but also requires bigger knowledge, as the game gets more concrete. 6...b6 is another variation, transposing into a kind of Queen’s Indian with a complicated positional struggle. 7.0-0 b5 8.a4 c6 9.b3

9...c3 This makes most sense to me. The idea is to give back the pawn and follow up with a quick ...c5 to get active play. I am certain this is the most practical way to play. If Black plays something like 9...cxb3 10.Nxb3 Bb7, White could play something like 11.Bd2 Bxd2 12.Qxd2 followed by Na5 and Rfc1 with a lot of compensation for the pawn. Some may play like this with Black, hoping that they would have an extra pawn at the end of the day. But I do not see this as a very realistic way to play for a win. White always has compensation and will at least get back the 314

pawn. Despite all these reservations, I should say that it is certainly also unattractive. 10.Nb1 Bb7 11.Qc2 bxa4 This was my idea. I was certain that Fabiano had prepared an improvement over his own game: 11...Nbd7 12.Nxc3 Rc8

13.Ne1 Qb6 14.Nd3 ½–½ Caruana – Almasi, Reggio Emilia 2009. 13.Ne5!? seems a very likely improvement. 12.Rxa4 12.Nxc3 was also possible. Black is not short of options and could try something like: 12...axb3 13.Qxb3 a5 14.Na2 c5 15.dxc5 Bd5 where Black has no problems whatsoever. Total liquidation is possible, with an imminent draw. 12...c5 13.Nxc3 a5!N Against Kveinys I kind of blundered and was quickly in trouble after 13...Bc6 14.dxc5! Bxc3 15.Rh4!?. The game continuation is a significant improvement. The bishop is supported and Black is ready to develop the knight, with a comfortable position out of the opening.

315

White already has practical problems. The key issue is that the b-pawn is on b3. If it was on b2, his position would hang together easily. The stability of the knight on c3 would be greatly enhanced. What we see now is that it is not so easy to bring the pieces into the game in a harmonious way. The circumstances were not in Fabiano’s favour here. This game was his last chance to fight for first place among the Rising Stars and the qualification to Melody Amber. So it was not surprising that he was slow to realize that he needed to look for a way to equalize quickly, before he would get into trouble. Actually, this is quite an interesting discussion. One view is that you should always fight and will always get chances. But reality is often different. In this game White never had any chances to be better, but had to defend for many hours. A realization that this was the way things were heading would have been appropriate. There are objective laws of chess, which no amount of self-help promotional catchphrases can defeat. 14.Na2?! 14.dxc5 Na6 15.Ra1 Nxc5 16.Na2 would be a way to seek exchanges and a draw. It looks cowardly, of course, but sometimes it is better to live as a coward than die for nothing. 14...Be4! A typical idea for this type of position. The white queen does not have a good square to go to and the bishop is very active on e4. Even without understanding concretely why it could be advantageous to play like this, it makes sense to play it. Analysis shows that the main difference is that after 14...Nbd7 15.Bg5!? Black does not have 15...Nb6, as after: 16.Nxb4 Nxa4 316

There is 17.Nd3! Nb6 18.dxc5 with compensation for the exchange. 15.Qd1 The engine says that 15.Qb2 is preferable and that after 15...Nbd7 16.Nxb4 cxb4 Black is only a little better. Mainly the queen is not in the way, which is a very poor ambition to have on behalf of the queen. 15...Nbd7

Now the position is not at all easy for White.

317

16.Nxb4 Caruana offered a draw, probably affected by the difficulty in finding anything that was worth playing. White is already somewhat worse and it is hard to imagine how he will ever win the game. After the game I was asked by a journalist if this draw offer was a sign of cowardice, which I rejected. Rather I see it as good understanding of the position. There are no targets in the black position. 16.dxc5? Nxc5 drops the exchange. And after 16.Bd2?! Black can either win the exchange with 16...Bc6 or fight for a positional advantage with: 16...Nb6 17.Nxb4 cxb4 18.Ra1 Qd5, where the bishop is poorly placed on d2 and Black is experiencing an almost obscene level of harmony. 16...cxb4! This consolidates Black’s control of the dark squares. 16...axb4? would be a difficult move to explain. At least accept the draw offer immediately and don’t waste any time. 17.Bb2 Against 17.Ra1 I had intended 17...h6 to keep the bishop out of the game, but the computer gives 17...Nb6 18.Bg5 Nbd5 targeting the c3-square, which is also very strong. 17...Nb6 18.Ra1 Qd5

The advantage is already quite big. In a way Black’s idea is very simple. It is easy to see what I

318

should do: play ...a4 and get the knight to c4. But I have to prevent White from organizing any counter-chances. Actually it is pretty astonishing that White plays natural-looking moves and before move 20 he is already in a lot of difficulties. Caruana’s situation was very difficult. He came to the game with a hope to fight for first place and in the opening he already has to fight hard not to lose quickly. 19.Ne1 Passive, but there are no other moves. White would love to play f3 & e4, but cannot do it in another way. After 19.Nh4 Rfc8! Black’s advantage is increasing.

This is the first really important moment for me in this game (my opponent experienced such a moment on move 14). Although I wanted to prevent unnecessary counterplay, I fell prey to simplistic thinking and played what I thought I “had to”, rather than what I wanted to. This is a far more common mistake than you would think. A player assumes something and does not question his own assumption and for this reason misses something he would see immediately if he had actually been thinking. 19...Bxg2? I thought that this move was forced, but actually I missed a tactic I should not have. When looking at this again, I immediately questioned my logic, realizing that: 19...Rfc8! This was much more energetic. 20.f3 White has to go for this critical move, or everything he is doing is pointless. 319

20...Bc2!! 21.Nxc2 Qxb3 22.Nxb4 22.f4 Nfd5 23.Nxb4 Qxd1 24.Rfxd1 axb4

This leaves Black entirely dominating; ...f5 and ...Nc4 are both coming. 22...Qxb2 23.Rb1 Qc3 24.Na2 Qe3† 25.Rf2 Nfd5 26.Rb3

320

26...Nc3 27.Nxc3 Rxc3 28.Rxc3 Qxc3 Black is already in conversion mode, although White should still fight. 20.Nxg2 Qb5 Creating a passed pawn. 21.Nf4 a4 22.bxa4 Nxa4 22...Rxa4 23.Qd3 would also give White a chance to get closer to a draw.

23.Qb3?! White would have been able to solve his problems more easily with: 321

23.Qd3! Qxd3 23...Qc6 is harmless, unless White falls for 24.Rfc1? Nxb2!. 24.Nxd3 Nxb2 24...Ne4 25.Rfc1 and White is ready to bring in the king. Finally he is close to equalizing. 25.Nxb2 b3

26.Nd3 Black’s advantage is minimal. Still it is not easy to enter an endgame with this b-pawn lurking around on the 3rd rank. Emotionally it is easier to block it. 23...Nd5! Bringing in the knight is very logical.

322

24.Nd3? Against: 24.Nxd5 I was planning to play: 24...exd5 This is the most logical. My queen is more active and if we were to exchange into a pure rook endgame, the white king would quickly run to b3, making the passed pawn a liability as much as a strength. 25.Rfc1 Nxb2 25...Qxe2 26.Ra2! would give Black little. The b-pawn will be gone in a moment. For example: 26...Qb5 27.Rca1 Nb6 28.Rxa8 Rxa8 29.Rxa8† Nxa8 30.Bc3 26.Rxa8 Not taking transposes as Black would take on a1. 26...Rxa8 27.Qxb2 b3 28.Rc3 Rb8

323

This position looks very unpleasant for White and I would have been surprised if Caruana had gone for it. But upon closer inspection, White has nothing to fear. The rook will shuffle along the third rank aimlessly, asking what Black wants to do. Not much is probably the realization! If Black was to rush his queen to a2, White would play Rc2!, chasing it away. If Black first had removed his king from the 8th rank, White would always answer ...Qa4 with a rook move that makes it possible to play Rd2 or Rb1. The queen is perhaps a clumsy blockader, but she is also very effective. 24...Nac3 24...Ndc3 is likely to transpose into a very similar endgame. In some lines this might have been more accurate, but on the other hand, I forced him to take with the bishop on a1, which also looks bad. During the game these considerations are luxury problems, not worth wasting your time on. Black is much better in both cases. 25.Rfe1 After 25.Bxc3?! Nxc3 26.Qb2 Rab8 White is struggling to find a way to deal with ...Rfd8, targeting the d4-pawn. Whichever way he finds to defend the pawn, it will involve a lot of concessions. Often two knights defending each other are not very effective. Here it is mainly because the knight on d5 is blocking the d-line. 25...Rxa1 26.Bxa1 Rc8 Black has achieved full harmony with his pieces and should be looking for a way to use them. 27.Bxc3 324

Caruana decides to get rid of his bad bishop while he has the chance. Earlier it was too early; later might be too late. 27...Rxc3

28.Rc1 Caruana decided that this was the right moment to get rid of the rooks as well. 28.Ra1 h6 29.Qb2 was also possible. Black has an easy plan, playing ...b3, ...Qc4, ...Kh7, posing White a lot of problems. It is not yet a forced win, but it is getting closer. This does not mean that exchanging the rooks is a good idea, just that the position is bad. And in bad positions there are rarely any good moves. 28...h6 Of course I was tempted to sacrifice the queen with 28...Qxd3 29.exd3 Rxc1†, but after 30.Kg2 there is no meaningful way to continue. White puts the queen on b5 and waits. He will always have annoying checks. The pawn will be on b3, but will it ever get to b2? 29.Rxc3 Nxc3 It is relevant to think about what this ending would look like as a knight ending, as often Black would be happy to get rid of the queens on the grounds of their ability to create counterplay. I had an ending that is somewhat similar just a few months later. Although not directly transposing, I still find it relevant. We will consider this on page 179. 30.Nc1 325

30.Nxb4 Nd5 and Black wins.

30...Qg5! Good technical play in the spirit of Capablanca. Winning slowly. It is easy to fall under the illusion that Black should try to win on the queenside only. But an experienced player will know that the easiest way to convert an advantage is to create more weaknesses in his opponent’s position and thereby increase the advantage. Specifically here the f3square, which fits the knight like Cinderella’s glass shoe. I was also thinking about this line: 30...Qa6!? Threatening ...Qa1. 31.Kg2 Qa8†! 32.f3 Qa3 33.Qxa3 bxa3 34.Kf1 a2 35.Nb3

326

I was not sure that it would be sufficient to win. My feeling was that it was, but it would require very precise calculation, especially because the king is out of the game at the moment. Looking deeper at the position confirms my feeling that Black is winning if he plays good moves. The white king will have to go to deal with the a-pawn and meanwhile Black will be able to attack the white kingside. 35...Nd5!? This is very concrete, so obviously suggested by a computer. I would believe that there are other less demanding winning lines, but as the game is likely to be very sharp, depending on the utmost accuracy at some point, I find it a fitting illustration of how the knight ending plays out. 36.Ke1 36.e4 Nb4 37.Ke2 Nc6 38.Kd3

327

38...e5! 39.dxe5 Nxe5† 40.Ke3 g5 wins in the long run. The white kingside is too compromised for White to be able to eliminate the a-pawn. 36...Ne3 37.Kd2

37...Nf1†!! This idea is truly spectacular. Black gives up the knight in order to have passed pawns on both sides of the board. 38.Ke1 38.Kc3 Nxh2 followed by the creation of a distant passed pawn is not a serious defensive try for White. 38...Nxh2 39.Kf2 Kf8! The king needs to go to the queenside. There is no time to waste. 40.Kg2 Nxf3 41.Kxf3 41.exf3 h5! will ensure that there is a permanent passed pawn on the far kingside that will prevent White from going to the queenside with the king. 41...f5! The following endgame hangs on one tempo. 42.g4 g6 43.gxf5 gxf5 44.e4 There are other lines, but no other results. 44...fxe4† 45.Kxe4 Ke7 46.Ke5 h5 47.Na1 h4 48.Kf4 Kd6 49.Nc2

328

49...Kc7! 50.Kg4 Kb6 51.Kxh4 Kb5 Black wins by exactly one tempo. 52.Kg4 Kc4 53.Kf4 Kc3 54.Na1 Kb2

If White was able to play 55.Kd2(!!), then 55...Kxa1?? 56.Kc2 would even win for White after 56...e5 57.d5. But he is not and has to resign... 31.e3 Qb5 At this point I had the wrong impression of the game. I was under the impression that the game should be won more or less immediately, which led me to look for forcing wins, rather than continue in the spirit of Capablanca and tie down my opponent. 329

The computer points out that Black had a nice finesse with 31...Qf5! 32.Kg2 Qe4† 33.Kg1 g5 and Black keeps improving. 32.Kg2 This walks straight into an unpleasant check. But slowly waiting for Black to play ...g5-g4 would also be torturous. Still 32.Qc2 with the idea Nb3 was forced, after which I still had to squeeze the life out of him slowly. I would do it of course. But I would have to be very cautious not to let him slip away at some point. Now Black gets the d2-square for the knight and there is no need for Capablanca anymore. 32...Qc6† 33.Kg1 Ne4 34.Nd3 34.Qb2 Ng5 is equally horrendous. The queen ending after 35.h4 Nh3† 36.Kh2 Nxf2 37.Qxf2 Qxc1 is hopeless. 34...Nd2 This is very strong, but I could also have picked up a pawn with: 34...Qb5!? 35.Nc1 (35.Kg2 Nd2 is an improved version of the game. Black will get ...b3 quite quickly.) 35...Nd2 36.Qd3 Nf3† 37.Kf1 Qc6, which is of course similar to the game, but if I wanted to pick up the h-pawn, this is the most accurate way to do so. 35.Qb2 35.Qxb4 Qc2! is important to notice. 35...Nf3† 36.Kf1

330

36...Qb5? This is still good enough to win, but I am making the conversion harder – meaning more reliant on accuracy down the line. And this in a situation where accuracy was not demanding. As said, the way I played did not give me a good opportunity to take the h-pawn. 36...Nxh2†?! 37.Ke2 Qf3† 38.Kd2 Ng4 39.Ke1 is still a big advantage for Black, but only because he can create a passed pawn with ...g5 and ...h5. Still it is definitely a step in the wrong direction. What I did in the game was not truly wrong, but I could have done it more accurately with 36...Qc4!, where I get the pawn to b3 without giving him the c3-square. This is a half-move variation and really something I should have seen. But this is of course the case with mistakes. They are not about what we can see and not see. But about how often we get it right. Carlsen gets it right slightly more often than the rest of us, which is why he is Number One. No one gets it right all of the time. Black should win rather easily, as I would not have seen, but worked out afterwards. 37.Ke2 b3 38.h4

Black has many ways to cross the finish line from here, but if I had this position during a game, I would look for a human-style forced win. And it is there: 38...g5! When we were analysing this position I saw this immediately, proving I would have found it in the game as well. 39.hxg5 Nxg5! 40.g4 Ne4 41.f3 Nf2! 42.Kxf2 Qxd3 and White needs to resign. 37.Ke2 Ng1† 38.Kd2 Nf3† 39.Ke2 b3 40.Qc3 40.Qa3!? was offering slightly more resistance. The key idea is that ...Ng5-e4 no longer exists. Black should still win after 40...Nxh2, but at least Black would have to show some accurate moves.

331

40...Qh5? This looked very tempting with so little time on the clock and it is not a bad move as such, as Black is still winning. But still, 40...Ng5! would have more or less won immediately. The key line is 41.f3 Qf5! and if the white knight moves, 42...Qb1! comes. Probably Fabiano would have come up with 42.Qc8† Kh7 43.Qb7, offering a bit more resistance.

Black is still winning after 43...Qh3 44.Nf2 Qxh2 45.Qxb3 Nh3 46.Qb1† f5 47.Qf1 Ng1†!, where both the f- and g-pawns are falling. 332

41.Nf4

41...Nxd4†? This is probably the biggest mistake in this whole knight and queen ending. At move 41 I should definitely have spotted 41...Qb5†! forcing a repetition, giving me more options. Basically, I would have a more realistic chance to spot 42.Nd3 Ng5!, which is not so easy to spot. It takes time to realize how strong it is. We already pointed out in Positional Decision Making in Chess in the game with Malakhov (page 43) that moves 41 and 42 are common places to err. Sadly I did not have a chance to read this yet, as at this point we had not written the book. 42.Kd2 Nf3† 43.Kc1 Qxh2 44.Nd3 Qh1† 45.Kb2

333

45...Qd1? This is a very understandable decision, but in retrospect it is not clear at all that the exchange of queens is in Black’s favour. White’s queen is not creating any counterplay at this point. 45...h5! 46.Qxb3 g5 was strong, although it is less winning than it once was. 46.Qxb3 Qxb3† I already made this decision, but of course I could have changed my mind. 47.Kxb3

334

The knight ending is of course very promising. But there are certainly also real drawing chances for White. The small amount of pawns is a problem, as is that the creation of a passed pawn would require further exchanges of pawns. My evaluation of this ending now is that a draw is at least as likely an outcome as a win with accurate play. But the key word is of course “play”. Both sides have to make moves and just because Black has misplayed his position does not mean that he should stop doing his best. A lot of the following analysis is to a great extent computer assisted. The endgame is very complex and what we were thinking during the game is very different from what deep analysis shows afterwards. There you try to avoid obvious mistakes and to implement the plan of creating a passed pawn in the h-file, and preferably keeping the king in front of the h-pawn in order to control the promotion square. 47...Kh7 48.Kc2 Kg6 49.Kd1 Kf5 50.Ke2 Kg4

Black has certainly implemented the most natural plan. But of course ...e5 could have been tried at some point. During the game I thought it would be winning if I got the king to g4. Now I am less convinced. 51.Nc5 h5 52.Nd7? 52.Ne4! was necessary, when the game hangs in the balance. 52...f6? 52...e5! would have blocked the white knight out of the game. Kf1 is always met with ...Kh3. 53.e4

335

53...f6! Black is winning. Next follows ...Nd4† and either ...Kf3 or ...Kh3-g2. Then it is time for ...g5 & ...h4. 53.Nb6 f5 also wins for Black. The plan is ...Ng5 and ...f4 in the right moment, forcing White to take twice on f4, when the h-pawn dominates events and Black will get his king to f3 or to h3-g2/h2 at some point, preparing the coronation of the pawn. 53.Nf8? Fabiano had a chance to force a draw here. 53.Nc5! e5 53...Ng5 54.f4! and there is no ...Ne4 is the difference from the game. 54.Ne6 g5 54...g6 55.Nf8! would not give Black chances to create any passed pawn either. 55.Nc5

336

55...Nh2!? 55...f5 56.Nd7! is an important finesse. Without this, White would be totally lost, but it is there and he is fine. 56.Nd7! Kf5 57.e4†! White has reached an endgame he can hold. Black’s winning chances are closely related to the active king.

53...Ng5! 53...e5? with the idea ...Nh2 and ...Kh3 is devious. But White can save the game with 54.Ne6! g6 55.Nf8 and again Black has to go for 55...h4 56.gxh4 Nxh4, which holds no chances.

337

54.Ng6 54.f4 Ne4 55.Nxe6 g5 and Black wins. The two passed pawns are unbeatable. 54...e5 55.Kf1? 55.Ne7! was the only move. White is probably holding on, but it would take very deep analysis to get to a solid conclusion. 55...Kh3 Black had a beautiful shortcut available in 55...Nf3!.

56.Kg2 h4! is a key point. After 56.Ne7 Kh3 57.Nf5 g5 Black wins; ...h4 and ...Kh2 will follow – and at the right moment, ...Nd2†, forcing the white king away. 56.Nh4 Ne4 57.Ng6 If 57.Nf5 Black continues with natural play: 57...g5 58.Kg1 Kg4 59.Ng7 h4 60.gxh4 gxh4 followed by ...Kf3 and Black wins.

338

57...Kh2 58.Nh4 g5 59.Nf5 Kh3? This was a very poor decision. The nearing of the second time control of course affected my reluctance to take chances. I saw and should have played 59...g4!?, which is easy to say now, but during the game it was not so easy to decide on sacrificing a piece. The problem was to some extent that I felt it was winning all the same. But both in practical terms and in objective terms, this was a terrible misevaluation. As we have seen above, the endgame was constantly in the balance and very difficult to play for both sides. Having seen something concrete, I should have gone for it. Of course, it is not impossible to imagine a scenario where something direct fails and then the argumentation would be reversed, so this type of logic does not work during the game. What did work was that I saw and understood the strength of this continuation, even though I did not have enough time to work it all out. My real mistake was for that reason a misevaluation of the knight ending, which consisted after playing it for a few hours and analysing it back in 2010. It is only with the tools available to us ten years later that we can determine the correct evaluation with the level of accuracy achieved here.

339

There are three meaningful variations we should look at. White was threatening 60.Ng7, so it is important to see that Black has some tactical options like 60...h4 61.gxh4 Nxf2 or 60...Nxg3† 61.fxg3 h4 for example. After 60.Nh6 Nxg3†! 61.fxg3 Kxg3

62.e4, Black wins again with very logical, but also very necessary moves. However, they are not too hard to work out. (62.Kg1 Kf3 63.Nf5 transposes to the 60.Nh4 Nxg3†!? line below) 62...Kf3! 63.Ng8! h4 64.Nxf6 h3

340

Threatening ...h2. 65.Kg1 g3 Followed by ...h2† and ...Kf2. 66.Nh5 however forces Black to find:

66...h2† 67.Kh1 g2†! 68.Kxh2 Kf2, when mate is near. Against 60.Nh4 it makes sense to stop for a moment.

341

60...Nd6! This is the simpler way to do it. 61.Ng6 Nf5 White now has to make a concession. Against most moves ...Nxg3† comes with additional power, as the white pieces are worse placed. And against 62.e4 Black wins another pawn with 62...Nd6. Black also wins after 60...Nxg3†!?, but the line is unnecessarily complicated: 61.fxg3 Kxg3 62.Nf5† Kf3 63.Kg1 Ke4 64.Ng3† Kxe3 65.Nxh5 f5 66.Kg2 e4 67.Ng7 (67.Kg3 Kd2!) 67...f4 68.Nh5 f3† 69.Kf1

Because the white knight is coming back to g3-e4, Black’s cleanest win is 69...Kd4!? 70.Ke1 e3 71.Ng3 Ke5 with total domination and eventually a breakthrough with the king along the h-file.

342

60.Ng7? At this point White could have saved the game with: 60.Ne7! h4 60...Kh2 61.Nf5 and there is no obvious way for Black to make progress, although he can still think up problems for his opponent. 61.gxh4 gxh4 62.f3 This is analogous to the game with a small but important difference. 62...Ng3† 62...Nd2† 63.Kf2 e4 64.Nd5 f5 65.fxe4 Nxe4† 66.Kf3 and White holds. 63.Kg1 f5 64.Nd5!

343

The difference. White takes control of the c3-square. 64...Ne2† 65.Kf2 Nc1 66.Ne7! Nd3† 67.Kg1! After the exchange of pawns, White will make a draw. 60...h4 61.gxh4 gxh4 62.f3

This is the final difficult moment for me in the game. 62...Ng5? This is a very typical mistake. I was hesitating too much at many moments in this game and eventually when I had to do something, I was not ready for it. Sometimes changing your approach to the game is really difficult. Black has three ways to win this game. 62...Nd2† is very similar to the game. Except the knight is much better placed on d2, where its control of the f1-square shines through in the following variation: 63.Kf2 e4! 64.Ne8 f5 65.Nd6 exf3 66.Nxf5

344

66...Kg4! 67.Nh6† Kh5 The key reason for putting the knight on d2 is that after 68.Nf5 Black has access to the g5-square and wins after 68...Kg5! 69.Nd6 Kg4, when the knight is dominated. And after 68.Nf7 h3 the pawn is unstoppable on account of 69.Kg3 h2 70.Kxh2 f2. A very artistic way to win the game was: 62...Ng3† 63.Kg1 Not very difficult, but still beautiful is: 63.Kf2 Kh2 64.e4

64...Nh1†!! and the black king comes to g2 or g3, winning the game. 63...f5 64.Ne6 Ne2† 65.Kf2

345

65...Nc3! 66.Ng7 Nd1†! 67.Ke2 Kg2! The h-pawn is crowned as the winner. 62...Nd6 63.Kf2 f5 wins more slowly. Notice that the white knight is unable to attack the e-pawn and the h-pawn. 63.Kf2 e4 This was of course my idea.

64.fxe4? Fabiano missed the final chance to save the game. 64.Ne8!! f5 65.Nd6! and White makes a draw. 346

(But not 65.Ng7? f4! and Black would win.) 64...Kg4 65.Ne8 Nxe4† 66.Kg2 f5 67.Kh2 h3

68.Ng7 Against 68.Nc7 I was planning 68...Nf6 69.Ne6 Kh4, although 69...Nd5! is also very tempting. 68...Nd2 69.Kg1 Nc4 The simplest. 70.Ne8 70.Kf2 Nxe3! 70...Nxe3 71.Nf6† Kg3 72.Nh5† Kh4 73.Nf4 Kg3 74.Nh5† Kg4 75.Ng7 f4

347

76.Ne8 f3 77.Nf6† Kf5 78.Nh5 Ng4 79.Ng3† Kf4 80.Nf1 Ke4 81.Ng3† Ke3 82.Nf1† Kf4 83.Kh1 f2 84.Nd2 Ne3 85.Nf1 Kf3 86.Ng3 f1=Q† 0–1 A good question is if you want to view my mistakes in this game as many mistakes or if you want to group them as three mistakes. Carelessness on move 19. Exchanging the queens was definitely a mistake and a lot of my thinking regarding this was just wrong. Finally, the reluctance to do something active in the knight ending. Especially on move 59. I care less about the various finesses as uncovered by the computer. Analysing the game afterwards is useful for a lot of things. These kinds of discoveries enhance my feeling for chess and in this case especially the geometric relationship between the pieces. However, analysing my own actions and especially looking for systematic mistakes and misunderstandings (the exchange of queens for example) gives me a chance to improve my approach to the game, which I see as more farreaching than some finesses in a knight ending, no matter how interesting they are.

A similar knight ending The following game shows a knight ending with the same pawn structure, although in a slightly better set-up for the defender. It is not directly relevant to the Caruana game, but it is interesting to look at as an addition to it. Boris Gelfand – Kiril Georgiev Plovdiv 2010 348

41...Nxa4 42.Nxc6 h5 43.d5?! Unhappy that he had the d5-square, I probably made a mistake by reducing the number of pawns on the board. 43.Kf3 with slow improvement was better. 43...exd5 44.Ne7† Kf8 45.Nxd5 Nc5 46.Nf4 Nd7 47.Nd3 Ke7 48.Kf3 g6 49.Ke4 Nb6 50.Kd4 Ke6 51.Nf4† Kf5 52.Nd5 Nd7 53.Ne7† Ke6 54.Nc6 Nb6 54...f6!? was already possible, with the intention to liquidate the game. 55.Nd8† Ke7 56.Nb7 Ke6 57.Nc5† Ke7 58.f3 Na8 59.Ne4 Nc7

349

60.g4 According to the computer White loses his advantage with this move, but it would criticize any attempt to do something, making it entirely misleading. The computer is a useful tool, but if you do not know how to interpret its findings and take everything it says as instructions from a divine being that should be obeyed, you are the tool. 60...Ne6† 61.Ke3 hxg4 62.fxg4 Kf8 63.Nf6 Nc5 64.Kd4 Ne6† 65.Ke4 Nc5† 66.Kd5 Nd3 67.Kd6 Nf4 68.Nd5 Nd3 69.Ne3 Kg7 70.Kd5 Nf4† 71.Ke4 Ne6 72.Nd5 Nc5† 73.Kd4 Ne6† 74.Ke3 Nc5 75.Nf4 f6 76.h5!?

76...g5? 350

Having been under pressure for a very long time, Black errs. 76...fxe5 77.Nxg6 Kf6 was one way to make the draw. Now White wins. 77.Nd3 Ne6 78.Ke4 Kf7 79.Nb4 fxe5 80.Kxe5 Nf4 81.Nd5 Nh3 82.Nc7 Kg7 83.Ne6† Kh7 84.Kf6 Nf4 85.Nxg5† Kh6 86.Nf7† Kh7 87.Kf5 1–0

351

Chapter 8 Karjakin

Tal Memorial, Moscow 2018

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

352

Let’s start with an easy one (see page 185)

How should White proceed? (see page 187)

353

After a long think I found a strong plan here (see page 187)

How should White play for an advantage? (see page 189)

354

How did Karjakin defend against Rxe6? (see page 192)

Black could still have held the game at this late point (see page 195) Boris Gelfand – Sergey Karjakin Nalchik 2009

355

1.d4 d5 2.c4 c6 3.Nf3 Nf6 4.Nc3 a6 The Chebanenko was fashionable at the time. One of its attractions was that it was very easy to prepare for Black. 5.e3 b5 6.c5 6.b3 is another possibility with many games played. I personally played it against the legendary GM Vlastimil Hort in the 1990 Olympiad in Novi Sad. 6...Nbd7 7.a3 This prepares b2-b4. Black’s main idea is of course that 7.b4 is met with 7...a5 and Black is fine. I had 7.Bd3 e5 against Aronian in a sharp game. 7...a5 This is natural. Later on we shall see some of the advantages of including the moves with the apawns.

8.Qc2!?N This is a prophylactic idea, but also I wanted to play e4 quickly. 8.Bd3 e5 9.dxe5 Ng4 10.Bxb5! cxb5 11.Qxd5 Rb8 12.h3 was Aronian’s great preparation against Grischuk in Bilbao 2009. After 12...b4 13.axb4 axb4 14.Ne4 Black’s position was already desperate and Aronian won the game on move 42. 356

Of course it was possible to follow in Aronian’s footsteps, but I was happy to play something new. 8.Qc2 was the idea of Maxim Rodshtein and after analysing it together with Alex and Maxim, we came to the conclusion that it was a clever idea that deserved testing. I am not sure if this was played by an amateur beforehand, but it was definitely the first game played with it on a high level. 8...Qc7 Preparing ...e5, forcing White to go for the central break himself. Karjakin’s thinking might have gone something like this. 8...e5 does not work, so I have two reasonable options, 8...Qc7 and 8...g6. It is almost impossible to decide during the game which is best, so it is better to play one that looks reasonable and save time for the problems that will inevitably come later no matter what. 8...e5 This does not work now. 9.dxe5 Ng4 10.Nd4! Ndxe5 10...Bb7 11.f4 Bxc5 12.Be2 was also unpleasant for Black.

Here my idea at the time was: 11.Ncxb5! Ten years later the engine says that 11.h3 Nf6 12.f4 is even stronger, but from a practical point of view, it makes more sense to take the extra pawn and enjoy a good pawn structure to match. White follows up with Bd2 and Rc1. 11...cxb5 12.Bxb5† Bd7 13.c6 Nxc6 14.Nxc6 Qc7 15.Qa4 357

White has won a pawn. A later game was played in a critical line. 8...g6 9.e4 (9.g3!? is the 2019 engine suggestion) 9...dxe4 10.Nxe4 Bg7 11.a4 b4 12.Bc4 0-0 13.0-0 Nxe4 14.Qxe4 Nf6 15.Qxc6 Bd7 16.Qb6 Qxb6 17.cxb6 Rfb8 18.Ne5 Be8 Black later won, but this was not because of this position, where Black is probably close to equal, Eljanov – Movsesian, Sochi 2012. 9.e4! Nxe4 10.Nxe4 dxe4 11.Qxe4 Nf6 12.Qe5!

This was the key point of our preparation. 12...Qxe5† Avoiding the exchange of queens with 12...Qb7 is not necessarily bad. White is better after 13.a4!?. The concept for White is that after Black has played 7...a5, White will always be able to create a good square for his pieces on c4. 13...b4 14.Bc4 is an example of this. 13.Nxe5 Bb7 14.a4 As said, this is essential. If Black’s pawn was still on a6, he would have no problems whatsoever. 14...b4 Now it is a good question what to do. I decided to keep the option open on which piece to put on c4. 15.Bf4! The other advantage of this move is that White gets the bishop to the critical diagonal before Black has time to play ...Nd5. 358

15...Nd7? This is a really poor move. The knight belongs to d5. In the game I got an advantage and the way I played was not the only opportunity. Later on Black players relied on: 15...Nd5 I guess the experience there was unpleasant too, as the games dried up pretty quickly. I am not sure Black’s position is really bad. But it is definitely sad. 16.Bg3 g6 17.h4! Bg7 18.h5 g5

19.Bc4! 359

Ensuring the other bishop gets a life too. 19.0-0-0? Bxe5 20.Bxe5 f6 21.Bg3 Ba6 would be what Black is hoping for. The knight on d5 is a prototypical good knight, while the bishop on g3 has very limited scope. 19...Ba6 20.Bb3 Bxe5 21.Bxe5 f6 22.Bg3 Black is very close to equality, but it is a defensive set-up. In the one game played from this position, White got everything you would ever desire. 22...e6 23.Kd2 Bc8 24.Rae1 Kf7 25.Bd6 Bd7 26.g4 h6 27.Rhf1 Rac8 28.Re2 Kg7

29.f4!? 29.Bxd5 cxd5 30.f4 was a more accurate move order, but it had no influence on the game. 29...gxf4 29...Rce8! was a better defence. 30.Bxd5 cxd5 31.Rxf4 Rhg8 32.b3 Kf7 33.Ref2 f5 34.gxf5 exf5

360

35.Rxf5†! Not a very difficult sacrifice to commit to. 35...Bxf5 36.Rxf5† Ke6 37.Re5† Kd7 38.Rxd5 Kc6 39.Rf5 Rce8 40.d5† Kb7 41.Rf7† Ka6 Black is trying the last trick and he gets away with it. 42.Be7?? 42.Rf6 Kb7 43.Kd3 and White would have won easily.

42...Rg2† 43.Kd3 Rxe7! 44.Rxe7 Rd2†! 45.Ke3 Re2†

361

46.Kxe2 ½–½ Gasanov – O. Ivanov, Alushta 2009. 15...Ba6!?N is an attempt to equalize immediately by exchanging White’s good bishop.

The key tactical point is that after 16.Nxc6?! Black plays 16...Bxf1 17.Rxf1 Nd5 18.Bg3

362

18...f6! 19.0-0-0 Kd7 20.Nb8† Kc8 and we get a complex endgame where Black has two minor pieces for a rook and by no means worse chances. So for this reason White should play 16.Bxa6 and develop his pieces to see if there is anything to play for in the endgame. Compared to the game Black will get the knight to d5 quickly, so all the Bc7 ideas do not work. 16.Nc4!? First of all, this avoids the exchange of the knights. It also is threatening Bc7, although deeper analysis shows that it is an empty threat. I was tempted by 16.d5, but I did not like 16...Nxe5 17.Bxe5 f6! 18.Bg3 (18.dxc6 with equality would be more pragmatic, but then there would be no reason to play like this at all.) 18...cxd5 19.Bb5† Kf7 20.c6 Bc8 and the compensation for the pawn is at most enough to make a draw. 16...Ba6 Sergey takes my threat seriously. Analysis shows that Black could have played 16...e6 without being in any specific danger. Probably White should castle queenside. The key point is that 17.Bc7?! does not particularly work. 17...Ba6! 18.Nxa5?! Bxf1 19.Kxf1

363

19...Ra7! 20.Bb6 Nxb6 21.Nxc6 Rxa4 22.Ke2 (22.Rxa4 Nxa4 23.Ke2 gives Black more options. For example: 23...Nxc5!? when White has no dreams left of winning the game.) 22...Rxa1 23.Rxa1 Nc8 24.Ra8 Kd7 25.Ne5† Kc7 26.Nxf7 Rg8 27.Ra6 Be7 28.Rxe6 Rf8 29.Ne5 Rf6

And although I would think White has a fair share of practical chances in this endgame, I would not rate him as the favourite. 16...g6?! on the other hand is better for White on account of: 17.0-0-0 Bg7

364

18.g3! and Black will be very passive and under a lot of pressure. For example: 18...Ba6 19.Nb6 Nxb6 20.cxb6 Bb7 21.Bg2 with a big advantage for White.

17.Nb6! I had seen that this would give me a pleasant advantage, but I spent (wasted) most of my time on this move contemplating 17.d5 cxd5 18.Nb6 Ra7 (18...Bxf1 19.Nxa8 Bxg2 20.c6! which leads to winning complications for White was one of the ideas I spotted – which, like King Solomon’s seven hundred wives of royal birth and three hundred concubines, led me astray) 19.Nxd5 Bxf1 20.Rxf1 Nxc5 21.Nc7† Kd7 22.Nb5 Rb7 23.Ke2

365

White has compensation for the pawn, but the whole thing is the wild feverish thoughts of a fantasist off his medication for way too long. Stockfish says 0.00, while most humans will probably say “Wha?” I had other lines I was trying, but the whole thing was awfully impractical and should be shown as a warning, not an example to follow. White has full compensation and I do not think he can be worse. But it also lacks practicality. Although it was tempting to go for something forcing when ahead in development, it should be noted that White is also not ready for action. One thing that was very clear to me was that all the black weaknesses are on the dark squares, so there is no point to allowing Black to take on c4 with the bishop. The advantage of the two bishops has to be more than just an abstract idea. Just like being a grandmaster is about the ability to play good moves and not just a diploma to hang on the wall. So after 17.d5 had lost its attraction on me, I used elimination to play 17.Nb6. 17...Ra7 17...Nxb6?? 18.cxb6 Bb7 19.d5! would see Black blown away. 18.Bxa6 Rxa6 19.Nc4! Of course the intention was never to exchange this wonderful knight for the passive black knight.

366

19...Ra7! Black cannot allow Bc7. 19...e6 This is the critical move to consider, but since Karjakin probably was unhappy with keeping the rook on a6 anyway, the move he made in the game was easy to choose without calculating the following complicated variation: 20.Bc7 Be7 21.Bxa5 This is a critical move but not very practical-looking. Please notice that White can play something like 21.0-0-0 and retain all the pleasures of the position. The f4-f5 idea is still there and a5 is still weak. 21...e5!

367

I do not think either of us took this in at the time, but the computer points out that it is far less obvious than a first glance will tell you. However, all the obvious moves lead us in the right direction. 22.0-0-0 exd4 23.Rhe1 Nxc5 24.Rxe7† Kxe7 25.Bxb4 Ke6! This is the finesse that can be hard to see in advance. Black does not lose two minor pieces for a rook, but wins the exchange. However, this is a poor moment to stop calculating. 26.Bxc5 Kd5 27.Bxd4 Kxc4

White has a pawn for the exchange and no direct threat. But if we stop and look, we can see that he has a few more good moves to make before we evaluate the position. 28.Kc2! Rb8

368

28...f6 is probably best met with the simple 29.Be3, when b3 and Rd7 are on the agenda. White’s winning chances are considerable. 29.Be5! Rb7 30.b3† Kc5 31.Bxg7

White has an obvious advantage, but if it wins with accurate play is hard to say. What is obvious is that Black is going to suffer a lot defending this endgame and that counterplay is hard to come by. For example: 31...Rab6?? 32.Bf8† and Black is mated. I did not take 19...g6 seriously during the game. Forced by Jacob to look at it now, I quickly noticed that 20.Be5! leaves Black with no pleasant reply. 20.0-0-0 e6

369

The way I looked at this position during the game was that if Black is allowed to finish his development undisturbed, he will be able to hold pretty easily. For this reason I decided that I had to take action immediately. This is a very logical way of looking at the position, as one of White’s main assets is his lead in development. 21.Be5 The key idea here is to try to open the position by advancing the f-pawn. There is however a different way to look at things. The extra time can also be spent to increase the strategic reach. Stockfish kindly tells us that White could play 21.g4!? with the idea g4-g5, which would prevent the knight from ever reaching d5 and would give White a significant space advantage (to learn more about space advantage, please go to Positional Decision Making in Chess, where I think we explained how to exploit one not too badly). 21...Nf6 This is therefore the only move. 22.g5 Nd5 23.Bb8 It is not at all clear if White should push the rook to the 8th rank first, but it feels like he is in control if he does. 23...Ra8 24.Bg3 Be7 25.h4

370

White has an obvious space advantage and all the weaknesses in the position are in Black’s camp. If this was better than what I played in the game is for the reader to decide. Only, I beg you to form your own opinion based on a plan for White from this point and not the numeric value promoted by Stockfish. One argument, from Sam Shankland, is that White may be better, but breaking through without a clear pawn lever will be difficult. Another point comes from the young Indian GM Raunak Sadhwani. If Black were able to play both ...h5 and ...g6, he would seemingly have achieved a fortress. But he is struggling to achieve it. Against ...h5 White has g5-g6 ideas, and against ...g6 he has Be5 with h4-h5 coming. 21...Nf6 If Black played 21...Nxe5 22.Nxe5 Rc7, I would have stuck to my plan and tried to open the position with f4-f5 and putting the knight back to c4 at some point. White is better. 22.Bb8 I was not attracted to exchanging on f6. It would eliminate the strong black knight, but I preferred to retain the access to the e5-square. 22...Rb7 23.Bg3 Ra7 24.Bb8 Rb7 25.Be5 Ra7 Over the last four moves I repeated the position to gain more time and not end in too bad time trouble after burning a lot of it on move 17. 26.f4 371

It is time for the plan. 26...Nd5 This is the natural move. Nothing else really works. 26...Ne4?! 27.Rhe1 f5 28.g4 would give White a big advantage after something like 28...Kf7 29.gxf5 exf5 30.d5!, as Black cannot play 28...Nf2?! 29.gxf5 Nxd1 30.Kxd1, when Black’s position collapses. The knight is his only decent piece so he should hang on to it. After the game I also analysed 26...Be7, when White has a lot of sensible ways to prepare f4-f5, as well as to play it immediately. White can take on f6, play Rhg1 and g4&f5, play f5 immediately and finally, play 27.Rhf1, when after something like 27...0-0 28.f5 exf5

372

White has the luxury to choose between 29.Rxf5 and 29.Bxf6!? Bxf6 30.Rxf5 in both cases with a serious advantage. 27.Rhe1 A very natural move. 27...f6 This of course weakens the e6-pawn, but waiting was also deeply unpleasant. Against 27...Kd7 I had planned 28.Bb8 Rb7 29.Bd6 f6 30.Bxf8 Rxf8 31.b3 Ra7 32.g3 with a very nice position. But now the computer tells me that 28.f5! would have been possible. 28...exf5 29.Bb8 Ra8 30.Ne5† Kc8 31.Nxf7 Rg8 32.Bd6 is definitely progress for White. 28.Bd6 Kd7

373

This is a big moment in the game. I had to decide on either playing for a structural advantage (which is what I did), or to use my lead in development to do something now. It is hard to say which is the best strategy. 29.g3 I am not very happy with the counterplay he got in the game, but the alternative did not seem more convincing at the time. But afterwards I had the chance to go deeper. 29.Nb6†! Nxb6 30.cxb6

a) Black can sacrifice the exchange with:

374

30...Bxd6!? I am not sure this would have been the choice of Karjakin. As we have later seen, the “Russian Minister of Defence” is quite comfortable taking on seemingly painful positions, as long as he feels he can hold them. 31.bxa7 Bxf4† 32.Kc2 Ra8 I rejected this idea because I felt Black has compensation here. Black has two pawns for the exchange and White has no passed pawn or anything else of importance, so Black has compensation. But what the computer has taught us is that we should continue the sequence of forced moves till the end. The position has not stabilized as Black still has to eliminate the apawn, which will take a move. In the meanwhile White has energetic moves that change the structure and nature of the position. Notice that Black does not have time to take the a7-pawn before ten moves later. This focus on not stopping too early in a dynamic variation and giving such evaluations as “compensation” is one of the most important things we have learned from the engines. Which is natural, as creating long lines is what they are designed to do. 33.Re4! Bd6 34.Rde1 e5 34...Rxa7 35.Rxe6 Be5 36.R6xe5 fxe5 37.Rxe5 gives White a highly favourable rook endgame. This could be Black’s best chance, which is not a good sign. 35.Rh4 h6 36.Rg4 g5 37.Rd1 Ke6 38.dxe5 fxe5 38...Bxe5 39.Rc4 and White wins.

39.h4 Be7 40.Rc4 c5 41.hxg5 hxg5 42.Kd3 Rxa7 43.Ke4 Bd6 44.Rc2 White’s position should be technically winning. b) The alternative is passive defence. 30...Rb7 31.Bc5 Bd6 32.g3 h5 33.Kc2! After a lot of analysis it becomes clear that this move is the most promising. 33...h4 34.Rd3 hxg3 35.hxg3 Rh5

375

After 36.Kb3 followed by Kc4 White is in control, but he still has to find a way to break through. 29...Bxd6 Karjakin did not fancy just sitting there as White plays b3. Instead he tries to create counterplay via the b4-square. 30.Nxd6 b3! 31.Re2 Rb8 32.Rde1

32...Rb4! 32...Nb4 This leads to a long forced line I had more or less under control. 376

33.Kd2! Nc2 34.Rd1 Nxd4 35.Re4 Nc2

36.Nc4 Threatening Kc3†. 36...Nb4 37.Rde1! e5 37...Re8 38.Nd6 Re7 39.Nf5! and White wins. 38.fxe5 f5 39.e6†! An important move. 39.Rh4 Ke6! would be fine for Black. 39...Ke7 40.Rf4 g6 41.g4 Black is in a lot of trouble. 33.Rxe6 I also considered 33.f5, but after 33...Rxa4 34.fxe6† Kd8! Black has no problems. 33...Rxa4 34.Kb1 Ra8! I completely missed this move, including the rook in the game. 34...Rxd4 35.Re8 would give White a winning advantage.

377

35.Ne8? This felt natural. 35.Nb5?? would be a horrible blunder. 35...cxb5 36.Rd6† Kc7 37.Rxd5 looks good for White until you spot 37...Rd8! where nothing works for him. 38.Rxd8 (38.Re7† Kc8 39.Rxd8† Kxd8 40.Rxg7 Rxd4 41.Kc1 Rc4† 42.Kd1 a4 is also totally winning for Black.) 38...Kxd8 39.Rd1 Rc4 The material is equal but White cannot make a move at all. Black will play ...b4 and then advance the a-pawn with devastating effect. The best try was: 35.Nf5! White is hoping to make a mating net. The idea is Ne7 followed by Rd6† and Rxc6†.

378

a) 35...g6? 36.Ne7! transposes to the next variation with the slight difference that Black has played ...g6, which makes no sense at all. 36...Rxd4 37.Rd6† Kc7 38.Rxc6† Kb7 39.Rd6 and White wins. b) 35...Kc7 I believed this was the best move, trying to save a tempo. After the game I found this very difficult line: 36.Ne7!!

Now we have two main lines, of which one is rather beautiful: b1) 36...Rxd4?! The refutation is not easy to see in advance. 37.Rxc6† Kb7 38.Rd6 Re8 39.c6†! Ka7 40.c7 Rxe7 and now in order to avoid back-rank mate, White has to find 41.c8=N†!, which wins. (Instead, 41.Rc1? fails to 41...Nc3†!! 42.Rxc3 Re1† 43.Rc1 Rxc1† 44.Kxc1 Rc4† and it is Black who 379

wins.) b2) 36...Nb4 37.R1e3 Rd8 38.Nf5 g6 39.Nd6 Nd5 40.Rxb3 Rb4 41.Rxb4 axb4 42.Nc4 and the extra pawn will win the game. c) 35...Rd8! We found this defensive idea only the second time we checked the game. 36.Nxg7 (36.Ne7 Nxe7 37.Rxe7† Kc8 would leave Black no worse.) 36...Kc7 37.R1e4 Rb4

38.f5 Black has some compensation for the pawn. White still has the better chances, although he is far from winning. 35...Rxd4 36.Rd6† After 36.Nxg7, I did not find a good response to 36...Rc4, but Black is also OK after: 36...Rd2!? 37.Rd6† Kc7! 38.Ne6† Kb7 39.Rd7† Kc8 40.Rxh7

380

On account of 40...Ne3!! 41.g4 a4 42.g5 a3 43.bxa3 Nc2 and it is time for White to give perpetual check with the rook in order to avoid getting into trouble. 36...Kc8 37.Rxc6† Kb7 38.Rce6

38...Rd3? This is a losing mistake. The saving move was: 38...Rd2! I was nervous about this during the game and have not managed to find anything for White to 381

even try in the analysis. The coordination of the white pieces has disappeared. Before we were playing for an attack, whereas now it is unclear what White is trying to do. 39.c6† Ka7! The game is wildly unclear. One of the lines we looked at ended in a very likely result: 40.Rc1 The key defensive idea for Black is very nice. 40...Rxh2 41.c7 Rc2! 42.Rxc2 bxc2† 43.Kxc2

43...Nxc7! 44.Nxc7 Rc8 45.Re7 Kb6 46.Rxg7 Rxc7† 47.Rxc7 Kxc7 With a drawn pawn ending. 39.c6† Ka7 40.c7 a4 Karjakin is desperately looking for anything to do. 40...Rd2 was not working anymore. The extra tempo is fatal in the pawn ending: 41.Rc1 Rc2 42.Rxc2 bxc2† 43.Kxc2 Rc8 44.Kb3 Nxc7 (44...Nb6 45.Nd6!) 45.Re7 Kb6 46.Nxc7 Rxc7 47.Rxc7 Kxc7 48.Kc4

382

It is easy to see visually that White is winning. Here is a possible variation to back it up: 48...Kb6 49.g4 g5 50.f5 Kc6 51.b3 h6 52.h3 Zugzwang. 52...Kb6 53.Kd5 Kb5 54.Ke6 Kb4 55.Kxf6 Kxb3 56.Kg6 a4 57.f6 a3 58.f7 a2

59.f8=Q a1=Q 60.Qxh6 White wins. 41.Rc1 a3 He could have tried a last trick with 41...Nb6, tempting White to take on b6, when although White is clearly better, it is not easy to break through. But during the game I was planning to play 42.Nxg7 Rd2, when I can play 43.Nf5 or the even simpler: 43.Rxb6 Kxb6 44.c8=Q Rxc8 45.Rxc8 Rxh2 46.Ne8! and the endgame will win for White. There are still three pawns on the board. 383

42.bxa3 42.c8=Q?? a2† would not happen. 42...Nc3† 43.Rxc3 Rxc3

44.Kb2 44.Nd6?! Rxc7 45.Nb5† Kb7 46.Nxc7 Kxc7 47.Kb2 would be close to winning too, but also a terrible mishandling of the advantage. 44...Rxe8 The rook endgames arising after 44...Rc5 45.Nd6 would also be hopeless for Black. 45.Kxc3 1–0

384

Chapter 9 Stalemate

With Kramnik, and Aruna and Akhil Anand

Contents Let’s start on the sidelines Spooked by stalemate Desperate save Don’t stalemate your king when your pieces are free to be useless 385

Frustrated by stalemate Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

How did Black react to the check? (see page 200)

386

Balashov gave me a draw. What could he have played? (see page 204)

What is Black’s strongest defensive resource? (see page 210)

387

What does the last-minute escape plan look like? (see page 212)

How can White play for a win here? (see page 217)

388

White has a surprising way to exploit his advantage (see page 221)

How can White win this queen ending? This is a tough one! (see page 223)

389

White is about to queen. How can Black defend? (see page 224)

There are a few things that only appear in endgames and as a result of late-night drinking by a fraternity of study composers. (Except in the case of my great countryman Yochanan Afek, who very rarely touches alcohol and if so, only a small glass to be polite.) Namely zugzwang, fortresses and stalemate. The weirdest of these is of course stalemate, where one side wins in every sense of the word, except for being able to deliver checkmate. Stalemate is quite common in kids’ tournaments, but rarely something that happens in top-level games. The typical stalemate scenarios, such as king and pawn against king, and king, bishop and rook’s pawn against king, are not played out, and would not be something you would mention to your friends as a stalemate. “Real” stalemates in a way where they would be interesting to bring up in polite conversation involve a new scenario in some way. Something surprising and fun, at least during the game. It also includes a sense of inevitability. Either the player allows it, or he must backtrack aggressively, which is usually what happens. I had five interesting encounters with stalemate in my games, which I hope the reader will also find interesting. They cover various scenarios, such as being the spectator, desperate saves (mainly) and problems in the conversion. Or to say it in other words, I have been on all three sides of the table.

Let’s start on the sidelines The earliest stalemate in competitive play came at move 27, while the earliest in this book is at move 390

39. I sat next to the following game at the World Rapid Championship and was absolutely astonished by the finish. Even though I was focusing on my game with Fedoseev, I could not look away. Luckily, I managed to win a good game anyway. Pavel Ponkratov – Etienne Bacrot Berlin (rapid) 2015

30.Nxd6! Up to this point the game had not been anything remarkable, but with this move White set the board ablaze. 30.Qe2 would also have forced Black to find a few only moves, but they were far less spectacular: 30...Bxe4 31.Qxe4 Rh1† 32.Kg2 Rxb1 33.Qxb1 Nxf3! and the queen ending will end in a draw. 30...Bxc2 30...Qxd6? 31.Qe2 would lose a piece. For example: 31...Bh3† 32.Kg1 Re8 33.Re1 Nxf3† 34.Qxf3 Rxe1† 35.Bxe1 Qxa6 and White has the necessary 36.Qxh3. 31.Bxe5†

391

When you see a player make all the right moves, later confirmed by the computer, everything looks obvious. But Jacob has given the position to promising Indian juniors and their suggestions have broadened our understanding of the position and increased the impression Bacrot’s solution leaves behind. 31...Kh7!! The only move. 31...f6? 32.Rb7 is not a variation you need to calculate. But it is important to see that 31...Kg8 32.Rb7 is hopeless, as the queen cannot go to c6 on account of Rb8†. After 32...Qd8 White wins if he is patient. Simple and strong is 33.Bc4 Rh7

392

34.a5! and Black is entirely without counterplay. The real alternative is: 31...Kf8 The intention to play like Bacrot did in the game fails to a nice difference: 32.Rb7 Qc6 32...Qd8 33.Rxf7† Kg8 34.Rb7 is devastating.

33.Rb8†! This is the problem. If White takes on f7, we are back in the game. 33...Ke7 34.Bb7! Rh1†

393

34...Qxa4 35.Rxh8 and the threat of a check on e8 is deeply problematic for Black. 35.Kg2 Qxa4

36.Bd5!! Only this quiet move wins. 36.Kxh1 Qh4† 37.Kg2 Qxg5† 38.Bg3 Qd2† allows Black to escape with a perpetual check. 36.Re8† looks strong, but Black has a series of accurate moves that solves all his problems. 36...Qxe8 37.Nxe8 Re1!

38.Bg3 Re2† 39.Kf1 Bd3! 40.Nd6 Re6† 41.Kf2 Rxd6 and the draw is imminent. 36.Nc8†? Ke6 37.Nb6 was suggested by a Grandmaster (with the idea 37...Qa1 38.Re8† Kf5 39.Bc8† Kxg5 40.f4† Kh6 41.Rh8#). 394

But what he missed was that after 37...Rg1†! 38.Kxg1 Qa1† 39.Kf2 Kxe5! the black king is not in any danger at all. 36...Bf5 36...Rb1 37.Ra8! and the tempo decides. 37.Nxf5† gxf5

38.Bf6† Kd6 39.Rd8† Kc7 40.Kxh1 White wins on points. 32.Rb7 There are other moves, but none that people would play.

395

32...Qc6 33.Rxf7† Kg8 34.Bc4

White looks absolutely winning. His pieces are fabulously placed, and his next check will hurt badly. So Black is desperate. However, Bacrot showed the only path to salvation. 34...Bd3†!! A diversion to gain time. 35.Bxd3 Qd5 36.Bc4 Rh1† 36...Qd1†? 37.Kf2 Qd2† is not a perpetual, as White has 38.Be2!. And now either 38...Rh1 39.Re7! or 38...Rh7 39.Rf6 will be the sad end for Black. 37.Kg2

396

37...Qd2†! 38.Kxh1 Qg2† Draw agreed. ½–½ 39.Kxg2

Spooked by stalemate The first game of mine in this chapter did not end with a stalemate. But it was decisive for the final stages of the game and could have been added to the history books as one of the few cases where 397

someone managed to win a lost position with a stalemate trick! A funny and entirely unrelated story: A decade ago, I woke up from surgery. The first thing I saw as I slowly came to, was the doctor. He introduced himself, told me that everything had gone well and then asked if I remembered him. I said “Of course! In 1982 we played a game in the Najdorf.” Already this was surprising to him, but then I told him all the moves till the end of the game. He was totally in shock. Now, I must confess that this is probably the only game of mine from the early 1980s that I remember, for a reason I cannot guess. Back to the game. Boris Gelfand – Yuri Balashov Minsk 1986

44.Bd2! Good prophylaxis, trying to get the bishop to g5. I do not remember if I saw 44.Qg5 Kh7 45.Bb4 Nf4! 46.Qxe7† Kh6 and White has no advantage at all. I should note that Alex recently went through some of my games and found that also 45...Qb1† 46.Kh2 Qc2 holds the balance. 44...f4! A good active try for counterplay. 44...Nh4 looks active, but White delivers checkmate: 45.Qxe6† Kh8 46.Qf6† Kg8 47.Bf7†! The key move to see. 47...Kh7 48.Qh6# 398

45.Bg4 Nf8 46.Qxf4? This is too cautious. During the game I remember rejecting the following line because of a stalemate trick. 46.Bxe6† Nxe6 47.Qxe6†

At this point there are two variations worth analysing. 47...Kh8 This looks poor, but the winning line is not as simple as one might expect. Certainly, there are many lines after which White is winning in the long term. But the direct, undisputed line requires some accuracy. 48.Qxe7! Qg6† 49.Kh2 Qh5† 50.Kg2 Qg4† 51.Kf1 Qd1†

399

52.Be1 Changing the position, preparing for another round of checks. 52...Qd3† 53.Kg1 Qg6† 54.Kh2 Qh5† 55.Kg2 Qg4† 56.Kf1 f3! This is of course what we should be afraid of. The win here is simple enough, but you would have to trust it from afar. 57.Qf8† Kh7 58.Qf7† Kh8 59.Bc3! Qg2† 60.Ke1 Qg1† 61.Kd2 Qxf2† 62.Kd3 Qe2† 63.Kd4

The white king has escaped the checks, but at a cost. Black now has a strong passed pawn of his own. To allow this requires nerves of aluminium. 63...f2 64.e6! f1=Q 65.Kd5†! The discovered check with the king decides.

400

More challenging was 47...Kf8. The strongest move is 48.Bb4!, which is what I wanted to play. I rejected it because of: 48...Qb1† 49.Kh2 Qh7† 50.Kg2

50...Qg6†!! 51.Qxg6 f3† with stalemate to come. I will not spoil it for the reader and instead ask you to find the mistake in this variation. 46...Qb1† 47.Kh2 47.Bc1!? was the last chance to fight for an advantage, but maybe not much of a chance. 47...Qxb2 48.Qe3 Alex was not overly optimistic about White’s chances after 48.Qg5† Neg6 49.Bxe6† Nxe6 50.Qxg6† Ng7 51.Qd6 Qb3. Indeed, it is hard to see how he would break through. 48...Neg6 49.Qc3 Qa2 50.Bh5 Qd5 51.f4

401

51...b4! 52.axb4 On 52.Qxb4 my opponent was definitely planning to play 52...Nxe5 with equality. But a funny line also leading to a draw was: 52...Nh4!? 53.Kg3 Qg2† 54.Kxh4 Qh2† 55.Kg5 Qg3† 56.Bg4 Nh7† 57.Kh5 Qh2† 58.Kg6 Nf8† and White cannot escape the perpetual, on account of 59.Kf6?? Qh4#. 52...a3 53.Bf3 Qd8

54.Be3?? Draw agreed??

402

½–½ I think I realized that I no longer had any winning chances and made a random move. My opponent, happy to have escaped, accepted it uncritically. The correct 54.Be1!, preventing the queen coming to h4, followed by 54...a2 55.Qa3 would have led to an immediate draw. So, my escape was rather lucky. After: 54...a2 55.Qa3 (55.Bf2 loses to 55...Nxf4 56.Qa3 Qd2) 55...Qh4† 56.Kg2 Qe1!

Black wins because of the double threat of queening the pawn and taking the bishop. The only way to avoid both is merely a short delay: 57.Bd4 Nxf4† 58.Kh2 Qh4† 59.Kg1 Ne2† and Black wins, as he does in a dozen other ways... Years later I ran into Balashov and he reminded me that I had avoided his clear trick. But I probably fell for it, seeing the danger, but not accurately understanding that it was just a trick and not some great defence that could not be breached.

Desperate save The next game was played towards the end of the Grand Prix series. I needed 2½/3 to qualify for the Candidates, as did my opponent. In the opening I played a novelty that should lead by force to perpetual check after what he played. But because of the tournament situation, I tried to look for something else and as a result ended up in a pretty desperate situation. I was suffering the whole game and after the time control the position is lost, although less lost than it had been earlier. My opponent would have to show some technique to win the game.

403

This game shows various defensive ideas in a lost complex position and how you should not give up before you have tried absolutely everything. Dmitry Jakovenko – Boris Gelfand Khanty-Mansiysk 2015

41.Kh1! A very clever move chosen after quite a long think. White avoids a lot of counterplay and has a winning position. 41.Qe6? would allow 41...Qg5† with perpetual check. The careless 41.b4? would allow counterplay: 41...g6! 42.Qe4 Qg5† White can go further over the edge with 43.Kh2? Nf6 44.f4

404

44...Neg4†! 45.Kg1 Qh4 and it is Black who wins. 41...Qh6 41...g6 42.Qe6 is already deadly. After 42...Kg7 43.Qg8† Kh6 44.Qh8† Kg5 45.Qxh4† Kxh4 46.a4 it is easy for White to win. One of the points of White’s last move was that after 41...g5 he can play 42.Kg2! and with no g5square and no control over the f6-square, Black cannot easily create counterplay. The following line: 42...Qd4 43.Be4 Kg7 44.Qxg5† Kf7 45.Qe7† Kg8

46.Bh7†! Kh8 47.Bc2! Qd5† 48.Kh2 Qf7 49.Qxf7 Nxf7 50.a4 Nxd6 405

51.b4 when White is winning was given by Miroschnichenko. Still some accuracy is needed, but it is also difficult for Black to resist. 51...Nb8 52.Bd3 Nc6 53.b5 Na5 54.Kg3 Ndb7 55.b6 Nd6

56.Ba6 Ne4† 57.Kh4 Kg7 58.b7 Nc6 59.Bc4 Nb8 60.a5 and White wins. 42.f4 g6 43.Qe6 Qf8!

406

Trying hard to offer resistance. 44.Bb3? 44.fxe5? Qf1† 45.Kh2 Qf2† 46.Bg2 Qf4† 47.Kg1 Qe3† 48.Kf1 Qd3† would not allow the white king to escape. But let’s look deeper at the most critical options. 44.a4 This is what I was afraid of. But deep analysis shows that Black is holding on with the nail tip of his pinkie! 44...Nd3! The critical move. Only after deep analysis can we determine that this indeed holds. After 44...Qc8 45.Kg2 Qc2† 46.Kg3 the white king runs away.

407

45.Qxd7 45.Qf7 Qc8! gives Black sufficient counterplay. 45...Qxf4 46.Qc8† 46.Qe8† Kg7 47.Qe7† Kh6 48.Bg2 Qc1† 49.Kh2 Qf4† 50.Kg1 Qf2† 51.Kh1 Nf4 52.Qe4

52...Nxg2 53.Qxg2 Qe1† 54.Kh2 Qe5† 55.Kg1 Qxd6 56.Qc2 Qg3† 57.Kh1 g5 is also a draw, I think. 46...Kh7 47.Bg2 Qxd6

408

48.a5 This is the computer’s main choice, which shows how difficult it is for White to convert his advantage. Deep analysis shows that 48.Qb7†! Kh6 49.Qf3 with a clear advantage is stronger, but the lines are not clear-cut here either. 48...Nf2† 49.Kg1

49...Qd1†! 50.Kxf2 50.Bf1 Qd4 51.Qc3 Nxh3† 52.Kh2

409

52...Qg1† 53.Kxh3 Qxf1† is of course dangerous in practice, but also a tablebase draw. 50...Qd2† 51.Kf3 Qd1† 52.Kf4 Qd2† 53.Ke5 Qxa5† 54.Bd5 Qe1† 55.Be4 Qg3† 56.Kd4 Qf2† 57.Kc3 Qe1† 58.Kd4 Qf2† 59.Kc4

59...Qxb2 This position is objectively a draw. We know from the tablebases that with the king on f3 or g3, White would win. But not with the king on g4. Chess is mysterious sometimes. The key point is of course that the exchange of queens would lead to a draw, unless it could be followed immediately by h4-h5, pinning the g-pawn and securing the transformation of the white pawn away from the flank. The only way White would win is if the king were on g3 and queen on e6, forcing the black

410

queen to passive defence on g7. Then the white king would march to b6 somehow. Look it up in Lomonosov... Instead the engine comes up with an unbelievable prophylactic move: 44.h4!!

Not only does this prepare White for what follows and prevent ...g5, it also puts Black in something akin to zugzwang. It now makes sense to divide the lines into two branches. a) 44...Qd8 This loses, but not so obviously. For example: 45.b4!? The key idea comes after: 45...Qxh4† 46.Kg2 Qd8 46...Qg4† 47.Qxg4 Nxg4 48.Be6 Ngf6 49.Bxd7 Nxd7

411

50.Kf3! with a winning ending. It depends on tempos. No other move here wins. 47.fxe5 Qg5†

48.Kf3! The king runs away. Besides saying that the king hides in front of his own pawns, there is no explanation or understanding connected with the following variation; it is a spacebar exercise in digital accuracy, but it feels more appropriate to include it than to omit it. 48...Qh5† 49.Ke3 Qg5† 50.Kd3 Qg3† 51.Kc4 Nxe5† 52.Kb5 Qd3† 53.Kb6 Qd4† 54.Ka6 Kg7 55.d7 Qd3† 56.b5 Qa3† 57.Kb7 And White wins.

412

b) 44...Qc8 This is one of the two moves that does not lose immediately. 45.fxe5 This allows a lot of checks.

45...Qc1† Finding the way to escape these is a big challenge, unless you are gifted with digital accuracy. It is not at all certain that Jakovenko would have been able to find the win from here, had he played the first few moves of this variation. This is partly what good defence is about. Not playing the best move according to the engine, but making it as difficult as possible for your opponent to avoid mistakes. 46.Kh2 Qf4† 47.Kg2 Qd2†

413

48.Kf3! This triangulation makes it possible to bring back the bishop. 48...Qd1† 49.Kg3 Qe1† 50.Kh2 Qd2† 51.Bg2 Qf4† 52.Kg1 Qe3† 53.Kf1 Qd3† 54.Kf2 Qd4†

55.Kg3 The king is about to escape the checks, so Black has to measure his response. 55...Nxe5 55...Qe3† 56.Bf3 Nxe5 (56...Qe1† 57.Kg2 Qd2† 58.Kh3 and the checks are gone) 57.Qf6† Kh7 58.Qf4 Qe1† 59.Kg2 and White is on his path to victory. 56.Qd5 Qg4† 57.Kf2 Qxh4† 58.Ke3 Qe1† 59.Kf4 Nd7 60.Qd4† Kh7

61.Bd5 414

White has established control. It is impressive how the centralized pieces dominate the board. Still a lot of moves would have to be made after this 17-18 moves long variation to win the game... 44...g5! I played this after 25 minutes of thinking. I felt I needed to generate some form of counterplay at any cost. 45.f5

45...Nd3? I was quite surprised to find out after the game that I could have held here with far less desperate means. 45...Kg7! 46.Qe7† During the game I thought this endgame was entirely lost. 46.Kh2 Qh8 with the idea of ...Qh4 also gives Black a lot of counterplay. (I was thinking of ...Qc8 only. To put the queen on h8 seems like a typical computer move, meaning that I did not look at it at first, but once I look deeper, I can see the ideas). White does not have immediate threats after 47.Qe7† Kh6, while Black is ready for ...Nd3!? or ...Qa8!? for example, after which White would be in more risk of losing. 48.Kg3 is met with 48...Nf6! and Black is holding. 46...Qxe7 47.dxe7 Nf6

415

48.Bd5 White can try other ideas here of course, but none of them are effective. 48.a4 Nc6 followed by ...Nxe7 is fine for Black. So is 48.Kg2 Ne8 49.Ba4 Kf7 and Black is in time. 48...Ne8 49.a4 Kf6 50.a5 Kxe7 51.a6 Nc7 52.a7 Nd3 53.b3 Kf6!

There are other paths to the draw, but cleaning out the white pawns is definitely the most effective. 54.a8=Q Again, other options exist, but none are better. 54...Nxa8 55.Bxa8 Kxf5 56.Kh2 Kf4 416

Black will make a draw. The white king is too passive and there are so many scenarios where Black makes a draw. For example, against a king march to e2 (if even achievable), ...Nc1†xb3 leads to an endgame with the wrong bishop. 46.Qxd7 Qa8† 47.Kg1 47.Kh2?? Nf4 is not happening of course. Although White still escapes with a draw after 48.Bd5!. 47...Nf4 48.Kf2

48...g4! 417

This is the last chance. Of course, it is a rather lame trick, but on occasion even these work in toplevel events. 48...Qg2† This is the most obvious try – but the white king runs away. It is not that easy to see, but once on the board, Jakovenko would have found his way for sure. 49.Ke3 Qe2† White also runs away after 49...Qg1† 50.Ke4 Qg2† 51.Kd4 Qf2† 52.Kc4. 49...Qxb2 50.Qc8† Kg7 51.Qc7† Kh6 52.d7 shows the extent to which White is able to ignore the black checks. 50.Kd4

50...Qxb2† Another typical line is this: 50...Qd2† 51.Kc4 Qd5† 52.Kb4 Qd4† 53.Kb5 Qe5† 54.Ka6 Qe2† 55.Kb7 Qe4† 56.Qc6 and wins. There are quite a lot of ways to escape the checks here. The most human to me seems to avoid letting the knight come in. 51.Kc4 Qe2† 52.Kc5 Qe3† 53.Kb5 Qd3† 54.Kb6 Qd4† 55.Ka6 Qd3† 56.Kb7 A typical zigzag. White wins.

418

49.hxg4?? I assume he was thinking he was winning, when taking this pawn. Maybe thinking I am smart enough for a double bluff? It is not always nice to ask your opponent after the game what they missed, so I will leave this to the reader’s imagination. 49.Ke3 Qf3† 50.Kd4 resembles the note to 48...Qxg2 (and the game without any stalemate possibilities). 49.Kg3 Qf3† 50.Kh4 Qf2† 51.Kg5 Nxh3† 52.Kg6 Nf4† 53.Kf7

This line brings the king to safety, after which it’s all over. After 53...Qh4 the computer gives me 419

weird quick ways to win, but certainly any human would choose: 54.Qe8† Kh7 55.Qg8† Kh6 56.Qh8† Kg5 57.Qxh4† Kxh4 58.d7 and it’s over. 49...Qg2† 50.Ke3

50...Nd5†! 51.Kd4 51.Bxd5 Qd2† 52.Kf3 Qf2† 53.Ke4 Qe3† 54.Kxe3 is also stalemate. 51...Qf2† 52.Kxd5 Qd4† 53.Kxd4

A nice save. If I had found 45...Kg7 and saved the game, no one would remember this game. Now it will be remembered by everyone. 420

½–½

Don’t stalemate your king when your pieces are free to be useless The following game was from the Melody Amber tournament (see Dynamic Decision Making in Chess page 70), against my good friend Lev Aronian (see Positional Decision Making in Chess page 178). In the game I made the mistake of allowing my opponent to install a passed pawn on f7, protected by another on g6, stalemating my king on h8, which was prevented from moving by the pawn on g7. The main issue with this was not only that I was playing for a win, but also that it should have been impossible to get rid of all my pieces. But I was saved when my opponent did not see the winning continuation. Levon Aronian – Boris Gelfand Nice (blindfold) 2008

1.d4 d5 2.c4 c6 3.cxd5 cxd5 4.Nc3 Nf6 5.Bf4 Nc6 6.Nf3 a6 7.Rc1 Nh5 8.Bd2 Nf6 9.e3 Bg4 10.h3 Bxf3 11.Qxf3 e6 12.Qg3 Nh5 13.Qf3 Nf6 14.Bd3 Bd6 15.g4 Rc8 16.g5 Nd7 17.Qg2 Nb6 18.f4 Nb4 19.Bb1 Nc4 20.a3 Nc6 21.0-0 h6 22.g6 f5 23.Be1 Qf6 24.e4 Ne3 25.e5 Nxg2 26.exf6 Ne3 27.Rf2 Nxd4

Who says the Exchange Slav is boring? In these rapid blindfold games, it does not make sense to show off your serious preparation. Rather the players play something generic and try to make the games exciting along the way. As you can see, Lev and I succeeded in this case.

421

28.Bd2? John Nunn demonstrated that the best way to play for White was the following forcing line: 28.fxg7 Rg8 29.Nxd5! Rxc1 30.Nf6† Kd8 31.Nxg8 Rxe1† 32.Kh2 Nd5

33.Nxh6 Nf6 34.g8=Q† Nxg8 35.Nxg8 Bf8 36.Nf6 Rxb1 37.Rd2 Ke7 38.Ng8† Ke8 39.Nf6† with a draw. 28...Nc4 29.Nxd5 exd5 30.Bc3 Bc5 31.Kh1

31...0-0? Too soon! It is rare you castle at move 31 and find it was premature! In a later volume we shall see me castle even later in the game... 422

31...gxf6! 32.Rd1 Ne3 33.Re1 Ne6 was totally winning. 32.f7† Kh8 33.Bxd4 Bxd4 34.Re2 At the time I did not think so, but the f7-pawn gives White sufficient compensation for the piece.

34...Bxb2? This was my intention. I had wrongly assumed that the pawns on the queenside would be very strong and I was on the way to winning the game. At the same time Aronian believed he was winning (as he often does) and was getting excited. The fact that he was winning is a bit surreal to me even today, but the analysis shows this to be the case convincingly. After a calm move like 34...Rcd8, the position remains deeply unclear. 35.Rxc4 dxc4 36.Rxb2 c3 According to my notes John Nunn called 36...b5 37.Bxf5 Rcd8

423

38.Rc2 equal. I blame my note-taking more than the good Doctor! Black is without counterplay and White will win slowly but surely. 37.Re2 b5 38.Bxf5 Rcd8

We have reached a phase of the game where White must win it. My judgement of the position was catastrophically wrong. Meaning that the phases of the game seen in hindsight are not the same as the players believed during the game. 39.Bc2? 39.Re3! would have been a shortcut. The c-pawn just goes. 424

39...a5 40.Kg1 a4!? A desperate attempt to create counterplay with two split passed pawns that the bishop cannot control. 40...b4 41.axb4 axb4

42.Kg2 leaves Black lost. White will manoeuvre the king to the queenside to look after the pawns, after which the bishop and rook combined will win the game with Ba4 and Re8 followed by Rx(tothe-left) and Be8. 41.Kf1 b4 42.axb4 a3 43.Bb3 Rb8 Even at this point White is winning, although there would have been a lot of nice tricks, suitable for blindfold games. However, there are two fortresses that White must avoid. The first is well-known, while the second is a clever expansion on it that probably has never been considered before, but came up more than once in my analysis for this game. However, as it did not appear in the main lines, I do not want to have a collection of random-looking moves on the page to justify its inclusion.

425

These two fortresses are rather generic of course, the first being a version of the famous fortress Sam Shankland resigned against Anish Giri. The second is a version of this that came up in some of our analysis. A slightly more advanced position which arose from our analysis is the following:

426

Black is dead lost, but he can try a last trick: 1...h4†!? The key idea is simple. After 2.Kxh4? Black escapes with 2...Rxf7! 3.gxf7 g5† 4.Kxg5 Kg7

And very soon he will play ...c2, ...Kxf7 and hide in the corner. Instead White can win with both 2.Kg4 and: 2.Kf4 This allows Black to lose in the most spectacular way. 2...Rxf7† Quiet play continues after 2...Rd8 3.Bc2 Rf8 4.Ba4!, when Black is in zugzwang. The threat of

427

Ba4-e8 is deadly. We could stop here, but why not include a final zugzwang: 4...c2 5.Bxc2 Rc8 6.Ba4 Rf8 7.Be8 and the game is over.

With the h-pawns on the board, there is no fortress... 3.gxf7 g5† 4.Kf5 Kg7 5.Ke6 Kf8

6.Kf6! Black is in zugzwang. He may be the only one to get a queen, but it does not help him one bit. 6...g4 7.hxg4 c2 8.Bxc2 h3 9.g5 h2 10.g6 h1=Q 11.g7# Let’s return to the game.

428

For some reason Lev was not able to see the winning continuation. 44.Ke1? In order to win, White must start with a few accurate moves. 44.Ra2! Rxb4 45.Rxa3 Rxf4† 46.Ke2 After this White just must avoid some small tricks in order to convert the advantage. 46...Rd4

47.Ra6! This is the key move, transferring the rook to the ideal square on c6. 429

47.Ke3?! is the only other winning move, but not after 47...Rh4 48.Be6? (48.Bc2 somehow still wins) 48...Rh5! 49.Kf4 Rc5

50.Ra1 c2 51.Rc1 h5 52.Bb3 Rc6 and White will not be able to break through. 47...Rd2† 48.Ke3 Rb2 49.Ba4! Rh2

50.Ra8! Not strictly necessary, but good looking! 50...Rxa8 51.Be8 Rxh3† 52.Kf4 Rh4† 53.Kg3 And that’s that. 44...Rxb4 430

45.Re8 There are many other options, but they all lead to the same outcome. 45...Rb8 46.Rxf8† Rxf8 47.Kd1 Ra8 48.Bd5 Rd8 49.Kc2 h5 50.Bc6 Threatening Be8. 50...Rf8 51.Kxc3

51...a2 52.Kb2 Rb8†! It is important to get the rook outside the pawns. And the check means that there is also time to 431

push the h-pawn. After 52...h4? 53.Be8 Black is in zugzwang, not stalemated. 53.Ka1 h4!

54.Bb7 Aronian offered a draw. Black will give “perpetual rook”, which can only be avoided by taking the rook and stalemating Black. ½–½

Frustrated by stalemate The final game of this chapter has a stalemate as a pivotal part of the theme in the way my opponent played, but it is about much more than this. Mainly I see it as a mini-course in converting a material and positional advantage, when your king is open to harassment. This is by no means easy and I was not successful during the game, despite it having been adjourned, giving me plenty of time to analyse it together with my seconds before the resumption of play. There is also a psychological dimension to understanding what happened. I was winning for most of the game, although I seem to remember the position being close to equal at the start. When you have been winning for most of the game you can start to feel as if the game is already won and you are defending your win, rather than converting your advantage. You then start to look for ways to finish off your opponent that are bordering on the risk averse, which can be detrimental. This sort of emotional bias is common in chess, as it is a game played by humans. It is important for us to reflect on these experiences as well when we analyse our games, so that we can improve not just our skill, but also our mental approach. 432

Boris Gelfand – Vladimir Kramnik Sanghi Nagar (6) 1994

One of the biggest achievements in my early career was to win a Candidates match against my good friend Vladimir Kramnik by 4½–3½. Kramnik won in great style in the third game and I struck back in the fourth. At the time this game was played, the match was thus level and it was therefore of great importance that I convert my advantage at the resumption of the game, which was adjourned at this point (for our younger readers: before computers became really strong, we had to play for six hours, take a rest and continue the game the next day...). 61.Kh3 This was the sealed move. 61...Qf6! The only way to keep the game alive. To some extent Kramnik and I viewed where Black’s chances were in the same way. I was uncomfortable with the prospect of a queen ending, on account of the almost never-ending checks one must reckon with. Actually, to such an extent that it should be considered a liability. I had failed to win two advantageous queen endings against Adams and Beliavsky the same year and was negatively biased beyond reason. I overcame this irrationality later and won an interesting queen ending against Hodgson (see Decision Making in Major Piece Endings page 13).

433

I have earlier annotated this game for My Most Memorable Games for Olms and described there how initial optimism deteriorated in the analysis with Alex and Valery Atlas. Finally, we did not manage to construct a winning plan and by the time I wrote that book, I did not have a clear one either. But time is a friend of the studious. In the preparation for this book, Alex had looked at the game again and Jacob and I also analysed it closely, looking for new ideas, finding not one winning plan, but several. I should underline that simply pressing the spacebar will certainly give a line the engine considers winning, but would not be something you would be able to rely on in an adjournment. We need a plan. For example, Stockfish 10 recommends 62.Qe3 as the best move. It is certainly not a bad move, but what is the follow-up? Moving around may preserve the advantage, but it will not convert it. I am sure that a plan can be constructed with this move, and many others, but it also must make sense to me for me to think it was how I should have played. 61...Kh7 allows 62.Qb6 with a transition into a winning rook endgame. With active rooks and an extra pawn, as well as the black king being cut off, I do not think I have to justify this with variation overload... 62...Qxb6 63.axb6 Rb7

64.Rc7! is the key point. The black king is obviously misplaced on h7. A big part of the story of this game is that I had the position after 61...Kh7 in the game. Two moves earlier:

434

Like a good boy I played 60.Kh2 Kg8 before sealing 61.Kh3, repeating the position to ensure that I could carefully check the position at home. The bizarre thing is that I had seen 60.Qb6! and evaluated it as totally winning. Kramnik had also seen it and was desperate not to allow the chance again. This was one of the times where following this standard advice was ill advised. This is the nature of chess. There are never any short-cuts that can allow you to stop thinking. Quickly upon resumption we arrived at the biggest decision point.

62.Rcxd5 After the game Kramnik expressed some anxiety about his position after 62.f4 (not a move I 435

wanted to make voluntarily) 62...Rd6 63.Rcxd5?!, but the queen ending is not that convincing. 63...Rxd5 64.Rxd5 Rxd5 65.Qxd5

65...Qe7! 66.Qa8† Kh7 67.Qxa6 Qe4! and it is clear that things have moved in the wrong direction, even if White still is winning with accurate play. You also have to remember under which circumstances this accuracy has to be displayed. Alex checked 62.Rexd5!? again and it is worth going deeply into the lines, as they are truly fascinating.

a) First of all, Black can play 62...Qe7 or some other waiting move, avoiding the queen ending Alex was aiming for. I think this would be a fairly practical approach.

436

b) After 62...Qg5, which is that type of practical waiting move, the win is not obvious.

63.Rxd7! Rxd7 64.Rc8†! 64.g4? Rd8! and the game still has to be won. I allowed the engine to run for a bit (in other words, had lunch randomly at this point) and was not as impressed with the moves it was suggesting as with its evaluation of the position. White is clearly not in control at this point. And now the computer tells us that Black has to enter into the hopeless queen endgame, on account of: 64...Kh7 Again and again in what follows, we shall see how vulnerable the king is on h7. 65.g4 The reason is that after the natural reply: 65...Qf4 White has a stunning mating combination.

437

66.Qc3!! Here there is no proper defence against Qc6-a8, checkmating the black king and keeping control of f3 at the same time. 66...Rxd4 Black is threatening to deliver checkmate with ...Rd2. But White has a winning defence. 67.Qc6!

With the double threat of Qa8 and Qg6†!! killing it. c) Alex had revisited our old analysis on: 62...Rxd5 63.Rxd5 438

Black’s choice is if he should take on d5 first or not. Back in the mid-1990s we gave: c1) 63...Qg5 Here Alex found the idea of pushing the d-pawn rapidly. 64.Rxd8† Qxd8 65.d5 Qxa5 66.d6!

This is the end of Alex’s notes. To be helpful to the reader, I shall fill in the blanks, which are not at all obvious. Take the f5-pawn or not? c11) 66...Qxf5† This is obviously the human test. 439

67.g4 Qc8 Preventing the queen from getting to c7 immediately. 68.Qd5 Qd7

But White has a different way to take control of the d7-square. 69.Qa8† Kh7 70.Qe4† Kg8 71.Qe7 Qc6 72.Kg3 Qc1 A desperate attempt for counterplay.

73.Qe8†! Why this move is important is easy to see in the variation 73.d7? Qg1† 74.Kf4 Qd4† and Black makes an immediate draw. 73...Kh7 74.d7 440

Evaluating this position by calculation is probably impossible. To do it by experience however is not hard at all. With Qe4 being so powerful in so many variations, it is hard to believe that Black would have a perpetual here, and no surprise, he does not. 74...Qg1† 75.Kf4 Qc1† 76.Kf5 Qc5† 77.Qe5 Qc2† 78.Qe4 Qc7 79.Qd5 Qc2† 80.Ke5 Qc7†

81.Kd4 White wins. c12) 66...Qd8! This was suggested by Stockfish and is far from stupid.

The idea is to run with the a-pawn. The following variation is beautiful and can be found after deep thinking, but it is not something that would come naturally to me. 441

67.Qd5 Threatening d6-d7. 67.Qb4 Qd7 68.Kg2 is another computer win, which no human would ever suggest. 67...Qd7 Things are more difficult with the f5-pawn on the board, as there is no check on e4. And giving it away again does little, as Black will get the g7-square for the king. But there is another surprising plan. 68.g4 a5

69.f4!! The only winning plan. 69...a4 70.g5 The idea becomes clear. Mate is threatened. 70...Qe8 70...hxg5 71.fxg5 Kh7 72.g6† Kh6 73.Kg4! and mate is near. The text move is a final, desperate attempt. White has a good deal of winning lines here, but I personally prefer that Black cannot run. 71.Kg4! Qe2†

442

72.Qf3 Qxf3† 73.Kxf3 Kf8 74.d7 Ke7 75.f6† One of the pawns will queen. But the comfort of Alex’s great winning line is destroyed if we change the move order a bit. c2) 63...Rxd5 64.Qxd5 Qg5!

This is a serious defensive try that confirmed my reluctance to go a queen ending. Similar moves targeting the f7-pawn could lead by force to this position, should Black wish it. The more human alternative will set White up for disappointment. c21) 65.Qa8† Kh7 66.Qe4 f6 67.Qd3 443

67.g4 Qd2 68.d5 Qd1 and White has no winning chances left. 67...Qxh5† 68.Kg2

68...Qe8! An excellent triangulation. Black must keep the queen active. How instructive this is cannot be overstated. 69.d5 Qb8 70.d6 Qb2† The dangers posed by this type of ending are well illustrated by the following line: 71.Kh3 The alternative line is not much more comforting: 71.Kf1 Qa1† 72.Ke2 Qe5† (72...Qb2† 73.Qd2) 73.Kd1 Qa1† 74.Kc2 Qa2† 75.Kc3 Qxa5† 76.Kb3 Qb6† 77.Kc2 Qf2† and White will not escape the checks. 71...Qa1 72.d7 White could still return with the king of course, but the outcome is the same. 72...Qh1† 73.Kg4

444

73...Qc1!! And White needs to make a draw urgently, to avoid the mate coming after 74...h5†!. 74.f4? And this is not it. 74...Qh1! White is mated on the next move: 75.d8=Q h5# c22) 65.Qe5 Qxh5† 66.Kg2 Qg5 Practically, this seems best. 67.Qe8† Kh7 68.Qxf7 Qd2† 69.Kh3 Qxd4 70.Qg6† Kh8 71.Qxa6

445

71...Qd5! Domination and centralization at the same time. So White has to go for multiple exchanges: 72.Qc8† Kh7 73.Qc2 Qxf3 74.f6† Kh8 75.a6 Qh1† 76.Kg4 h5† 77.Kg5 Qd5† 78.Kg6 Qg8 79.Qc6 gxf6† 80.Kxf6 Qg7†

White wins in 61 moves after best play. I guess about 55 of these will be choosing the right square for the king after a check. The wrong square is likely to set you back 10-20 moves... For those of us who prefer chess to snakes and ladders, this is not a serious winning strategy. Jacob and I came up with a different idea. What if White first consolidated on the kingside? 62.g4!

At this point we have looked at two options as well. 446

a) 62...Kh7 This is strongly met by taking control of the c-line. 63.Qc2 63.Qb6 Qg5 64.Qxa6 Qd2 is the kind of counterplay I was seeking to avoid. 63...Kg8 64.Rc6 Rd6 65.Rc7 Kh7 66.Ree7

This is the type of set-up White is looking for. He can now move in before striking. 66...Kg8 67.Qc5 Rf8 68.Kg3 Rfd8 Black can do nothing but wait, so White chooses the best moment. 69.f4 Rf8 70.Kh3 Rfd8

447

71.Rxf7 Qxf7 72.Rxf7 Kxf7 73.g5 White’s attack is irresistible. Next comes Qc2 and f6, for example. This line is indicative of what White is dreaming about, but keeping Black restrained is not that easy. Especially without the help of an engine that has calculated all cunning tries to the end. b) 62...Qd6 63.Kg2 63.Qb6 does not work when the king is defending the f7-pawn. 63...Qf6 64.Rcxd5 Qc6!? Trying to play like the game. 64...Rxd5 65.Rxd5 Rf8 is the engine’s choice, but White is of course in complete control here.

After something like 66.Qe3, White is winning. Conversion will take time, but this is one of the most attractive versions I have seen of this ending.

448

65.Rxd7! Qxd7 66.Qb6! A dominating move. 66...Kh7 The most resilient. 66...f6 is the set-up Black would like, but at this point White would play 67.Qe6†, ending the game immediately. It is every schoolboy’s wish for 66...Qxd4 to be the main line, as Black is mated with:

67.Re8† Kh7 68.Qg6† fxg6 69.fxg6# 67.f6! 67.Re4 would be less direct, with the idea of taking on a6. But after 67...f6 I would argue that 449

Black has made more progress. 67...Qxd4 An admission of guilt if there ever was one. But at this juncture it is hard to see a way for Black to avoid the rook ending. That it wins for White is no surprise, but it does require a bit of technique, so it is worth looking at it until the end. 68.Qxd4 Rxd4 69.Re7! Kg8 70.Re8† Kh7 71.fxg7 Kxg7 72.Ra8 Rd6 73.Kg3 Re6 74.Ra7 Rd6 75.Kf4 Kg8 76.Ke5 Rc6

77.f4!? 77.Rd7 Kg7 78.Rd6 is a more cautious approach, but at some point White should commit to some sort of action and the main line is pretty easy to work out. I may be wrong, but it seems to me that White does not need a brain to convert the advantage here. 77...Rc5† The only critical move. Passive play is not available, nor is it something that would discourage White from playing like this: 77...Kf8 78.Kd5 Re6 79.f5 Rf6 80.Rc7 is a basic illustration. 78.Kf6 Rxa5 78...Rc6† 79.Ke7 Re6† 80.Kd7 Kg7 81.f5 and the a-pawn falls. 79.Ra8† Kh7

450

80.g5! Other moves win as well, but there is something about forcing mate in a rook ending. This variation may feel comforting and just, but – and there are two big buts – this was found only after prolonged work, with engines. And more importantly, people more imaginative than us will most likely be able to find other ways of making the job hard for White. 62...Qc6! The following line is not as critical as more or less everything else in the annotations to the game. 62...Qg5 63.Rxd7 Rxd7 63...Qxh5† 64.Kg2 Rxd7 65.Re8† Kh7 66.g4 Qg5 67.Qb8 g6 68.Qe5 and White wins. 64.f6 Qxf6 65.Re8† Kh7 66.Qc2† g6

451

67.hxg6†! Originally I had given 67.Qe4 as winning, but we simply missed that after 67...Rxd4 68.hxg6† Kg7! White has no advantage at all. I doubt I would have made such a blunder at the board... 67...fxg6 68.Qe4 White is winning. Either in a rook endgame after 68...Qf5† or at once after 68...Rxd4? 69.Qb7†, when he wins the queen. These analyses are extremely deep of course, and back then we struggled to find even a quarter of all of this, working without computers and with limited time. But part of my problem was that I overestimated how winning my position is. As discussed above, I wanted to win it without allowing even a hint of counterplay, which turns out to be unrealistic. As we have seen, it is necessary to allow some counterplay and then show precision. I was determined to keep the pawns on f3 and g3 so as not to allow any unnecessary checks. This was mainly a psychological decision and in retrospect not a good one. But once you are looking with the computer, things that were previously more or less impossible become easy. 63.Rxd7 After 63.Rc5 Qf6 White has nothing better than to put a rook on d5. 63...Qxd7

452

64.Qb6? Back in the 1990s I analysed 64.Qd3!?, but looking deeply with the engine now, I have found that White has an even more accurate continuation with: 64.Qc4! Qd6 65.f4 Rb8! Black has to look for active options, as White would love to have the d-pawn dominate the position with something like: 65...Qd7 66.d5 Kh7 67.Qc6 Qa7 68.d6 Qg1 69.Qd5 and Black is dead lost. 66.Rc5 Qd7

67.Kg4! This is the safe square for the king. The key point about the construction of the white set-up is 453

that the black queen is not managing to get active. 67...Rb5 Designed to prevent Rc7. 68.Re5 The moves here are somewhat random, but show possible play. 68...Kh7 69.Qd3 Rb8 70.Re7 Qd5 71.Qe4 Qb3

72.f6† Kg8 73.fxg7 Kxg7 74.Ra7 Qd1† 75.Kh4 And in the (not so) long run Black is dead. 64...Kh7

454

65.g4?! This allows Black to get off the hook a little too easy, but it was already too late to win the game against accurate defence. 65.Qxa6 Qxd4 would give White different ways to create problems for Black: a) 66.Qe2

66...Qa1! The queen is perfectly placed here behind the a-pawn and ready to harass the white king. 67.Re7 67.a6 is quickly dealt with: 67...Rd4 68.f4 Ra4 69.f6!? The only try. 69...gxf6 70.Qd3† Kg7 71.Re7

455

The threat of Qg6† is impressive, but Black has a defence: 71...Qh1† 72.Kg4 f5†!! 73.Qxf5 Qd1† 74.Kh4 Qh1† 75.Qh3 Qxh3† 76.Kxh3 Rxa6 and after minimal discomfort, Black will make the draw. 67...Rd4 68.Qe1 But not 68.Rxf7?? Qh1† 69.Qh2 Qxf3 and it is Black who wins. 68...Qa2 The queen remains an irritation for White. Understanding what the active black queen is able to do here is key to understanding the position. 69.Qc3 Ra4

70.f6! 456

The last try. Black has to display accurate defence: 70...Qb1 71.Re1 Qf5† 72.g4 Qxf6 73.Qd3† The rook endgame is a draw, just. 73.Qxf6 gxf6 74.Re7 Kg7 75.Ra7 f5! and the g4-pawn will be easy to attack. 73...g6 74.hxg6† Qxg6

75.Qxg6† In practical play, White would perhaps try 75.Qd5!?, when the advantage is there and very playable, with chances for Black to make mistakes. You should always allow the opponent to dream of problems of his own, and not just try to force your will on him. 75...Kxg6 76.Re5 This rook endgame might look promising, but the mechanics of it will quickly be revealed. 76...Rf4 Just waiting. 77.Kg3 Ra4 78.Rf5

457

This is the set-up White would love to call stable. If the king can be shuttled to the queenside, the game wins by itself. But this cannot be achieved. 78...Kg7! Probably the only move Black can get wrong at this point. Or more precisely, Black may be cautious about entering this position, because he would feel in zugzwang, having to play this move. 78...Ra3? 79.Kf4 would allow the king to cross the barrier. And 78...Rb4? 79.Rf4! is immediately over. 79.Kf2 Black has three ‘only moves’ that save the game. Any time-wasting would lose. But for a grandmaster these moves are very intuitive. 79...Ra3! 80.Ke2 Kg6!

458

Again, it is close to zugzwang, but the king can return to g6 at the right time. After 81.Kd2 h5! Black draws. b) 66.Re7 Qd5!

This powerful centralization of the queen is seen in many lines. 67.Kg4 Rb8 The same with the activation of the rook. I am sure the reader will by this point have a feeling for Black’s counterplay. 68.Qe2 Rb4† 69.f4 Ra4 70.a6 Ra3 71.a7

459

Black is almost in zugzwang. He cannot allow Qf3 or Rxf7 and 71...f6?? is refuted by 72.Rxg7†! and the rook on a3 is hanging. But Black can still move the king and there is no way for White to break through. Jacob spent some time analysing 65.Kg2!? with the idea of slowly improving this position. But he was unable to find a win there as well. Black can play slowly with 65...f6, but after something like 66.Re2 Qd5 67.Re7, White is definitely playing for a win. The real problems arise if Black decides to enter the rook ending: 65...Qxd4! 66.Qxd4 Rxd4 67.Re7 Rd5

At this juncture White has two tries that look rather alike.

460

a) 68.Rxf7 Kg8 69.Ra7 Rxa5 70.g4 This looks promising, but after 70...Ra4! we have not been able to establish a way to play for an advantage. The key position arises after: 71.Kf2 Kf8 72.Ke3

72...a5! The only move. The key point is that the only waiting move not to allow the white king to cross the fourth rank 72...Kg8?, loses to 73.f4! with the idea g4-g5. The forced line goes: 73...Ra3† 74.Ke4 Rg3 75.Rxa6 Rxg4 76.Rg6! Rh4 77.f6 Rxh5 78.fxg7 Ra5 79.f5 and once White plays f5-f6, Black is paralysed, waiting for Rh6-h8 to finish the game. 73.Ra6!? 73.f4 does not work now. 73...Ra3† 74.Ke4 Rg3 and Black holds. 73...Kf7 74.f4 Ra3† 75.Ke4 Rg3 76.Rg6 a4 77.f6 gxf6 78.Kf5

461

78...Rb3! Keeping the rook active. The g4-pawn is no longer a target. 79.Rxf6† Kg7 80.Rg6† Kh7 81.Ra6 Rb5† 82.Kf6 Rb4 83.Ra7† Kg8 There is no meaningful way for White to make progress. b) 68.f6!?

This looks awfully clever. Forced play continues like this: 68...Rxh5 69.Rxf7 Kg6! The only move. 70.Ra7 Rxa5 71.fxg7 Kh7 72.f4

462

The only way for White to make progress. 72...Rf5 73.Kh3 Kg8 74.Kg4 Rf7 75.Rxa6 Kxg7 White has an extra pawn, but with limited material, Black can hold the draw with accurate defence. I held this ending twice, against Kramnik and Karpov (see page 118 in Decision Making in Major Piece Endings).

65...f6! And just like that, all of White’s hopes are gone. There was no chance that Black would go for a bad rook endgame with: 65...Qxd4? 66.Qxd4 Rxd4, where White wins efficiently with: 67.Re7 Rd5 68.f6! Kg8 69.Re8† Kh7 463

70.Rf8! Rd7 71.fxg7 Kxg7 72.Ra8 and we reach the endgame covered above. 66.Re4 66.Re2 Rc8! 67.Qxa6 Qd5 is described in the next note.

66...Qc8! Kramnik thought for 40 minutes at this point. His brilliant defensive idea was not planned and was found at the board. Today the computer can show us that Black also survives after: 66...Qd5 67.Kg3 464

67.Qxa6 Ra8 68.Qe6 Qxa5 gives White two extra pawns, but no winning prospects. His king is too open. The typical line looks like this: 69.Qf7 Qc3 70.Qg6† Kh8 71.Re8† Rxe8 72.Qxe8† Kh7 73.Qe4 Qd2 74.d5 Qd1 and White cannot make progress. 67...Rc8!

68.Qe6 Qb7 Black is ready to create counterplay against the white king again. The two extra pawns offer White no advantage. 67.Re7 Qc1!

Kramnik’s brilliant idea. 465

68.d5 Maybe I should have allowed Kramnik to show his idea in all its glory with: 68.Qxd8 Qh1† 69.Kg3

69...Qh2† 70.Kxh2 Stalemate. 68.Kg2 Qd2† 69.Kg3 also does not work. After 69...Rxd4 70.Qc7 Black saves himself with: 70...Qg5 and White has no advantage. Trying to avoid the stalemate by giving up the rook first with 68.Rxg7† Kxg7 69.Qxd8 almost works. But after: 69...Qh1† 70.Kg3 Qe1† 71.Kf4 Qc1† 72.Ke4 Qc6†!

466

The white king has failed in its endeavour to run to safety. 68...Qf4 68...Rc8 was also possible. The game could have ended 69.d6 Rc2 70.Kg3 Qd2 and White’s only move is 71.Rxg7†! with a perpetual. 69.Kg2 Rc8 70.Qxa6 Rc3 71.Qe2 Qc1 72.Qf2 Qd1!

73.Re1 Qxd5 74.Ra1 Rd3 75.Ra2 Rxf3 76.Qxf3 ½–½ 467

A disappointing result. After this the seventh round was drawn too and in the 8th round Kramnik made an ill-advised pawn sacrifice in the opening (he blundered a pawn on move five, but I know you will not believe this is possible for such a great player, so I am playing it safe). I punished this in the traditional way, by taking the money and running. I slowly exchanged lots of pieces, while keeping my advantage and eventually won a bishop endgame. There is a funny story connected to this game. There was a riot because of local problems and Alex was panicking that we would have to abandon the game and replay it from the beginning later, which would be a novel way to lose an extra pawn. Luckily for me, the police calmed down the protesters. With this chapter discussed and analysed, I was able to take the defensive ideas so fresh in my mind to a game against one of the bright new stars. This game deserves a chapter on its own, which naturally follows this one...

468

Chapter 10 Stalemated

India 2020 – Spot the future champions!

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

469

White could have refuted the piece sacrifice here. How? (see page 239)

How can White increase his advantage? (see page 242)

470

What is the next step? (see page 245)

Black has a strong defensive resource here. What is it? (see page 250)

471

How should White exploit the situation? (see page 254)

How can Black hold? (see page 256)

472

What is the next move? (see page 258)

How did Black save the game? (see page 260)

Vidit Gujrathi – Boris Gelfand

473

Poikovsky 2018 This game was played against a young rising Indian star in the 19th edition of the tournament held yearly in celebration of Anatoly Karpov. It is amazing that they have managed to keep a tournament going in the middle of Siberia for so long, and can in this way be compared to the Wijk aan Zee tournament, which is held in a Dutch seaside village and has an even greater tradition. This was the first time I played in Poikovsky. It was one of the strongest fields yet in the tournament and I was very excited to play, especially with a great mix of young and experienced players. In the end I shared 2nd-3rd place with Nepomniachtchi, half a point behind Jakovenko. This was half a point ahead of Vidit, so this game can be said to have been crucial for my good finish. Vidit is a strong player as you will see in the game. He has a classical style and good technique, like his friend and I think sometimes analytical partner Anish Giri. Not many players add to opening theory in a consistent way, but Vidit is one of them. 1.e4 c5 2.Nf3 Nc6 3.Bb5 g6 4.Bxc6 bxc6 This is how I prefer to play this position. 4...dxc6 is of course perfectly playable, and can also be considered safer. 5.0-0 Bg7 6.Re1 Nh6 Most people became aware of this move when Boris Spassky played it in his match against Fischer in 1992. However, Dautov had already played it a year earlier. 7.c3 0-0

474

8.d4 Strangely, this move is the less popular option. After 8.h3 I played a lot of games with 8...d5 and 8...f5, with a lot of dynamic play. Even 8...c4 is interesting, intending to sacrifice the pawn for dynamics. For example, 9.Qa4 Rb8 10.Qxc4 d5 with big complications in Oparin – Dubov, Moscow 2018. Stockfish, who does not know about psychological pressure and does not make tactical mistakes, thinks White is better. In practice things are less obvious. Dubov won the game. 8...cxd4 9.cxd4 d5 10.e5 f6

11.Qc2!? This was Vidit’s innovation for this game. It was an interesting way to pose new problems. The idea is prophylactic, preventing Black from taking control of the e5-square, as is the main plan. I had some previous experience in this line: 11.exf6 exf6 12.Nbd2 Re8 13.Nb3 Nf7

475

14.Nc5?! This move allows Black to seize the initiative, forcing White to find some defensive moves. 14.h3 was more prudent, when the chances are probably balanced. 14...Bg4! Stockfish running for long enough pretends that the position is equal. It would be more prudent to say that accurate play would have allowed White to keep the balance. 15.Rxe8† Qxe8 16.h3 16.Be3!? 16...Bxf3 17.Qxf3 Qe1† 18.Kh2 Ng5 19.Qe3 19.Qg4 Qxf2 20.Bxg5 fxg5 21.Qe6† Kh8 22.Qxc6 would have kept the balance, something that is not easy to verify during the game. For example: 22...Rg8 23.Ne6 Bxd4 24.Nxd4 Qxd4 25.Rd1! gives a lot of counterplay – or a perpetual. To both see this and trust it is not without difficulty. 19...Re8 20.Bd2

476

20...Qe2! 21.Qxe2 Rxe2 Black was better in Inarkiev – Gelfand, Magas (rapid) 2016.

11...Bg4 The key idea behind Vidit’s novelty comes after 11...fxe5 12.Bxh6! (12.Nxe5 Ng4! would work out for Black. The key line is 13.Qxc6 Bxe5! 14.dxe5 e6 with the threats ...Qh4 and ...Nxf2. The rook on a8 is not hanging, as White has no time to take it. White would thus be relegated to a desperate and probably unsuccessful defence.) 12...Bxh6 13.Nxe5

477

This position would favour White. Black has a lot of weaknesses and no clear scope for the bishops. Probably the knight duo will outshine the bishop pair here. But 11...Qb6 was also possible. I also considered 11...Bf5!?. In the end you can choose only one move. 12.Nbd2 Qb6 13.h3

At this point I decided to sharpen the game, rather than to enter into a passive defence. 13...fxe5!? 478

Changing the character of the game entirely. It was perfectly possible to play 13...Bxf3 14.Nxf3 Nf5, when I felt Black was slightly worse. 14.hxg4 Nxg4

15.Nc4!? This is a clever move, but I was not concerned. I mainly considered: 15.Qb3 I was planning to play: 15...exd4 One of the things I looked at for Black was to play 15...e4, but I felt that we would reach the following endgame by force: 16.Nxe4 Qxb3 17.axb3 dxe4 18.Rxe4 Nh6

479

19.Bxh6 Bxh6 20.Rxe7 Rfb8 21.Rexa7 Rxa7 22.Rxa7 Rxb3 23.Rc7 Rxb2 24.Rxc6

I believed that Black should be able to make a draw with accurate defence, but had no intention to go for it. Who wants to end up in such a position straight from the opening? You are only playing for a draw and there is no guarantee that you would make it. And this is relying on there being nothing you have overlooked. Actually, it turns out that White has another tempting option in the above variation. I thought that Black would have enough counterplay after 19.Rxe7 Nf5 20.Rexa7 Rxa7 21.Rxa7 Rd8.

480

But apparently this is not the end of the line. White can play 22.Be3!, when taking with the knight on d4 loses to an elementary tactic, and the endgame after 22...Bxd4 23.Nxd4 Nxd4 24.Rc7 is very unpleasant as well. 16.Rxe7 d3 17.Qxb6 axb6 I was not sure about this position at all. It is not so easy for White to get his pieces into the game. If he manages, Black would certainly be in trouble, but first he would have to achieve this. Computer analysis suggests that Black has reasonable compensation after: 18.a4 c5 19.Rb7

19...Ra6! 19...Rae8?! looks active, but after 20.Nf1 White is ready for a4-a5 with an advantage. 481

It is quite possible that there is some way for White to pose problems for Black, but he would have to find it first. And I did not think it was that simple. And in later analysis with engine assistance, it also does not come easily. But it is one thing to prove at home that there is an advantage, it is another to prove it at the board. I am not sure that a human alone or the engine alone can prove an advantage, but combined it is likely. After the game the computer showed a very counter-intuitive idea: 15.Qd3!! The concept is that White wants Black to play ...e4 and there is not really anything else he can do.

15...Rae8 looks active, but after 16.Qe2! Black has nothing else to do. After 16...e4 17.Nxe4 dxe4 18.Qxe4 Nf6 19.Qe6† Kh8 there is no reason to make it difficult (20.Ng5!?) in order to understand the position. After 20.Re2 White has a clear advantage without complications and we can consider Black’s sacrifice refuted. But again, this was found by the computer. The players were nowhere close to this during the game. 15.Ne4!? was suggested by Sam Shankland, as was the fantastic counter-reaction: 15...exd4! 16.Nc5 e5! 17.Nd7 Qc7 18.Nxf8 Rxf8

482

And with ...e4 coming, Black has considerable counterplay. The position is entirely unclear. 15...dxc4 The following was pretty forced: 16.Qxc4† Kh8 17.Qe6 Rf5 18.Re4 Nf6 The computer’s suggestion 18...Nxf2 makes no sense. There are no pieces to create an attack. 19.Rxe5

483

19...Rxf3 I felt no attraction to 19...c5 20.dxc5 Qxe6 21.Rxe6, which is simply a pawn down. 20.gxf3 Qxd4 I spent a lot of time on 20...Nd5 21.Re2 Bxd4, but after 22.Bg5! I was not happy. White gets his pieces into the game and is clearly better.

I liked my position at this point, thinking I should be able to hold pretty easily. 21.Be3 It is hard to see this move as being wrong. White develops his pieces. At the same time, it is hard to ignore Stockfish, which is insisting that 21.Re1!? is the more accurate move. It all relies on tactical play: 21...Rf8 22.Qxe7 Nd5

484

23.Be3! Qxb2 23...Nxe3 24.Rxe3 Qxb2 25.Rd1 is totally hopeless. After 25...Qxa2 26.Rd7 Qa1† 27.Re1 Qf6

28.Re6! Qxe7 29.Rexe7 White will take the pawns and then improve the king. This is nowhere near a fortress. Even if the rooks are exchanged after the inevitable fall of the queenside pawns, White will win by advancing the first f-pawn to be exchanged and then the second. 24.Qe6 This endgame is apparently close to finished. Black cannot improve his position and after concrete play with: 24...Nxe3 25.Rab1 Qc2 26.Qxe3 Qxa2 27.Rb7 Qd5 28.Qe7 Rg8

485

29.Kg2 White is totally dominating. 21...Qxb2 22.Rc1 There were other moves, but again, this seems pretty natural. 22...Rf8 23.Rxc6 Ng4 Hitting the rook on e5 three times. 24.Re4 Nxe3 25.Rxe3 Bf6

486

At this point I was still pretty optimistic. 26.Rc8! Vidit played this almost instantly. We shall debate it below. But first we shall look at a move I was concerned about during the game: 26.Ra6 I confess to starting to develop some anxiety about my position looking at this. 26...Bh4! This is the critical move, leading straight into a sharp endgame. During the game I had a brief moment to think that maybe I should just hold the position with something like 26...Qd4 but after 27.Kg2 it is not at all easy for Black to make a move. I don’t like this at all.

27.Qe5† White has nothing else. After 27.Re2 Qb1† 28.Kg2 Qd3! Black is seriously annoying White. The general idea for Black in this type of position is to prevent his opponent from coordinating his pieces, which would allow him to make his material advantage count. 27...Qxe5 28.Rxe5 Rxf3 29.Re2

487

This endgame looks like it could be close to lost. White’s next moves are likely to be Rxa7, a3 and Ra2, preparing a fierce march up the board with the pawn. But Black has some moves to make first and it appears he is not without counterplay. The following line is not to be seen as “analysis,” but more of a pondering about how the game may continue. Black has a passed pawn of his own after all... 29...Rc3! 30.Rxa7 Rc1† 31.Kg2 Kg7 32.a4 Ra1

The most natural move to me seems to be to dominate the bishop, but Black’s counterplay comes freely: 33.Re4 33.a5 feels more critical, but also here it is possible for Black to create counterplay. 33...Bf6 488

34.a6 h5 35.Ra8 g5 36.a7 g4 Black should hold, but once again it is possible for White to go over the edge. 37.Rd2 Ra3 38.Rd7 looks critical, but after 38...h4

White cannot execute his plan, but should prevent a disaster. After 39.Rc8?? Black plays 39...h3† 40.Kf1 Ra1† 41.Ke2 h2, winning. 33...Bf6 34.a5 h5 35.a6 h4 36.Ra8 Ra3 37.Re3 Ra2 38.Rd3 g5 39.a7 g4 40.Rd7 g3 Black has sufficient counterplay to force White to look for a draw, as after 41.Rb8 gxf2

42.a8=Q?? Black wins with 42...h3† 43.Kf1 Rxa8! 44.Rxa8 h2 and one of the pawns promotes. 26...Rxc8 27.Qxc8† Kg7 My thinking during the game was that although traditionally the exchange of one pair of rooks 489

limits the counterplay Black will be able to create, there are other relevant factors in the position. After the exchange of the rooks it is not that easy for White to create threats against the black king, while at the same time the white king is finding it hard to hide, leaving it open to being annoyed endlessly. Apparently, I relaxed too much and did not fully grasp the dangers I was facing.

I think this is a good moment to discuss the abstract reasoning behind various exchanges. First of all, we saw from the variation with 26.Ra6 that the exchange of queens in some circumstances might not be horrible for Black. The presence of three rooks on the board allows him to create counterplay. However, if both the queens and a pair of rooks were to come off, Black would be dangerously close to lost. The argument in favour of 26.Rc8 would be to get the exchange of the rooks in early and be able to force an exchange of queens later, by creating threats to the black king. This is not so much a narrative of how the game “should” go, but a strategy White is pursuing. Personally, I had no intention of exchanging queens. I had a strict “catch me if you can” strategy. And after 26.Rc8 I would avoid losing my a-pawn! So, I was not unhappy at all. My opinion now of this endgame is that Black should hold after either exchange, but that it may be trickier this way around. There is a danger we always dislike most the problems we had to solve, but none the less, I stand by this evaluation. 28.Qe6 The queen takes up a dominating outpost. In some lines the queen may be better placed on a different square; but as Black is not planning to do anything, White will be able to make adjustments along the way. 28...Qc1† 490

One of the calculations I had to make was when to advance the h-pawn. In some positions it would create counterplay and in others it would desperately weaken the g6-pawn. I decided to keep it back for now and let Vidit show his hand first. This is maybe the first decision in this game I got entirely right. It is not that Black should never advance the h-pawn; but in most situations, it is likely to simply be a weakening move. Although I got this right, the defence was still very difficult and you can rightly claim that I did not manage to defend the position, but simply managed to pick-pocket half a point from my opponent on the way out of the tournament hall... 29.Kg2 Qc5 30.Re4 Placing the last piece on the optimal square. 30...a5

At this point Vidit executed the only idea he has to break the black fortress. By advancing the fpawn, he will be able to create threats against the black king and, in the course of events, force the exchange of queens. Amusingly checkmate is not the main aim, but winning the endgame is. Ulf would be proud! 31.f4! Qc2 It was not possible to avoid the advance of the f-pawn. After something like 31...Qb5 32.f5 Qxf5 (32...gxf5 33.a4 Qc5 34.Rc4 only makes things worse.) 33.Qxf5 gxf5

491

34.Re3! the a-pawn appears to be doomed. It can only be saved with 34...e5 35.Rd3 Be7 36.Rd5 Bb4, but the endgame after 37.Rxe5 Kf6 38.Re8 is obviously lost. 32.f5 g5 33.a4 The optimal reorganization of the white pieces seems to be: 33.Qc4 Qd2 34.Re6 Qb2 35.a4 Qd2 36.Qe4 Qd7 37.Rc6 Qd2 38.Rc8 Kh6 39.Qc4

But it is still not easy to see what White is going to do here if Black just waits. The king is safe enough. We let the engine run to a serious depth and it was not able to inflict damage on Black. Essentially Black holds, but of course it is a close shave.

492

33...Qd1 34.Rc4

34...h5!? As we saw above, Black can also play passively, but it is hard to sit and wait. White will shuffle his pieces endlessly, hoping to somehow encircle the a5-pawn at the right moment. I did not desire this fate and decided to play for active counterplay. I think this was a quite reasonable decision and would not blame my later mistakes on this, although I have been oscillating between thinking this was a good idea and not. Each path has its own dangers. It is possible to say that Black can go for passive or active defence in this position and both are viable, but he cannot sit between these two chairs. And maybe this is what I did. My opinion now is that it did not make a lot of sense to play actively. If I am waiting, I might suffer no more than the boredom of doing nothing. But after active play I must avoid making mistakes. It is much harder to make mistakes if you are meant to do nothing and you do nothing. Part of the equation is that I somehow forgot that advancing the h-pawn would weaken the g6square and give White additional targets. Of course, it is not hard to understand, but where our focus is, is a big part of the thinking process. I should add that time trouble was not an issue. 35.Qe4 g4? But this is certainly not the right move. It is quite natural to put the pawns on the light squares when you have a dark-squared bishop. But in this case, Black should keep the options open and both play for a fortress and counterplay. By advancing the g-pawn, it will also become a potential target later. 35...h4! was much stronger, generating necessary counterplay through the ...h4-h3† threat. It is easy to generate lines by pressing the space bar, but it is not the way humans think. Imagine that the rook 493

is on c8 and the queen on c4 and it is Black to move.

...h3!† and the draw is secured. I must admit that putting the pawns on the same colour as the bishop is so counterintuitive that I didn’t even consider it during the game.

36.Rc7? Missing a big shot. At this point White could have made Black’s life very difficult with 36.Rc5!. The key point is that 36...h4? is met with 37.Rd5! and the g4-pawn falls. White wins. 494

The defence for Black is now excruciatingly difficult and against perfect play, it should fail. Our confidence in the following rather extensive variations is not absolute. I also apologise for their depth. It is one of those rare moments where the truth takes precedence and we will look at a truck-load of engine-corroborated moves, rather than trying to understand the position by human perception only. We already talked about the weakness of the a5-pawn and the g6-square. This is the principle of two weaknesses. On the other hand, Black has active counterplay. At this point Black has two options.

36...Qd2 37.Rc8 37.Rd5 Qc3 38.Rd8 Kh6 is not progress for White. The exchange of queens is not favourable here. 37...Qd1 37...Kh6 38.Qe3† and the endgame is hopeless.

495

38.Rb8!! The quiet move is extremely difficult to find. Grandmasters to whom we gave this position to solve found prophylaxis to be an unlikely idea. The point of White’s clever prophylactic move is that 38...h4 would now be met by 39.Rb1! and the g-pawn is lost. 38.Ra8? h4! would draw. White would also not be able to win after the forcing 38.Qe6?! Qf3† 39.Kg1 Kh6!, where the only serious try is 40.Qe3† Qxe3 41.fxe3, but then Black draws with urgent counterplay.

41...h4! 42.Rc5 Bg5! 43.Kf2 h3! and Black is surviving, on account of 44.Rxa5? Bxe3†!. Therefore the more normal 44.Kg3 Bxe3 is reasonable and leading to a draw. 496

38...Kh6 38...h4 39.Rb1 h3† 40.Kg3 and White wins. 39.Qe3† Kg7 40.Qb3! This endgame is winning for White. The following variation is not completely forced, but may be close to best play and is certainly a good illustration of what White is trying to do. 40...Qxb3 41.Rxb3 Kh6 42.Rb6 Threatening to go to a6 and pin the bishop, while attacking the a-pawn. 42...Kg5 43.Rb5! Bc3

44.f6†! Without this, the draw would be imminent. 44...Kxf6 45.Rc5! 45.Rxh5 Be5 should also win, but feels a bit less accurate. 45...Bd2 46.Rxh5 Bg5 47.Rh8 Black is unable to defend both the a- and g-pawns. Once the g-pawn inevitably falls, White will win by squeezing Black slowly, while keeping the pressure on the a-pawn. Black’s other main try is: 36...Qd8 37.Qd5 Qb8

497

38.Rc4!! More prophylaxis against ...h4. The strategy for White is to combine prophylaxis with slowly manoeuvring the pieces into place. Superhuman accuracy is needed. At this point the trap is 38.Rxa5? Qb4!! defending the g4-pawn in advance. Black survives after: 39.Ra8 Kh6 40.Rg8

40...h4! and there is not enough danger against the black king or potential in the a-pawn to play for a win. For example: 41.a5 Qf4! 42.a6 h3† 43.Kf1 Qc1† 44.Ke2 h2 45.a7 h1=Q 46.Qxh1† Qxh1 47.a8=Q Qxa8 48.Rxa8 Kg5 49.Ra5 Kf4 with a draw. 38...Kf8!? I do not think anyone can find such a move, but it is the most resilient it seems. 498

38...Qb6 loses to 39.Rc8! Kh6 40.Rg8. Again we see the need of the h5-square for the black king in conflict with the job of the h5-pawn of defending the g4-pawn. 40...Bg7

41.Qe6† Simplest. 41...Qxe6 42.fxe6 Bc3 43.f3 gxf3† 44.Kxf3 Black is dominated and has too many weaknesses. White wins. 39.Qe4 Qb3 40.Rc5 Qh3† 41.Kg1 g3 42.Rc8† Kg7

43.Qg2! 43.fxg3? Qxg3† 44.Qg2 Having achieved the desired exchange of queens, White may seem to be winning, but Black somehow survives after: 44...Bd4†! 45.Kf1 Qxg2† 46.Kxg2 Kf6 47.Rc4 Ke5 48.Kf3 Kd5 49.Rc8 Bb2 50.Kg3 Bf6 51.Kf4 h4 52.Kg4 e6

499

53.fxe6 Kxe6 This endgame is a tablebase draw. 43...Qxf5 44.Qxg3† Kh6 45.Rc4 Qd5

46.Qe3† Kg6 47.Qc5 Qd3 48.Qc6 Kg7 49.Qe4 Qxe4 49...Qd6 50.Kg2 Kh6 51.Rc6 Qd2 52.Rc5 h4 53.Qf5 and White wins. 50.Rxe4 h4 This endgame wins with a bit of accuracy. The two flank pawns are too much to defend for the black bishop. 51.Re6 Kf7 52.Ra6 Bc3 53.Rc6 Bd2 54.Rc4 Bg5 55.Rc5 Bd2 56.Rh5 White wins.

500

While this win is superhuman, we should not forget that the defence is equally so. In practical play it is hard to press with White, but also not simple to defend with Black either. Jacob gave this position to some of his strongest students and, even after having failed to solve the problems themselves and being shown the moves, they found the whole thing confusing. Particularly 38.Rb8!! and 38.Rc4!! seem out of reach for the human mind.

36...Qd6? After Vidit missed a difficult win, it is my turn to miss a simple draw. During the game I saw the idea of: 36...h4! 37.Rxe7† (37.Rc4 h3† 38.Kh2 Qd6† is also a draw.) 37...Bxe7 38.Qxe7† Kg8 39.f6 Qd5†

501

With the idea of ...Qf7 or ...g3† and Black makes the draw. But I was quite unsure about it and missed some details completely. Probably it is because I believed that my position was already poor. Sadly, these things happen. 37.Ra7 After this White is winning, with some work to do. Essentially, the black position cannot be held, due to the weakness of the g6-square and the g4-pawn in many lines. 37...Qc5 38.Rd7 Qa3? 38...Qc3 was a little more resilient. I think White is winning, but it is not so easy to prove at this moment. 39.Rd3 Qc5 The c2-square would have been much better. Now White can invade with the rook without facing immediate consequences. 40.Rd8 Qc3 The end of the time control, not that it mattered greatly. We both had plenty of time and our mistakes should be blamed solely on our abilities and not be explained away.

We are now entering a phase where White has many chances to win the game, but where one job remains: to finish off his opponent. 41.Qe6?? Quite a serious mistake, totally unnecessarily allowing Black a lot of counterplay.

502

41.Qd5! This was a simple version of the same idea. Black has no defence. 41...Kh6 42.Rg8! During the game I was worried about 42.Qd2†. I intended to play 42...Qxd2 43.Rxd2 Bc3 44.Rd5 h4, but was very sceptical about my chances.

But thinking more deeply about it here and moving the pieces around the engine board, it seems we will always end with a version of the endgame with only pawns on the queenside as a worstcase scenario, where as long as the black king is not hopelessly cut off, the draw is achievable. 42...Bg7 43.Qe6† Kh7

In my later analysis I realized that White can play 44.Rd8 here and after the check on f3 there is 503

no follow-up on d1. For a moment I believed that the reason I had escaped was that we both missed this resource. But in our work for this book, this came out as total nonsense. White can move the rook to anywhere along the 8th rank and win, as long as it does not blunder the rook. From a human perspective, the most natural is therefore: 44.Rc8 Hitting the queen and keeping the rook defended. 44...Qf3† This is nothing but a few revenge checks. 45.Kg1 Qd1† 46.Kh2 Black is busted. He can only avoid mate by: 46...Qd6†

The endgame after 47.Kg2 Qxe6 48.fxe6 Kg6 49.Rc5 is trivially lost. In this line the advantage of having the rook on c8 is apparent. On d8 it prevents a check on d1. On e8 it attacks the e-pawn. Only b8 is pointless, except that it still forces the exchange of queens, and is thus good enough to win the game. It is clear that I had a near miss here. Somehow Vidit and I had mutual chess blindness, where we saw the line, but somehow did not consider it as critical as it was. Vidit for sure saw the exchange of queens, and rejected it. As we can see, without good reason. Maybe this is where he lost faith in 41.Qd5 altogether? Or did he miss that I had strong counterplay with ...h5-h4 in the game continuation? I think this is the most likely. After all, without it, he would win immediately. 41...Kh6 42.Rg8 Qf3† The big difference between putting the white queen on e6 and d5. White should not have allowed this check.

504

43.Kg1

At this point it is time for calculation; more specifically, for comparison. Black has to play ...h4, but should he give the check first or not? 43...Qd1†? It is Black’s turn to make a serious tactical mistake. It is easy to explain. When I made my move I thought the position was a dead draw, missing his 46th move entirely. I am sure you can follow my line of thinking: if it is a draw, why bother looking at the exchange of queens? Why allow extra options? But of course, this is exactly what I did. 43...h4! I eliminated this, so as not to give my opponent the option of exchanging the queens.

505

In fact, this would have set a nice trap, although there is little chance he would have fallen for it. 44.Qe3†?? I rather suspect that Vidit would have played as in the game with 44.Rg6† Kh5 45.Rg7!?, but in that case I have real counterplay with 45...h3!. Actually, White should hurry and make a draw while he still can! A small theme here is that the king is safest behind the pawns. It does not matter if this is on the 8th rank or the 4th rank. 46.Qf7† Kh4 47.Rh7† Kg5 48.Rh5† Kf4 49.Qc4† and there is either a perpetual check or an endgame that is entirely safe for both sides. 44...Qxe3 45.fxe3 Kh5 This is completely lost for White.

506

The pawns are too strong, and the bishop is able to come around and help. There are many possible lines, so I shall give just one: 46.Rg6 h3 47.Kh1 Kh4 48.Kg1 g3

Black wins. There is no defence against ...Bf6-c3-e1-f2, winning, that does not allow ...Bg5 to come with force. 44.Kh2 h4 Apparently White is not able to make progress, but my opponent found a powerful resource.

45.Rg6† Kh5 46.Rg7!! Kh6 The big difference from 43...h4 is that Black no longer holds after: 507

46...h3 47.Rg8! There is zero chance my opponent would play 47.Qf7†? Kh4, but I just wanted you to notice that White is lost suddenly.

47...Kh6 The most resilient. 47...Kh4 48.Qe3 Kh5 49.Qf4! quickly leads to a lost endgame. White wins with: 49...Qd6 50.Qxd6 exd6

51.f4!! A great move, eliminating Black’s counterplay. 51...gxf3 52.Kg3! (this is simplest, although 52.Kxh3 also wins) 52...Be5† 53.Kxf3 h2 54.Kg2 Black will soon have to let go of the a- or the h-pawn, leaving him with no chance to save the game. 508

48.Qe3† Kh7 49.Rg6 Qd6† 50.f4! Qd1 50...gxf3†? 51.Kxh3 wins. 51.Qf2

When we were analysing this, we first thought that Black was in zugzwang here, but actually Black has a waiting move and White has a threat. After 51...Kh8 White wins with 52.Qa2!. 51...Qxa4 52.Rxg4 There are now endless ways for the game to transpose into the same ending. This is one of them: 52...Qd1 53.Rg3 Qc1 54.Kxh3 Qb2 55.Qxb2 Bxb2

We checked this carefully and can say with certainty that White is winning with optimal play. And this is what it looks like: 509

56.Kg4 Kg7 57.Rb3 Bd4 58.Ra3 Bb6 59.Ra4 Kf6 60.Rc4 Kf7 61.Kg5 Bd8 62.Rc6 Kg7

63.f6†! exf6† 64.Kf5 Kf7 65.Ra6 Black is in zugzwang this time. If the king goes to e7, White takes on f6. If it goes to g7, the king comes in with Kf5-e6-d7 trapping the bishop. White wins. 46...Qd6†?! 47.Qxd6 exd6 was hopeless and I did not really consider it. 48.Ra7 Bc3 49.f6 Kg6 50.f7 Kg7

Black is in a sort of fortress set-up. But after 51.Kg2, White will play Ra7-e7-g4 and take the black kingside pawns. After this the win is a matter of not-too-difficult technique.

510

47.Rg6† Repeating the position once to gain time on the clock, as everyone has learned to do since their childhood. 47...Kh5 48.Rg7 Kh6 49.Rxe7

49...Kg5? This is lost by force, but during the game I did not see a mate so what am I to do? Luckily my opponent did not see it either. If I had seen a mate against 49...Kg5 or even suspected that it was there, I would definitely have gone for the alternative. 49...Qd4 This allows the transition into an endgame, but it was not that clear to me during the game that winning this would be that easy either. It turned out that I was right, as the ‘win’ we found evaporated when we looked deeper into the position. 50.Qe3† Qxe3 51.Rxe3 Kg5

511

52.Rb3 It is crucial for the understanding of the position to know that 52.Rd3 Be7 53.f3 gxf3 54.Rxf3 Bf6 is a draw:

Stockfish 10 starts out all excited, stating +7.72, but is unable to back it up. Black has an impenetrable fortress. 52...Bd4!! This discovery was made while writing this book. Black gives up the a-pawn in search of counterplay on the kingside. 52...Be7? loses in only one way, to a move which is extremely difficult to find.

512

53.Rb6!! (53.f3 we already know is a draw. 53.Rb5 Bb4 54.f6† Kxf6 55.Rh5 g3† also leads to a draw.) 53...Bd8 (53...Bb4 54.f6 Kg6 55.Kg2! and White will either take the a5-pawn or the black pawns on the kingside, not allowing any ...g3† tricks.) 54.Rb5 Bc7† 55.Kg2 Bd8 56.Rd5! Manipulating the bishop to a poorer square. 56...Bc7 57.Kg1! A nice waiting move. 57...Kf6

58.f3! gxf3 59.Kf2 h3 60.Kxf3 h2 61.Kg2 Kg5 62.Kh1 Kf6 63.Rc5 White wins. 53.Rb5 53.Kg2 Bc5 54.Rb5 should be met by:

513

54...h3†! Driving the white king backwards. (54...Bb4? 55.f6†! Kxf6 56.Rh5 h3† 57.Kg3! and White wins.) 55.Kg1 Bb4 Black draws. 53...Bxf2! 53...Bc3 also holds, but this is simpler. 54.Rxa5

54...Bg3†!! But only if you see this move! 55.Kg2 Bc7 56.Ra7 h3† 57.Kh1 Bf4 Black has created just enough counterplay. 58.Rh7 Bd2 514

White cannot make progress. Before moving on, let us also note that 49...g3†? gets nowhere after 50.Kh3!.

50.Re8? Missing a clever way to give up the f5-pawn in order to expose the black king to a strong attack. 50.Qe3†! Kxf5 51.Qe6† Kg5

515

52.Rd7! This move is nice, but not so difficult to see. The real problems arise later. 52.Qe3† Kg6 53.Re4 is the only serious alternative. However, after 53...Bg5 54.Qd4 Qxd4 55.Rxd4 g3† 56.fxg3 hxg3† 57.Kxg3 Be7 Black is in time to make the draw. 52...g3† 52...Qf3 53.Qd5† wins. 53.Kg2 Qc2 54.Rd5†! Kg6 55.Qg4† Kf7

After a series of precise moves, it is time for White to find a difficult “creeping death” move. 56.Qf5!! White can give checks and stuff, but eventually he will have to find this idea. The combination of 516

Rd7† and the exchange of queens is decisive – though not without a few complications. 56...Qc6 The last small piece of resistance. After 56...Qxf5 57.Rxf5 gxf2 58.Rxa5 it is not too difficult to understand that White wins. The black king is too far away from the queenside to fight the advance of the a-pawn. 58...Bc3 59.Rh5 Be1 60.a5 Kg6 61.a6 for example. 57.f3! 57.fxg3 hxg3 58.Qe4! also wins. 57...Qc4! Necessary in order to defend the e6-square. 58.Rd2!

There are other wins by now, but threatening both Rc2 and Qxa5 is cleanest. Finally, we have come to the end of the line.

517

50...Qc2? A grave error. I have no excuses. This was really poor. 50...Qd4! (d2 and d3 also work) and White has nothing. The endgame after 51.Qe3† Qxe3 52.Rxe3 Kxf5 is an elementary draw. Even 52...g3† 53.fxg3 hxg3† 54.Kxg3 Kxf5 is just a draw. 51.Qe3† Kh5 51...Kxf5 is not advisable on account of 52.Qe6† Kg5 53.Rg8† and mate.

52.Qf4! This is deeply unpleasant. 518

52.Re4? wins according to the computer, but once against Black can survive by: 52...g3† 53.Kh3 Qd1 54.Qe2† Qxe2 55.Rxe2 gxf2 56.Rxf2 Kg5 Reaching an endgame which we have already established to be a draw. 52...Qb3 53.Re3 This is good enough, but White also had another strong move at this juncture. 53.Rc8!

This almost feels like zugzwang, but Black can avoid doing anything with: 53...Qh3† 54.Kg1 Qb3 However, White is not relying on zugzwang, as he has a threat of his own. 55.Rc7! Finally there appears a threat of Rh7 mate, which White has been dreaming of endlessly.

519

55...Qd1† 56.Kg2 Qd5† 57.Kh2 Black is lost. He can give a pawn away with: 57...g3† 58.fxg3 hxg3† 59.Kxg3 But the inevitable can only be delayed with a bunch of obviously hopeless checks. Hopeless because the queen is too far away from the king. 59...Qg8† 60.Kf2 Qa2† 61.Kf1 Qb1† 62.Rc1 Qd3† 63.Kg2 Qe2† 64.Kg3

And it is either the exchange of queens, or 64...Be5 65.Rh1† with mate. 53...Qd1

520

We are now entering a phase of the game where all White must do is find the win. But this is proving to be not so easy! 54.Re6 The win is a version of the same thing over and over again. This time it is like this: 54.Re4! with a proper zugzwang. It is reproduceable, as White can play Kg2 and claim zugzwang once again. But there are no brute-force solutions at hand. 54...Bg5 55.Qc7 (or 55.Qe5) 55...g3† 56.fxg3 hxg3† 57.Qxg3 With the rook on e4 instead of e6, 57...Bf4 can be refuted with 58.Rxf4! and there is no stalemate. 54...Qf1 At this point Vidit understood that I was playing for stalemate, as it comes quickly after 55.Rxf6 g3†!

521

56.fxg3 hxg3 57.Qxg3 Qh1†!, but he did not see how to avoid it. 55.Re4 Qd1! The best practical chance. I could also have tried 55...g3† and those paralysed by the engines would probably prefer this. But after: 56.fxg3 hxg3† 57.Qxg3 Qxf5 58.Rf4!

I believed the endgame to be lost. The black king is cut off far away from the queenside. 58...Qe6 59.Qh3†! Qxh3† 60.Kxh3 White will keep the black king cut off, take the king to b5 and then get the rook around to take on a5. Please notice that White also wins if the black king is cut off on the f-file, but of course Black should try to use the bishop as a shield to get to the e-line, from where he will make the draw. 522

56.Qc7 I saw that he could play 56.Kg2! and we have the same zugzwang as on move 54. But again he did not find it. I presume he was looking for something proactive. 56...Kg5

57.Qf4† At this point I saw a nice win for my opponent: 57.Qh7! Qd6† 58.Kg2 h3† 59.Kg1 Qd1† 60.Kh2 Qd6† 61.f4†! gxf3† 62.Kxh3 and at the end of this beautiful triangulation, White wins. It is possible I would have tried a last trick. 57...Be5†! This seems entirely senseless, but it could confuse a player running out of time. 58.Rxe5 g3† 59.Kg2 (59.fxg3?? hxg3† and a check will soon emerge on f4, winning the white rook, securing a drawn endgame.) 59...Kf4!?

523

The trick here is to get White to defend all his bits with 60.Re3?, when Black surprisingly has a perpetual: 60...Qd5† 61.Kf1 Qd1† 62.Re1 Qd3† for example. The win is not so difficult though. After 60.Qxh4† Kxe5 61.Qxg3† the queen or the pawn ending is winning. Again, we notice that when the second f-pawn has not yet reached f5, the black king cannot take it and make its way back to c8 in time to prevent White from putting his king on b7. 57...Kh5 58.Re6? The last miss. After this there is nothing he can do to avoid the stalemate. 58.Kg2! still wins with zugzwang. 58...Qf1!

524

59.Qe3 59.Re4 no longer wins, as Black can claim a draw by threefold repetition due to 59...Qd1!. 59...Bg5! The stalemate really is the only defence. Black loses slowly after: 59...h3? 60.Qg3 Bh4

61.Rh6†! Kxh6 62.Qxh4† Kg7 63.Qxg4† Kf8 64.Qg3! Ke7 65.Qg7† Kd8 66.Qg8† Kd7 67.Qe6† Kc7 68.Qe4! Qxf2† 69.Kxh3 This queen ending is winning, as the black king is cut off. 60.Qd4 g3†!

525

60...Bf4† or even the ridiculous 60...Bf6!? also work. 61.fxg3 hxg3† 62.Kxg3

62...Bf4†!! At this point not surprising at all, but still a beautiful finish. 63.Qxf4 Qg2† 64.Kxg2 ½–½ As we had already worked on the chapter about stalemates for this book, I was unable to curb my excitement and immediately after the game told my opponent that this game would fit well in the book. Vidit was not happy. Not because of the book, which I don’t think he feels unhappy about contributing to, but simply because he missed a win. And I want to say that anyone in his shoes would feel the same.

Epilogue – Two Relevant Endings

526

I briefly mentioned that White is winning in this endgame above, but as it is somewhat nonstandard, I felt it was worthy of a closer look. Especially as the win is quite instructive. It is all about the rook dominating the bishop. 1.Kh3 h5 The pawn belongs here. After: 1...Bf6 2.Kg4 Black has to move the h-pawn anyway, as otherwise the position is a sort of zugzwang. Black cannot allow Kg5. 2...Kh8 2...Bd4 3.Kg5 Bxf2

527

4.Rd7! Bg3 5.f4 and Black is in zugzwang. 3.Rf7 Bd4 4.Kg5 Bxf2

5.Kh6!? There are slower and more human ways to win, but this works well. 5...Be3† 6.f4 Kg8

528

7.Rxh7! Bxf4† 8.Kxg6 White wins because the bishop will soon run out of safe squares. 8...Bg3 9.Rh3 Bd6 10.Rd3 Be7

11.Rc3! And wins. 2.Kh4 Bf4 Preventing the white king from penetrating via the g5-square, which would be deadly. 3.Rb4 529

3...Bd6 3...Bh6 This loses as well. 4.f4 Kf7

5.f5! Forceful and effective play. The slow route would be 5.Rb7† Ke6 6.Kg3 Bf8 when the win takes a while, but it will come: 7.Rb5 Bg7 8.Rg5 Kf6 9.Kf3 Bh6 10.Ra5 Bf8 11.Ke4 Bb4 12.Ra6† Kf7

530

13.Rc6! Dominating the bishop, when the win becomes relatively straightforward. For instance, 13...Ba5 14.Ke5 h4 15.f5 g5 16.f6 h3 17.Kf5 h2 18.Rc1 Bd8 19.Rh1 Bxf6 20.Rxh2 Kg7 21.Rg2 Kf7 22.Rxg5 and wins. 5...gxf5 6.Kxh5 Bf8

7.Rb5! Winning an important tempo. After something like 7.Rb6? Be7! Black is holding. 7...Kf6 8.Rb6† Ke5 9.Kg6 f4 10.f3

531

White is completely winning. A plausible continuation could be: 10...Bc5 11.Rb5 Kd4 12.Rxc5! 4.Rb6 Be7† 5.Kg3 Kf7 6.Kf4 Bf6

This has a lot of similarities with well-known fortresses, except that White has the breakthrough options with the f-pawn. 7.Rc6 Ke7 8.Ke4 Kf7 9.f4 Ba1 10.f5 g5 At this point direct play works. 11.Rh6 h4 532

12.f6! Simplest, although at this point anything wins. 12...Bxf6 13.Kf5 Bd8 14.Rh7† Kg8 15.Rb7 Kf8 16.f3 Be7 17.Ke6 Bd8 18.Rd7 Bb6

19.Kf6 h3 20.Kxg5 h2 21.Rh7 White wins.

533

Chapter 11 The Relevance of Endgame Studies

Gelfand, Smirin, V. Morokhovsky (sponsor of Pivdenny ACP Cup), Tkachenko and a bit of Rublevsky

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

534

White wins (see page 267)

A study-like win... because it is a study! (see page 269)

535

How should Black fight the pawns? (see page 274)

How can White win this endgame? (see page 275)

536

How should Black react here? (see page 276)

What is White’s strongest move? (see page 277)

537

White has a nice resource with all those pawns... (see page 278)

I have enjoyed solving endgame studies since my childhood. This training is useful for getting a good feeling for the potential of the pieces – not only in the endgame. But the main reason why endgame studies exist is not for their value for the practical player, but in their own right. The beauty of some studies continues to captivate the imagination of chess fans years and at times decades or centuries after their composition. They show us what is possible in chess and broaden our imagination. No other sport has its own poetry. The closest thing I can think of are trick shots in snooker. Many great players have composed their own studies. Reti, Smyslov, Timman and Benko, for example. I have in no way reached their heights, but have been involved in a few compositions over the years. Two can be found below. Boris Gelfand & Mikhail Zinar Original 2015

538

I got this idea of mutual zugzwang and I contacted Mikhail, who is a top expert in pawn endings. He took my idea and made this nice little study. The key idea is reciprocal zugzwang. 1.Kb6 Kc8 2.d3! The alternative approach fails like this: 2.b3? d3 3.h3 3.g6? does not work this time around. The difference from the parallel lines is that Black queens with a check on g1.

3...h5! 4.gxh6 gxh6 5.h4 h5 White is in zugzwang. He is fortunate that he makes a draw, a variation that is worthy of a study 539

in itself. 6.Kc5 Kxc7 7.Kxb4 Kd6 8.Kc3 Ke6 9.Kxd3 Kf5 10.Ke3 c5 11.Kd3! Kf4 12.Kc4 Kg3! 12...Kxf3 13.Kxc5 Kg4 14.Kd5 draws comfortably for White. 13.Kd5! Removing the important pawn on e5. 13...b6! 13...Kxh4 14.Kxe5 and White gets a queen on f8 just in time.

14.Ke4!! 14.Kxe5? looks similar, but after 14...Kxf3 15.Kf5 b5! 16.Kg5 (16.Ke5 Kg4! and Black wins) 16...b4 17.d3 Ke3 18.Kxh5 Kxd3 19.Kg6 c4 20.h5 cxb3 21.h6 b2 22.h7 b1=Q 23.h8=Q Kc4† 24.Kf7 Qf5† 25.Ke7 b3 Black has achieved a tablebase win. 14...Kf2 15.d3 Kg3 15...Ke2 16.Kxe5 transposes. 16.Kxe5 Kxf3

540

17.Kf5! Ke3 18.Kg5 Kxd3 19.Kxh5 Kc3 20.Kg4 Kxb3 21.h5 With the pawn all the way back on b6, it is White’s time to appreciate the adjudication of the tablebases: Draw. 2...b3

3.g6! The beautiful strike that makes the study. 3.h3? fails to 3...h5! and it will be White that ends up in zugzwang. 3...hxg6 541

Black cannot escape. After 3...h6, White plays 4.h3! and it will be Black that ends up in zugzwang. 4.h4 g5 5.h5! Again, deferring the move. 5...e4 Black has to try something, even though nothing works. 5...g4 6.fxg4 e4 7.g5! and White does not arrive first, but best. This is a common theme of pawn studies. 5...c5 does not work on account of 6.Kxc5 e4 (6...Kxc7 7.Kd5 and White wins in the good oldfashioned way by scooping up all of the pawns.) 7.fxe4! g4 8.Kxd4 g3 9.Ke3 and the white king catches the pawn. 6.fxe4

6...g4 Black will arrive first, but White, again, will arrive best. 7.e5 g3 8.e6 g2 9.e7 Kd7 10.c8=Q† Kxc8 11.e8=Q# I composed the next study together with my friend Sergey for the Memorial Tournament for Mark Dvoretsky’s 70th Birthday. The great Russian trainer sadly passed away in 2016, but his family and pupils organized a great event in his memory. 542

We achieved a respectable 7th place and won a €100 prize(!), I think. Some of the other contributions to the competition were amazing. Jan Timman won first place with an astonishing study. Sergey Tkachenko & Boris Gelfand Memorial Tournament for Mark Dvoretsky 2017

1.Kb5 There is a tradition with studies that they are composed going backwards as far as possible. I got the idea for this study when doing opening work with the young Russian Grandmaster Daniil Dubov. At some point it transposed into an endgame where the computer believed the position was equal. I did not understand why and proceeded to move the pieces around, trying to understand the position. Eventually, I got to the position we will reach at move 7 and thought this would be a good idea for a study. From the perspective of the practical player, you could argue the value of this tradition. When the first six moves, as in this study, are easy to find, the study does not become more interesting from the perspective of solving. Other moves at this point do not work. After 1.Nf6 Black plays 1...Ka4 and after 1.Ne7 Bb8 2.Kb5 e5 Black achieves sufficient counterplay as well. 1...e5 The only move that offers Black counterplay. After 1...Kb3 2.Ne7 Bb8 3.Nc6 White wins with 4.Nd4†. 543

2.Nf6 Necessary. 2.Ne7 would allow Black to survive with 2...e4 3.Nc6 e3. All of the endgames with queen and knight against queen are drawn. There is no miracle check followed by a knight fork. 2.Ka6 on the other hand fails to 2...Bb8 3.Nf6 Kb4 4.Kxb6 Kc3!. 2...Kb3

3.Ne4! It would be a mistake to play 3.Ka6? Bb8 when White has no way around: 4.Kxb6 (or 4.Ne4 Kc4 and the king comes back to secure the draw) 4...Kc3! Black draws, as we saw in the previous note. 3...Kc2 The toughest defence. 3...Bb8 4.Kxb6 Kc4 does not offer a lot of resistance. After 5.Nc5 the threat of Na6 and Kc6 forces Black into desperation. 5...e4!? 6.Nxe4 Kd5 7.Nc5

544

This is a theoretically winning position. The bishop cannot stop the pawn. 7...Bh2 (7...Kd6 8.Na6 and wins) 8.Na6 Threatening Nc7. 8...Bg1† 9.Kc7 Bh2† 10.Kd7 Black cannot prevent White from getting a queen. 4.Ka6 Bb8 5.Kxb6 Kd3 6.Nc5† Kc3!

This is the position I reached in my initial analysis! It takes the engine a number of seconds to find the win, even on a powerful computer, so I guess it is not entirely elementary – although for a human with some experience of studies, White’s move is not a great surprise. 7.Kb5! 545

The first step in a classical triangulation. 7.Na6 is premature. After 7...e4 8.Nxb8 e3 the endgame is a draw. 7...Ba7 8.Ka6 Bb8 9.Kb6 Zugzwang. It turns out that c3 is the only square where the black king will not run into a check once White goes for the black bishop.

The same position as the previous diagram, but with Black to play. The zugzwang decides. 9...Kd2 9...e4 fails to 10.Nxe4†, Nc5, Na6 and Nc7, blocking the bishop. 9...Kb2 10.Nd7 e4 11.Nxb8 e3 12.Nc6 e2 13.Ne5 wins. Against 9...Kd4, White follows the same path as the main line and wins easily because 12.Nc6† will gain a tempo. 9...Kc2 loses similarly, with 12.Nc6 and 13.Nd4† picking up the e-pawn. Finally, 9...Kc4 will allow 12.Nc6 and 13.Ne5† when the knight gets back in time. 10.Nd7 e4 11.Nxb8 e3 12.Nc6 e2

546

13.Nd4! Winning an important tempo, attacking the pawn. 13.Ne5? would however be wrong, as after 13...Ke3! Black also gets a queen and thus make a draw. 13...e1=Q 14.Nf3† White wins.

A study in a rapid game Often studies are inspired by themes from real games – mirroring the way that solving studies not only helps us improve our calculation and imagination, but also expands our minds as to what is possible on the chess board. There are many games from this book that could have been put in this chapter, but I decided to stick with one that stands out, partly because the lines are not as clear-cut as in a study. I want the readers to see the way solving studies can make us more creative, and then use this creativity in their own games, no matter how messy the circumstances. The game was played in the 22nd round of the 2008 Melody Amber tournament, where you played a blindfold and a rapid game against the same player, on the same day. This was the last game of the last day. It is not unfair to say that we were both past our “best before” date at this point. The blindfold game had ended badly for me, when I blundered horribly. We will join the rapid game at a time where I was heading for a second defeat. However, I managed to create counterplay and use a lot of ideas that could be used to create studies. I believe at least part of my inspiration came from ideas featured in studies I looked at in the past, 547

but no longer consciously recall. Our feeling for chess is, after all, developed by seeing a lot of chess, most of which we will later forget the details of. When I was young I met Alexander Koblencs, famous for his work with Mikhail Tal. He told me to remember not full games, but only the key moments. Before we look at the game, I want to say a few words about my opponent. I have played Shakhriyar Mamedyarov a lot of times, since he came onto the scene. The first game was in 2004 (see page 281). He was always a very dangerous player. I think his greatest strength is his feeling for harmony among the pieces, which he often uses to create an initiative. Earlier in his career Mamedyarov tended to take excessive risks in an all-or-nothing approach. This was a weakness, which he has overcome, including a solidity in his play, without which it would not have been possible for him to achieve all he has. I did not know that Mamedyarov was ranked clear 2nd in the rating list at the time of writing this chapter. But honestly, this does not matter to me at all. Rating is a result of average score over a long period of time, but what really matters for the serious player is to win tournaments. This is where Mamedyarov has impressed me. He won the Tal Memorial in 2010, Shamkir 2016 and 2017 and has been a part of the highly successful Azeri team, winning the European Team Championship in 2009, 2013 and 2017. In this way, Mamedyarov has impressed me more than some other unnamed top players who are often (rightly, I may add!) mentioned as possible challengers to the World Champion. All great players who have crossed the 2800 mark at some point, but none of whom have achieved the same level of tournament success as of yet. Mamedyarov is a three-time candidate. In 2011 he got a wildcard. I eliminated him in the first round of this knock-out tournament. Game 3 from this match will feature in a future volume of this series, when we will discuss my lifelong love of the Sicilian Najdorf. Mamedyarov qualified for the 2014 Candidates with a second place in the Grand Prix, making a decent 50%, ending in 4th place. The 2018 Candidates was played while we were writing this book. Mamedyarov was the top seed, having qualified by winning the Grand Prix. He played strong chess and tied for second, although of course in the Candidates there is really only one prize of importance. Boris Gelfand – Shakhriyar Mamedyarov Nice (rapid) 2008 1.Nf3 g6 2.d4 Bg7 3.c4 d6 4.e4 Bg4 5.Be2 Nc6 6.Be3 Bxf3 7.Bxf3 e5 8.d5 Nd4 9.0-0 Ne7 10.Bxd4 exd4 11.Nd2 0-0 12.Be2 c5 13.Bd3 g5 14.g3 Ng6 15.Kh1 Kh8 16.Qe2 Qf6 17.Rad1 Rae8 18.Bb1 Bh6 19.Qh5 Ne5 20.f3 a6 21.b3 b5 22.Rf2 Qg6 23.Qxg6 fxg6 24.Rdf1 Rb8 25.Kg2 Rf7 26.h3 Rbf8 27.Bc2 Rf6

548

By the time we join the game, both players were low on time. White’s position is horrible – no other word truly covers it. His position is under pressure from all directions, with a lot of weaknesses and passive pieces that cannot be improved. 28.Bb1 All White can do is wait, which brings us to an important point we know from the praxis and teachings of Akiba. When we have a large positional advantage, we should improve our position as much as we can, before acting. The poetic version of this advice goes simply by “do not hurry.” Being short of time, I doubt Mamedyarov made such considerations. I think he saw a combination that worked, without having a chance to second-guess himself by going deeper. 28...g4!? This is still a good move and hard to criticize, as it is a winning move. But I want to share the study-like triangulation with 28...Kg8! 29.Bc2 Kg7, where White cannot do anything meaningful other than 30.Bb1, when the combination from the game wins easily. 29.f4! When you have such a poor position as I did here, you will grab desperately at any chance to create counterplay that comes your way. Even if it costs you a rook. 29.hxg4? is entirely pointless. After 29...Bxd2 30.Rxd2 bxc4 31.bxc4 Rxf3 Black does not have to show any technique to win the game. 29...gxh3† 30.Kxh3 g5

549

Exploiting the pin down the f-file. But it is not an absolute pin... 31.fxe5! Sacrificing a rook for a cluster of passed pawns. 31...Rxf2 32.Rxf2 Rxf2 33.exd6

White is still objectively lost, but in the real world, with an opponent short on time and not a lot of time himself, things turned out to be more than a little tricky for Mamedyarov. 33...g4†! 550

Including the bishop in the game was not likely to be a difficult decision for Mamedyarov. As said, he was always great at making his pieces play. 33...Rxd2? is obviously poor. 34.d7 and the pawn promotes. Also bad is 33...Rf7?, when White wins with: 34.e5! g4† 35.Kxg4 Bxd2 36.e6 Rg7† 37.Kf5 Bg5, where White can win in a lot of ways, for example 38.e7 Bxe7 39.dxe7 Rxe7 40.d6 and the pawn will soon claim the rook. 33...Bf8? allows White to reveal his key idea. The pawns are incredibly strong after 34.d7 Be7 35.d6 Bd8, when he has 36.b4!, creating an extra passed pawn. 36...h5 is the last attempt, but White can play 37.g4 avoiding mate in one, after which White will quickly advance the c-pawn up the board. 34.Kxg4 Bxd2 35.e5

35...Kg7 The natural move, bringing the king in to deal with the passed pawns. 35...bxc4? is missing the point. After 36.e6 the pawns are too strong. Black can try to deal with them with 36...Ba5 37.e7 Re2, when White has 38.Bf5! preparing 39.Be6, winning. 35...Re2? would allow the pawns to claim a big trophy. White draws with: 36.Kf5 bxc4

551

37.e6! Ba5 38.d7 Kg7 39.d6 Kf8 40.e7† Black has to give up his rook and force a drawn ending. It should be mentioned that 35...h5† is also playable at this point, after which Black wins. 36.d7 Threatening to promote the pawn makes the game sharper. After the game I was thinking it might have been better to play 36.e6 but, upon closer inspection, this is not so. Black can win with 36...Ba5, when nothing works for White. 36...Rf8 also wins, but after 37.e7 Black has to find the funny-looking, but quite logical, 37...Rg8!, which wins an important tempo. 36...h5†?! It may looks good to drive the king back; but from a practical point of view, this move is the beginning of Black’s downfall. 36...Ba5 would have been fine. 37.b4!? would still be worth a try, but Black wins with: 37...Bd8! (37...cxb4?? loses in the same way as 38...cxb4? in the notes to the main game below)

552

The position after 38.bxc5 bxc4 39.d6 c3 40.c6 c2 41.Bxc2 Rxc2 42.e6 would be winning for White, if Black did not queen with check, when mate is close. For this reason, I was intending to try 38.cxb5!?, but after 38...cxb4 39.bxa6 d3! 40.a7 d2 41.Bc2 d1=Q† 42.Bxd1 Rxa2 Black wins. 37.Kh3

37...Ba5? Not knowing which way to get the bishop back to control the d8-square, Mamedyarov went the wrong way, clearly overlooking my reply.

553

37...Rf8? This would also have been bad. White wins in a long and complex variation: 38.d6 Rd8 38...Kf7 39.Bf5 is a common theme for this game. 39.e6 Black has to give up the rook to avoid an immediate loss. 39...Rxd7 40.exd7 Bg5

But this endgame is winning for White with accurate play. 41.g4! h4 41...bxc4 42.gxh5 c3 43.Kg4 Kh6 is winning for White, because of zugzwang. 44.Bd3 a5

554

45.Bc2! a4 46.Bd3 axb3 47.axb3 Bf6 (47...Bd8 48.Bg6 with the idea Kf4-e5 also wins.) White wins easily by creating another passed pawn on the queenside. 48.Kf5 White can also start with Bb1-c2 to improve his position, but this is good enough. 48...Bh4 49.Ke6 Kxh5 50.Kd5 (50.Bf5 Bd8 51.Kf7 Kg5 52.Bh7 works in a simpler way.) 50...Kg4 51.Kxc5 Bf6 52.b4 Kf4

53.d8=Q! (53.b5? Ke3 54.Kc4 Bd8 would be a draw.) 53...Bxd8 54.Kxd4 Bb6† 55.Kd5 Ke3 56.Bh7 White wins easily. 42.cxb5! axb5 43.a4 bxa4 44.bxa4 Bd8

45.g5! A key move that seriously restricts the black king. Black cannot take on g5 because the a-pawn will march to victory. 555

45...Kf7 Passive play is pointless. White can improve his position quite easily. 45...c4 46.Bf5 d3 47.Kxh4 Ba5 48.Kh5 Bd8 49.Be4 Ba5 50.Kg4 Bd8 51.Kf5 Ba5 52.Ke6 Bd8 53.g6 Kf8 54.Kd5 and so on... 46.Kxh4 Ke6

47.Kh5 Kxd7 48.g6 Bf6 49.Bd3 Kxd6 50.a5 White wins slowly and easily. 37...Bg5! was a much more natural follow-up to the ...h5 check, making full use of the g5-square. 38.b4 is not worth much now, without the gain of tempo. 38...bxc4! 39.bxc5 c3 40.Bc2!? The last trick. 40...d3! (40...Rxc2? 41.c6 Bd8 42.d6 and we end in a variation with perpetual check as in so many other positions.) 41.Bxd3 c2 42.Bxc2 Rxc2 43.c6 Kf7 Black wins. 38.b4!

556

I did not hesitate to grab this chance to muddy up the position. Later on, analysis shows that White is already back in the game, but while playing, all I could know for sure was that my chances were improving. 38...Kf7? Startled and short of time, Mamedyarov blunders horribly. 38...bxc4? would also be bad, as after 39.bxa5 Rf8 40.d6 Kf7 White wins with both 41.Bf5 and 41.Bg6† Kxg6 42.e6, in both cases winning a tempo by advancing a pawn with check. 38...Rf8? 39.bxa5 is also just bad. 38...Bd8!? This was a serious option. White is still doing well, having been given an extra passed pawn. 39.bxc5 bxc4 40.c6 c3

557

41.d6! White arrives just in time. 41.Bc2? does not work this time. After 41...d3! 42.Bxd3 c2 43.Bxc2 Rxc2 the white pawns have been brought under control. 41...c2 42.Bxc2 Rxc2 43.e6! The white pawns are unstoppable. Actually, it is Black who is escaping with a draw on account of the d4-pawn.

43...d3 43...Kf6? would lose to 44.e7! Bxe7 45.c7! and White arrives first, best and everything. 44.e7 d2 45.exd8=Q d1=Q

558

Black is a rook up still; White has to give perpetual check. 38...cxb4 This would have been the most natural reaction. 39.c5 Kf7 Black can also play 39...Rf1, when White makes a draw by simply pushing the pawns. 40.d6! Rxb1 41.e6 d3 42.e7 d2 43.e8=Q d1=Q 44.Qe7† with perpetual check. 40.c6 Ke7

41.d6† This position is interesting, as Black can easily go wrong and lose the game. 41...Ke6?! 41...Kd8 42.e6 Re2 43.e7† Rxe7 on the other hand is a direct draw. 42.Bg6 Rf1? 42...Bd8 is the only move, allowing a nice combination.

559

43.Bf7†!! Rxf7 44.c7 Rxd7 (44...Kxd7 45.e6† is the key point) 45.c8=Q d3! Black has to get out of the pin and can only do this by giving up this pawn quickly. 46.Qc1 Kxe5 47.Qe3† Kxd6 48.Qxd3† Kc6 49.Qg6† White can press, but Black will hold a draw with accurate defence. After Black’s mistake on the previous move, White wins. But only with very accurate play. 43.Kg2!! 43.Be8? allows Black to make a draw with 43...d3 as in so many lines. 44.c7 d2 45.d8=Q Rh1†! 46.Kg2 Rg1† with a draw, as 47.Kf2?? Bb6† would allow Black to queen with check. 43...Rc1 43...Rf8 44.Be8 and White wins. He will soon get two queens, the second with check. 44.c7! Rxc7 44...Bxc7 45.Bf5†! is beautiful, but not too surprising.

560

45.d8=N†! White gets a winning endgame. For example: 45...Kxe5 46.dxc7 Bxc7 47.Nc6† 39.bxa5 Ke7 40.cxb5 axb5 41.a6 In time trouble Mamedyarov had not fully understood the power of this passed pawn. 41...Rf8 42.Bf5! Black resigned. On 42...Rxf5 I would have played 43.d8=Q†! Kxd8 44.a7. 1–0

561

Chapter 12 Geometry

With Mamedyarov, Rublevsky and Mikhail Pleshkov

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

562

How should White fight for a draw here? (see page 281)

How can Black win? (see page 283)

563

Black to hold! (see page 284)

White has a remarkable geometric win (see page 287)

564

Winning the easy way would have been easier! (see page 288)

Very deep. How does Black win? (see page 288)

565

Here I missed a beautiful win (see page 291)

But this win I found! (see page 295)

You will not find geometric shapes in nature (with the exception of crystals and perhaps other micro levels). No squares, straight lines and so on. Everything is curved and rugged. Those shapes you will not find on the chess board. There everything is dominated by the geometric properties of the pieces. The only curves you find are those provided by the maker of the pieces. 566

Mamedyarov This was my first game of many with Mamedyarov. At the time he was only one of many promising young players. He had won the World Junior Championship the year before and would win it again the following year. Only a few years later he was one of the best players in the world. Shakhriyar Mamedyarov – Boris Gelfand Pamplona 2004 I had been better the whole game. Mamedyarov had defended well, but still had a job to do here.

The first thing we notice when we see this endgame is the potential for a transition into rook and knight against rook, which is very drawish. I drew with Bacrot, for example. Famously Kasparov won against Polgar, but her king was cut off in a bad way. Also, Dominguez had won a game against Onischuk in Biel: Alexander Onischuk – Leinier Dominguez Perez Biel 2008

567

When the players entered this endgame, White had the king in the middle of the board. So Dominguez had already made a lot of progress to get to here. But still, if Onischuk had done nothing and just kept the rook on the c-file, or any other file to the right of it but the d-file, the draw would still have been there. 88.Rd7?? Kc2 89.Ka3 89.Ra7 Rb2† 90.Ka3 Nb5† and the rook goes. 89...Nc6! It is easy to see that the rook would have been effective from the h-file now. 90.Ka4 Rb4† 91.Ka3 Rb5! 0–1 Of course, after reading this chapter, you will know that it is desperately important not to be on the corner square. Just one step away from the corner and White holds. So, while you will be better equipped to play this particular endgame, there are many other such endgames, where you cannot know all of the finesses. We know this from the great wealth of endgame studies composed over the last two hundred years. 65.Nd4†? You may have the feeling that White should be able to draw in many ways here, but if he does not get out of the corner, he will not survive. 65.Rxa7? Rh5† allows mate directly. 65.Kg1 is thus one of the two moves that hold. The following variation is basic: 65...Rd3 66.Rxa7 Ne2† 67.Kh2 Rxb3 68.Ra8 Nf4 and White draws with three moves, 69.Ra1, 69.Ra2 and my choice, 568

69.Rf8, when the knight is unable to do any harm. But let’s look more deeply at the dangers after: 65.Kh2 a6 66.Rf8 66.Kg1 is perhaps more practical, but during the game, how would you tell? Only by calculation. You go in the direction where you see the fewest dangers. 66...a5 67.Nxa5 Rh5† 68.Kg1 Rxa5

White is clearly in a critical situation. After 69.Rf6 Kg3, mate is coming. We will consider two options: a) 69.Kh1? Once again, the king cannot survive the imprisonment in the corner. 69...Kg3 69...Rg5 also wins. In general, there are a lot of ways for Black to get to the same place. 70.Rg8† Kf2 71.Rh8 Rf5 72.Rh2† 72.Kh2 Ne2 transposes. 72...Kf1 73.Rh8 Ne2! The net is slowly created. 74.Kh2 Kf2 75.Kh3 Rf4

569

This is the key position. Black now has the freedom to slowly manoeuvre the knight into place. 76.Kh2 Anticipating the check on g1. All other moves transpose. 76...Ng1 77.Kh1 Back in the corner, ready to hear the hammering of the nails. 77.Rh7 Nf3† 78.Kh3 Ng5† and the rook falls. And 77.Rh6 is met with 77...Rg4 with imminent mate – which of course was also there after 77.Rh7. 77...Nf3 78.Rh3 Ra4 Mate follows. b) Instead, White has to play 69.Rf7!, when the position remains a draw. 65...Rxd4 66.Rxa7

570

66...Rd6! I managed to take advantage of some amazing geometry. I should add that Black is also winning after: 66...Rd8 67.Rf7 Kg3 68.Rg7† Kf2 69.Rh7 Ne6!!

The first point is that after 70.Rf7† Kg3! 71.Rf1 Nf4!, Black wins. If the rook goes away on the first rank, Black wins with ...Rh8†, ...Ne2† and ...Rh1†. And after 72.Rg1† Kf2, mate is near. And after 70.Rh2† Kg3 71.Rg2† Kf3, the king is trapped in the corner in another unfortunate way. The king and knight are an excellent team. Now 72.Rg6 Nf4 73.Rg7 Rh8† leads to mate. And 72.Rb2 is met with 72...Nf4 73.Rb3† Kf2 74.Rb2† Ne2, and Black wins. 571

67.Ra2 White is lost no matter what. 67.Rf7 Kg3 68.Rg7† Ng6! also sees the rook shut out. Again, if the king were not in the corner, it would not get mated. 67.Ra3† Kf2 transposes to the next annotation. 67...Rh6† 67...Kg3 was also good enough. After 68.Ra3† Kf2 we see why the king is so badly placed in the corner. Had it been on h2, White would draw with 69.Ra2† Ne2, but now the move 70.Kh3 is not available to him. 68.Rh2 68.Kg1 Ne2† 69.Kf1 Rh1# 68...Nh3!

The crowning moment of the game. White is mated. 69.Ra2 Nf2† 70.Kg1 Rh1# 0–1 The next game also features rook and knight, but in very different circumstances.

572

We already encountered Wang Yue on page 39 and in Positional Decision Making in Chess on page 120. He is a tenacious player and a really excellent defender. Boris Gelfand – Wang Yue Dagomys 2010

I had been pressing for a long time and at some point I was winning, but eventually we reached this endgame. My understanding of the position was that it was a draw and for this reason I took some chances trying to pose him some final problems. 66.Kc4!? I allowed the knight fork, sort of bluffing. I found no other way to improve the position, as after 66.Kb4 Black draws easily with 66...g5 67.hxg5† Kxg5 and the white king is too far away. For example: 68.Kc5 Nh4 and it is time to shake hands already. 66...g5? Once again we see that it is very tricky to work out when you should go for counterplay and when you should be happy with what you have and trust that your position can withstand the opponent’s winning attempts. It was of course tempting for my opponent to win back the exchange. After 66...Nd6† 67.Kd5 Nxe4 68.fxe4 there is a choice:

573

Black cannot play 68...g5 on account of 69.e5† Kf7 70.hxg5 h4 71.Ke4! and White wins. But Black can change the move order. 68...Kf7! Now there is nothing after 69.e5 Ke7 70.e6 g5 71.hxg5 h4 72.g6 h3 as Black would queen with check, so White has to go forward. 69.Kd6 g5! But now Black gets the above variation except without the white king being able to run back to stop the hpawn. 70.hxg5 h4 71.e5 h3 72.e6† Kg7 73.e7 h2 74.e8=Q h1=Q with an obvious draw. Another way to hold the game would have been to play for a fortress. After 66...Ne7 67.Kc5 Nf5 68.Kd5 Ne7† 69.Kd6 Nf5† 70.Kd7

70...Kf7, White is unable to make any progress. Attempts to run around with the king will create the perfect time for Black to go for counterplay. The following variation illustrates the potential of the 574

black defence excellently. 71.Ra4 Kf6 72.Ke8 Ng7† 73.Kf8 Ne6† 74.Kg8 g5! 75.Kh7 gxh4 76.Kh6 h3 77.Rh4 Kf5

78.Kxh5 Nf4† 79.Kh6 Ne6 80.Rxh3 Kf4 and once the f3-pawn falls, Black makes a draw. All he has to avoid are simple traps. 81.Kh5!? Nd4?? 82.Rh4†, for example. There is no way for White to make progress, so instead Black will shuffle the knight around as he wants. 67.hxg5† Kxg5

68.Kd3! This was the point of the trick. It is quite possible that Wang Yue expected me to play 68.Kd5, when after 68...Nh4 the draw is 575

immediate. Instead Black suddenly finds himself having ruined his solid pawn structure by exchanging White’s main weakness, and he provided White with a passed pawn. 68...h4 68...Nh4 no longer makes any sense. 69.Ke3 is there. White also wins after 68...Nd6 69.Re6 Nf5 70.Ke4 Ng3† 71.Ke5, where Black is totally dominated. A key line goes like this: 71...h4 72.Re8 h3 73.Rg8† Kh4 74.Rg4† and it is all over. 69.Rg4† Kh5

70.Ke4! But this time the king has to go forward. It would of course be great to bring the king to h3, after which Black could be in a fatal zugzwang almost immediately. But there is no path to get there: 70.Ke2? h3 would allow Black to escape. White can still lose after 71.Rg8? h2 72.Rh8† Nh6!. And a draw arises after 71.Kf2 h2! 72.Kg2 Ne3† 73.Kxh2 Nxg4† 74.fxg4† Kxg4. 70...Nh6 70...Ng3† 71.Ke5 is hopeless. For example: 71...Nf1 72.Rg8 Ng3 73.Kf6 and Black has to resign quickly in order not to be mated. 71.Rg2 During the game I also considered 71.Rg1, but found the position after 71...h3 confusing 576

(71...Nf7 loses in a number of ways. The most elegant is: 72.Rg7 Nh6 73.Ke5 h3 transposing to line ‘d’ below):

a) 72.Ke3 Kh4 73.Kf2 h2 74.Ra1 Kh3 75.Ra5 looks tempting, but Black draws with 75...h1=N†. b) 72.Ra1 Kh4 73.Ra6 Kh5 74.Ra2 looked promising, but Black was able to escape with a subtle manoeuvre. 74...Kg5 (74...Kh4? would, on the other hand, lose to a beautiful manoeuvre. 75.Kf4 Nf7 76.Ra7 Ng5 77.Ra6 Kh5 78.Kg3! h2 79.Ra1 White will win a pawn and later the game.) 75.Ke3 Nf5† 76.Kf2 h2! With a draw after 77.Ra1 Kf4 or 77.Kg2 Nh4†. c) 72.Kf4 Kh4 73.Rh1 Nf7 74.Rh2 looks like a zugzwang position.

577

For example, White wins after 74...Nd8 75.Re2! and 74...Nh6 75.Ra2, in a similar way to the main line. But Black is able to escape with: 74...Nh8!! 75.Re2 Ng6† 76.Kf5 Kg3! 77.Kxg6 Kxf3 with a draw. d) 72.Ke5!! is thus the only way to go. Instead of trying to stop the pawn, White plays against the knight. 72...Kh4 73.Kf6! h2 74.Rg6! Kh5 75.Rg5† Kh4 76.Kg6 Ng8 77.Rh5† Kg3

It looks as if Black is making a draw, but after 78.Kf7! the knight is trapped. 78...Kg2 79.Kxg8 h1=Q 80.Rxh1 Kxh1 81.f4 and White wins. 71...h3

578

72.Rd2? Missing the final chance to win the game. It took me some time to understand it, even after the computer pointed it out. 72.Rh2! Kh4 73.Kf4 Ng8! The best try. 73...Nf7 loses to simple moves: 74.Ra2 Ng5 75.Ra6 Kh5 76.Kg3 transposes to the 74...Kh4? variation of line b above.

74.Rd2!! Only on this square, as White cannot allow ...Nd5†, which would be available if the rook went to 579

one of the other three queenside files. 74...Nf6 75.Ke3 The king is threatening to come round to take the h-pawn. 75...Kg3 75...Nh5 76.Rd4† wins quickly. And after 75...Kg5 76.Kf2 White will pick up the pawn quickly. 76.Rd6 Nh5 77.Rg6† Kh4 77...Kh2 After this White is obviously winning. The question is how he chooses to do it. 78.f4 seems reasonable. 78.Rg4#

This is the dream of course. 72...Kg5! 72...Kh4? 73.Kf4 and White would have won as in the lines above. Now Black is threatening to play either ...Nf5 or ...Ng4!?, both forcing a draw. 73.f4† Kh4 74.Kf3 Ng4 75.Ra2 Nf6 76.Ra6 h2 77.Kg2 Nd5 ½–½

Sakaev The following little gem is from the Vladimir Petrov Memorial rapid tournament in Latvia, which had been going for a number of years. One of the main organizers was Alexei Shirov – it is unusual that a 580

top player decides to become an organizer. It was a really pleasant tournament the year I played. I faced Karjakin, Mamedyarov and Ivanchuk (who won). I shared second with Karjakin and Rapport. Many other very strong players participated – Tomashevsky, Fedoseev, Malakhov and of course Shirov. Konstantin Sakaev – Boris Gelfand Jurmala 2015

We enter the game just in time to see me blunder. 46...Nxf4†? I could not find anything better, but felt that I should win this endgame in the long run. This is the nature of rapid chess. You cannot calculate everything accurately, but have to make a lot of decisions based on feeling. Especially when you get towards the end of the game, where you are most often living on increment. There is no point saving time to win the endgame, as you then do not have the time to get the winning endgame. 46...Rc1 would have won quickly. The key tactical points are 47.Rxd2 Re1# and 47.Kxd2 Rxd1† 48.Kxd1 Nxf4†, picking up the rook – if you want to call this tactical... The following moves are all more or less forced. 47.Bxf4 Rxh5 48.Bg5 e5 49.Rxd2 Rxd2† 50.Kxd2 Rh8 51.Ke2 Ra8 51...Rg8! is the only winning idea against the best defence, as we shall see below (where the rook goes to g7; but as you may be able to see, both are on the g-file). 581

52.Kf2 Only five years later did we discover that White has a much more resilient defence. 52.Bd2! This could have been tricky. 52...Rg8! After 52...Ra2? 53.h5!, White is already holding. So Black has to display a bit of patience and wait for just a second, asking White what he wants to do. 53.Bg5 Rg7! 54.Kd2

This is an important critical position. Even to get here is not obvious, as we have seen. But the 582

next few moves will really knock your socks off... a) When you see the winning line, the obvious question is of course: “Why should the rook be on the b-file and not the a-file?” The reason becomes obvious in the following parallel variation: 54...Ra7 55.Ke2 Ra3 56.Kf2 Kf5

At this point White can save the game with a bit of magic. 57.Bc1!! Rc3 And after 57...Ra1 58.Bh6 Ra2† 59.Kf3 e4† 60.Ke3 Rxg2 61.Bg5:

We have reached a fortress. Against 61...Ke5 62.Bf4† Kd5 63.Bg5 Rg3† 64.Ke2 Kd4, White always has 65.Bf6†, and Black will never get to give that decisive second-rank check as he likes it. 583

58.Bh6! Kg4 59.Bg7! The point. We now see with clarity that the rook was so much better placed on b3 than on c3 (and thus a3). b) 54...Rb7!! This is the deepest move of this already wildly complicated example. It would have been impossible to find over the board in rapid, and difficult in classical. 55.Ke2 Rb3 56.Kf2 56.Bd2 Rg3 57.Kf2 Rg8 followed by 58...Kd3 and Black will win. 56...Kf5!

It is important that Black wins a pawn before he starts pushing his own. As we saw above, if Black pushes the pawn to e4, White will be able to set up a fortress. This is a theme we see again and again. The king belongs in front of the pawn, so it can secure the squares the pawn can go to. White is in zugzwang, but it is not mutual. Black to play would win with ...Kg4. 57.Bh6 Kg4 The double threat of ...Rb2† and ...Kxh4 decides. This time, after: 58.Bg7 e4 Black wins the h4-pawn and the game (slowly), as White cannot accept 59.Bf6 e3† 60.Ke2 Kf5 and soon a check on the second rank will be deadly. The key point is that it is not a fortress now. White needs the pawn on h4 to support the bishop on g5, so it can check on f6 when the black king comes to d4, so ...Rb2† can always be met with Ke2-e3. It is easy to see the difference in what follows. 59.Bh6 Kxh4 60.Bf4 Kg4

584

61.g3 This certainly establishes a target, but what else? 61.Bd2 Rb2 62.Ke3 Kf5 63.Ke2 Ke5 and Black wins. 61...Kf5 62.Ke2 Ke6 63.Kd2 Everything loses. 63.g4 Kd5 64.g5 Rb2† 65.Ke3 Rg2 with zugzwang. 63.Be3 Rb2† 64.Bd2 Ke5 65.Ke3 Rb3† and once again the king penetrates. 63...Kd5 64.Ke2

64...Kd4 White must give up the g-pawn as well in order to avoid ...Rb2†, and not under good 585

circumstances. 52...Kd3 53.h5 Rg8 54.Bc1 e4 Black has made progress, but even here, with the seconds clicking away quickly, the position is not at all easy to win.

55.h6 After this move, White is threatening 56.Bf4 when the position is already a draw. 55.Bf4 Played immediately, this does not work. Black is in time: 55...e3† 56.Bxe3 Rf8† 57.Kg3 Kxe3 58.Kg4 Ke4 59.Kg5 Ke5 We can easily see the potential advantage of having the pawn already on h6. But here it is not. 60.Kg6 Rg8† 61.Kf7

586

61...Ra8! The simplest, but Black can win in many ways. The following lines all end with White resigning. 62.Kg6 62.g4 Kf4 63.h6 Kxg4 64.h7 Kg5 and it’s over. 62.h6 Kf5 63.h7 Ra7† 64.Kg8 Kg6 65.h8=N† Kf6 66.g4 Ra5 67.Kh7 Rg5 and the knight will die in the corner. 62...Ke6 63.h6 Rg8† 64.Kh7 Kf7 Almost stalemate. 65.g4 Rxg4 66.Kh8 Ra4 67.Kh7 Ra6 With mate on the next move.

587

55...Rf8†? Even if given plenty of time, I am not sure I would have been able to find the win. 55...e3† transposes to the game. 55...Rg4! This was the winning idea, preventing the bishop from coming to f4 while preparing to meet h6h7 with ...Rh4. 56.Be3 This is the most normal move. There are no other places on the diagonal and White cannot allow ...e3 without getting something in return. 56.g3 would block the white king’s advance. After 56...Rg8! 57.Bf4 Rf8 we have an elementary zugzwang. The white king will have to move, and Black will advance the pawn and win. 56...Rg6!!

This is the key move. 57.Bf4 57.h7 does not work. Black has 57...Rf6† 58.Kg3 Rf8 after which he will scoop up the pawn. 57...e3†! It is finally time to win the bishop. White would once again escape after 57...Rf6? 58.Kg3 e3 59.h7! Rf8 60.Bxe3 Kxe3 61.Kg4 Rh8 62.Kf5! Rxh7 63.g4 with classic shouldering keeping the black king out of the game. 58.Bxe3 Rf6† Here we can see the big difference. After 59.Kg3 Kxe3, the rook is perfectly placed and Black takes the h-pawn. White is not in time to create counterplay with the pawn still on g2. The alternative try cannot avoid transposition: 59.Bf4!? Rxf4† 60.Kg3 Rf6 588

Black now wins as he likes. For example: 61.Kg4 Rxh6 62.Kf5 Rh5† 63.Kf6 Ke4 64.g4 Rh6† And the rest is easy. 56.Kg3 e3 57.Bxe3 Kxe3 58.Kg4 Ke4 59.Kg5 Ke5 A draw was agreed before White would play 60.h7 or 60.Kg6, with an obvious draw. ½–½

Moscow geometry The Tal Memorial is the best tournament for me. It is strong and exceptionally well organized. They invited me there regularly, and since its inauguration in 2006 I have played almost more times in this event than all the other super tournaments combined. So, although I have heard that Bilbao and St Louis host wonderful tournaments, I am not able to confirm this through my own experience. The 2009 edition of the Tal Memorial was one of the strongest tournaments in history. And one that saw some of the best chess. I cannot say that spectators had the same experience, but as a player, this was my impression. Sometimes I had good results in this tournament in Moscow; at other times I had less good results. The best was of course when I won in 2013. We join this game in a situation where I have won a pawn and have just made the first move after the time control. We are in the fifth hour of play, a phase that is often associated with converting advantages. However, here is it not as easy as it looks.

589

Boris Gelfand – Levon Aronian Moscow 2009

I have played a lot with Aronian and he is a good friend of mine. You will be able to find our games throughout this series. Aronian collapsed, falling for a beautiful tactic, with compelling geometry. 41...Rb1? The correct defensive approach was to defend the a5-pawn and then seek to exchange pawns, getting into an endgame with few pawns that can be held. This sort of passive play is perhaps not very appealing, but at times nothing else is on offer. 41...Rc5! White has to find a plan now. It is possible to shuffle around with the pieces for a long time with 42.Rd7 Bb5 43.Rd8, in the style of Capablanca, hoping that the opponent would decide to weaken his position all on his own from either panic or boredom. This is probably a quite reasonable way to approach the position, but if we are to analyse it seriously, it makes more sense that we look at what active options would look like. It is not so easy for White to improve his position, with the a2-pawn being this weak. The way I look at this position is not as a winning position, but as a position where White has chances. It may be hard to win, but it is also hard to defend. I play this type of position according to how I evaluate it. When you are winning, you try to convert, which often means reducing complications as much as possible, if no direct wins are available. But when you think it is closer to a draw, you are fishing for chances and trying to trick your opponent or in other ways outplay him. Then there is the grey zone between the two, where 590

you do not know what the objective evaluation of the position is. There you play what you think is the best move and let things fall as they may. 42.h3!? One of the first moves that comes to mind. White is preparing to play g4 under favourable circumstances. The first thing to check is if 42.g4 works. And it no longer does. White is too reliant on Rxa5† options that are no longer there. All White is achieving is to split his pawns. After the text move there are three options to consider:

a) 42...Bb1? loses in an instant to 43.Nc4 Bxa2 44.Ne5. b) 42...Kg6? 43.g3!

591

This is rather subtle to see in advance. But if you are playing move by move (also called slowly improving your position) it is not unattainable. 43...e5 Black is forced to do this. 43...Bb1 makes things easier.

44.Nc4! Bxa2 45.Ne5† Kh6 The exchange sacrifice gives no hope at all. 46.g4 This gives White an almost mating attack and a very strong passed pawn on the g-file. On top of this, he can pick up the e6-pawn in short order. All this adds up to a decisive advantage, for instance: 46...fxg4 47.hxg4 Bb1

48.g5† Kh5 49.Ra8 and so on.

592

Black also cannot wait with 43...Kf6, when White still has a lot of ways to improve the position. For example: 44.g4 fxg4 45.hxg4 Kg6 46.Kf3 Kf6 47.g5† Kg6 48.Re7 followed by manoeuvring the knight to e5. 44.fxe5 Rxe5

45.Kf3 This endgame is winning for White with rooks on the board. Mainly because his rook is far superiorly placed. A key line is this one: 45...Bb1 46.Ra6† Kf7 47.Nc4 Rd5 48.Rxa5 Rxa5 49.Nxa5 Bxa2

50.Ke3 White wins on account of the dual threats of Kd2 and Kd4, trapping the bishop and winning the 593

pawn on b4 respectively. It is important for White that the pawn is already on g3. This tempo helps in various ways. c) 42...e5! Black has to exchange the pawn immediately in order to avoid White creating twin passed pawns. 43.Kg3 White has to keep control of the f4-square if he wants to fight for an advantage. There are now a number of ways to reach a rather generic endgame. We have chosen the one that seemed most informative to us. 43...Bb1 44.Ra6† Kg7 45.fxe5 Rxe5

46.Kf4 Re4† 47.Kf3 Re5 48.Nc4 Rd5 49.Rxa5 Rxa5 50.Nxa5 Bxa2

594

This endgame is of course better for White. We already know from above that if the pawn was on g3 and White was to play, he would be winning. But close examination here shows that Black is in time with ...Kf6-g5 & ...f4, eliminating the pawns on the kingside with a draw in sight. The evaluation of this endgame is really in the grey zone. White is obviously much better, but with concrete and active play Black holds against the concrete 42.h3. However, even the slightest hesitation and Black is lost.

42.g4! The geometry behind this move is very appealing. I was not really aware that everything else would lead to a draw during the game. You have one win and you are happy. 42...fxg4 Testing White’s attack in the most critical line. Black could have tried some other options, but none of them were appealing. 42...Rb2† 43.Kg3 fxg4 43...Rxa2 sees Black getting trapped in another mating net after: 44.g5† Kg6

595

45.Ng2! I find this move very pretty. 45...Kh5 46.Nh4 and mate. 44.Nxg4† Kf5

45.Rf7†! The most beautiful exploitation of the geometric patterns. 45.Rxa5† Ke4 46.Nf2† also wins rather effortlessly. The key point is that after: 45...Ke4 46.Nf2† White wins after 46...Kd4 47.Rd7† while if Black tries: 46...Ke3

596

47.Nd1†! This type of knight fork, on your own first rank, can be easy to miss. The best practical chance was probably 42...Kg6 43.Rxa5 Rh1, which is nothing more than hoping that something will show up later. Simple play like 44.gxf5† exf5:

45.Nd5 Rxh2† 46.Ke3 Bb1 47.Nxb4 should win for White, but he still has to make sure that there is no counterplay against the f-pawn. 43.Nxg4† Kf5 44.Ne5! This is one of the most amazing sights in this book. The bishop is sort of trapped in the middle of 597

the board. You can easily argue that the position here is not very difficult or even instructive, but I really like it all the same. Having a great feeling for geometry is one of the essential skills in chess, especially in the endgame, where there is more space for the pieces on the board for one-on-one battles.

44...Ke4 Going into a hopeless rook ending is the only thing that smells like continuing the game, although it cannot be called resistance. The many bishop moves are all losing: 44...Be4 45.Rf7# 44...Bb5 45.Rxa5 and everything hangs. 44...Bf1

598

45.Kf3! planning mate, as the bishop takes the square from the rook. 44...Bc2 45.Ke3 with the threat of mate and after 45...Re1† 46.Kd2, White wins a piece. 45.Nxd3 Kxd3 46.Rxa5 Ke4

This rook endgame is trivially won. There is no counterplay, although it is possible to end up in a rare situation where two extra pawns are not enough. Of course, you need to be careful to avoid Vancura and such things, but when you are two pawns up and the opponent has no counterplay, you should make it as simple as possible and keep advancing.

599

47.Re5† There were many ways to play this endgame. For example: 47.Rb5 Kxf4 (47...Rb2† 48.Kg3) 48.Rxb4† Kf5 49.a4 e5 50.a5 Ra1 51.Ra4 and it is time to resign. The game continuation is both slower and better. The best way to win a game is to win with the least amount of calculation needed and the least amount of counterplay. Although the above win is rather easy, it is not the type of win I would choose. Besides, winning more slowly increases the time I get to enjoy my winning position! 47...Kxf4 48.Rxe6 Rh1 49.Kg2 Rc1 50.Re2 Rc3 51.Kf2

51...Rc1 Jacob tried to find some resistance and really struggled. But we did come up with a pleasing line: 51...Rh3 52.Ke1 Kf3 53.Kd1 Rh4 54.Rc2 Ke3 55.Kc1 Kd3 56.Kb2 Rh3 57.Rg2 Kd4 Against 57...Rh4 58.a3 White will create a second passed pawn.

600

With Lautier, Henrik and Magnus Carlsen, Ponomariov, Aronian and Anand

58.Rg4†! This is not a difficult move of course, but it is better than 58.a3?? bxa3† 59.Kxa3 Kc5, where suddenly we have a drawn position! Both the black rook and king are ideally placed and White is 601

rather passive. But honestly, it would never happen. 58...Kc5 59.Rc4† Kb5 60.Rc2!

Now White can transfer the king to the kingside and slowly advance the h-pawn. With the black king in quarantine, you would expect Black to simply resign. 52.Rd2 Ke4 53.Kg3 Rc8 54.h4 Ke5 55.Re2† Kf5 56.Rf2† Kg6

57.Rf4 Rc2 58.Rxb4 Rxa2 59.Rb5 Rd2 60.h5† Kf6 61.h6 Rd7 62.Rh5 Rh7 63.Kf4 Kg6 64.Rh3 Kf6 65.Ke4 Ke6 66.Kd4 1–0

602

Chapter 13 Endings with Opposite Coloured Bishops

Oleg Pervakov showing the end of another masterpiece

Diagram Preview On this page you will find a few diagrams with critical moments from the coming chapter. If you want to compare your thinking with the games and analysis, you have the possibility. Take as much time as you need or want. This is not a test, but a chance to practise your analysis and decision making.

603

White’s win in this study is counterintuitive (see page 302)

An accurate move wins for Black (see page 304)

604

Find White’s nice drawing mechanism (see page 304)

Accuracy is needed! (see page 306)

605

What is the winning method? (see page 307)

Alex found a remarkable win for White here. Can you find it too? (see page 309)

606

Not so many options... (see page 313)

Please do better than I did here! (see page 314)

The following four endings, one study and three games from my own experience, contain more or less all the themes you will find in endgames with opposite-coloured bishops. Rather than a list, I will present you with them when they become relevant. I don’t think people are particularly good at remembering lists, but seeing ideas in practice leaves a stronger impression. Also, I expect that most 607

readers of this book will be familiar with most if not all of the themes, so mentioning them as they come into play makes more sense. The four endgames are presented in a sort of ascending scale. While the first is a study, it is still to some extent the simplest and clearest. Miroshnichenko & Pervakov 2016

I find this study by Pervakov, inspired by Miroshnichenko, to be a great illustration of the possible subtleties and some of the themes in endings with opposite-coloured bishops. The three main ones represented here are the willingness to give up material to increase the mobility of the bishop, blockade on one coloured square and of course fortress as a defensive strategy. The fortress theme in this chapter is one of the best known fortresses: the rook’s pawn with the wrong bishop. Brute force occasionally works against fortresses. Only the frail fortresses fall in this way. We shall see one of those below in my game against Shirov. The other main strategy is zugzwang, a theme almost solely found in the endgame and like stalemate, a part of the rules of chess that does not try to imitate anything in the real world. Often fortresses hold together only in one formation. Even a small change to the set-up makes it wobble and fall, if pushed in the right way. This is one of those times. 1.Ba4†! Not a difficult move to find. Studies often start with moves that the practical player would make quickly. When you try to solve them, you can either ignore this and practise your visualization, or 608

make the moves on the board you find to be forced and then think again, as you would do in a tournament game. Obviously, this involves the risk that you have missed resources along the way. 1.Bf5† Kc6! (1...Kc7 2.Be6 transposes to 1...Kc7 below) 2.Be6 Kb5 leads nowhere.

Evgenij Miroshnichenko is a warm and friendly person who really loves chess

1...Kc8! The most resilient defence. 1...Kc7 2.Bb5 leads us back to a generic winning position. 2...c3 3.bxc3 Bxc3 4.a6 c4 5.Bxc4 Be5† 6.Kg2 Bd4

609

7.Kf3! (There are other ways to achieve the same idea. But here the key point is that White wants to avoid 7.h4?? Kd6 8.h5 Ke7 9.h6 Kf6! with a draw.) 7...Kd6 8.Ke4 Ba7 9.h4 White wins. Black will not make it to f6 with the king. 2.Bb5 c3! Black has to get rid of the c-pawns desperately. After 2...Bxb2 3.Bxc4 Be5† 4.Kg2 the win is trivial, as the bishop is unable to quickly change sides and because the white king arriving on b7 would terminate the black bishop’s control of the a7square. 3.bxc3 Bxc3 4.a6 Be5† The key moment in the study. 5.Kg2!! With the amazing idea of Kh3! As we will see, mutual zugzwang is a major theme in this study... 5.Kh3? c4! 6.Bxc4 Bb8! leaves White in zugzwang. Black is threatening to run his king to the kingside, while the bishop slows down White’s king by aiming at the h2-pawn one last time. 7.Bd5 Kc7! Now 8.Kg2 would make it easy for Black’s king to make it to f6, and thus h8; and after 8.Kg4 Kb6! Black makes a draw because of the double threat. 5...c4 Black tries the same idea of line clearance. This time it is not enough. 6.Bxc4 Bb8! 610

6...Bd4 offers less resistance. White plays Kf3 and Ke4, followed by Kf5-g6 with a winning position.

7.Kh3!! This is the paradoxical move. White blocks his own pawn, but the logic is quite simple. White is aiming to go to g6 after which the h-pawn cannot be stopped. After 7.h4 the black king can run to the kingside and arrive in time. 7...Bd6 7...Kd7 8.Kg4 Ke7 9.Kf5 and White wins. 8.Bd5 Now there are other winning moves, but this is the simplest. 8...Kb8 8...Bb8 9.Kg4 and wins. 9.Kg2 White wins. First he will win the bishop on the kingside and afterwards he will queen the a-pawn, which promotes on the right colour square. 1–0 Whereas studies are carefully crafted to give us clarity and preferably only one solution, actual games are filled with viable alternatives all the time. This can make it harder to choose between different moves, where two may look of equal value, but one is not.

611

So, when we play chess, the critical mental attitude is always that of posing problems to your opponent. I managed to do this in the following game against a young Peter Leko. The initial position is certainly defendable for White, but he still has problems to solve. In the game it proved too difficult to do. The game does not look special when you play through it; but the beauty of chess is often what happened in the players’ minds during the game and not what was played on the board. Peter Leko – Boris Gelfand Dortmund 1996

We enter the game at a time when most knight moves are fine for White. As long as the piece is not squandered, he should not be in much danger. 69.Nb4 A natural decision to go into an opposite-coloured bishop ending. I did not really have any way to prevent it. 69...e3! Although you believe a position to be a draw, you should not remove the responsibility of proving it to be so from your opponent. 70.Be1! Not a very difficult move to choose for White, but still we should not underestimate that Leko had a chance to go wrong already here. 612

70.Nxd5? exd2 71.Nc3

71...Bc8! Only this accurate move wins. But after this, it is over. 71...Bc4? 72.c6 Kf4 73.c7 (Black would win after 73.Kc5 Ba6 74.Kd4 Kf3 75.c7 Kf2 76.Ke4 Bc8, where he keeps the f-pawn and wins the knight) 73...Ba6

White draws with this accurate move. 74.Kd5! (74.Kd7 Ke5 and the f-pawn wins easily) 74...f5 (74...Ke3 75.Ke6! is an immediate draw) 75.Kd4 Kf3 76.c8=B Bxc8 77.Kd3 White escapes with a draw. 72.Kc7 613

72.Kd5 Kf4 73.Kd4 Bg4 and Black wins. 72...Ba6 73.Kb6 Bc4 74.c6 Kf4 Black wins. 70...Nxb4 71.Bxb4

71...Bc4 Black could have posed White a lot of practical problems with: 71...Bc8!? 72.Bc3 Kg5

At this point White has to find a number of accurate moves in order to make the draw. 73.c6! e2 614

The best attempt. After 73...f5 74.Kc7 Ba6 75.Kb6 Bc4 76.Kc5 Bf1 77.Kd4 f4 78.Bb2 Kg4 79.Bc1 Black makes no progress. 74.Ke7!! Forcing the pawn to f5. 74...f5 75.Kd6 75.Kd8? Ba6 is hopeless for White. 75...Kf4 White now draws with a beautiful geometric theme.

We have reached one of the two drawing positions for White. This one is based on counterplay. 76.Kc7 Ba6 77.Kb6 Bc4 78.Kc5 Bd3 79.Kb6 Bc4 80.Kc5 Be6 81.Kd6 So, at the same time, the f-pawn has to remain on f5, as we shall see below. But on the other hand, the bishop is limited by the pawn in the same way as it is limited by the absence of a square to the left of the a6-square. 72.Bc3! The only move. 72.c6? Ke4 would see the f-pawn march straight to f1 unhindered. 72.Ke7? On the other hand, this loses to: 72...Kg5 73.c6 f5 74.c7 Ba6 75.Ke6

615

75...Kf4! The key move. The king frequently belongs in front of the passed pawns, in order to take control of the squares that the pawn(s) must cross on the journey its promotion. This was well explained in Small Steps to Giant Improvement by Sam Shankland. For the same reason, 75...f4? would make it too difficult to advance the pawns to the light squares: 76.Ke5 Kg4 (76...f3 77.Bc5 with a blockade on the dark squares) 77.Kd4 Kf3 78.Ke5! The key move. 78...Kg3 79.Be1† Kg4 80.Kd4 and Black has no way to make progress. 76.Kd5 Kf3 77.Ke5 Bc8

Black wins, as White can do nothing about the simple plan of ...e2 and ...Kg2-f1.

616

72...Kg5 After: 72...e2

White draws in a nice way: 73.Ke7 Kg6 74.c6 The logical move. 74.Kd6? Kg5 would make a monster of the f-pawn. 74...f5 75.c7 Ba6 76.Ke6 Kg5 77.Ke5 Kg4 77...f4 78.Be1 Kg4 79.Kd4 is analogous. 78.Be1 Bc8 79.Kd4 Kf3

617

80.Kd3 White is exactly in time. 80...Ba6† 81.Kd2 f4 82.Bh4 Black is too late. He did not manage to get the king to f1, so he cannot improve his position. This is the second drawing position. 73.c6! f5 Another good try was: 73...Ba6!?

White would have to find the draw we have seen earlier. 74.Kd5! 74.Kc5? would lose to 74...e2 75.Kd4 Kf4! 76.Be1 Kf3 and this time Black makes it to f1, winning. 74...f5 White is not in time to set up a fortress, so he has to rely on counterplay. 75.Kc5! 75.Kd4? Kf4 is hopeless. 75...f4 75...Kf4 76.Kb6! we have seen already. 76.Kd4! Kg4 77.Ke4 Black is not able to advance the pawns successfully. We see clearly why it was important to get the king to f3. 74.Ke5? This may look natural, but it is actually the losing move. 618

White had the chance to draw with: 74.Kc5 Bf1 75.Kd4 Kf4! The most testing move. 75...f4 would transpose to 77...f4 right below.

76.Bb2! The bishop needs to go to c1 to be able to control both the black pawns. 76...Kf3 77.Bc1 e2 77...f4 78.c7 Ba6 79.Ke5! Kg3 80.Ke4 Bb7† 81.Ke5 Black is unable to improve his position. 78.Bd2 Bh3 79.c7 f4 80.Be1 Kg2 81.Ke4 f3 82.Ke3 Kf1 83.Bh4 White has managed to achieve the desired fortress. But White also had a direct and elegant way to make the draw. 74.Bd4! Kf4 (74...f4 75.Ke5! Kg4 76.Ke4 with an imminent draw.) 75.Kc5 Bf1 76.Kb6! We saw this theme already. White draws. 74...Ba6

619

75.Bd4 75.Kd4 Kf4 76.c7 The manoeuvre that was a draw previously does not save the game here. The key point is that the f-pawn is still on f5, where the black bishop can defend it while the king goes off to win the bishop. White also succumbs after: 76.Bb2 e2 77.Bc3 Kf3 78.Ke5 Bc8 79.Kd6 Ke4! The simplest and also the most elegant. 80.Kc7 Kd5 Black wins. 76...Kf3 77.Bb4 e2 78.Be1 Kg2 79.Ke3 Kf1 80.Kd2

Black wins easily. All he needs is a waiting move. 80...Bb7! 620

80...f4? would drop the pawn for no reason. 81.Bh4 Bb7 (After 81...f3 82.Ke3 White has achieved the fortress. The time has passed for waiting around...) 82.Kd3! e1=Q 83.Bxe1 Kxe1 84.c8=Q Bxc8 85.Ke4 and White draws. 81.Bh4 Bc8 82.Be1 f4 White is in zugzwang. He needs to play both Bh4 and Kd3, but only after Black has advanced the f-pawn. By timing this correctly, Black can have the bishop on a6 or f5 at the right moment. For example: 83.Bh4 Ba6 As White does not have Kd3-e4, Black wins. 75...e2 76.Bc3 Kg4 77.Be1 Bc8 78.Kd4 Kf3

79.c7 Black is also in time after 79.Kd3 Ba6†! 80.Kd2 Kg2 81.c7 Kf1 82.Bh4 f4 and White is in zugzwang. 79...Kg2 80.Ke3 Kf1 81.Bh4 81.Kd2 f4 transposes to the final annotation on the previous page. Black wins. All Black would have to do is wait after 82.Bh4 and avoid playing 82...f3?? 83.Ke3 Bb7 84.c8=Q with a draw. Instead 82...Ba6 or 82...Bf5 wins. 81...e1=Q† 0–1 In the following game I was on the defensive. Again, we will see that the game would look 621

uninteresting to those who do not fully understand what is going on. While to the reader who takes in the details, a small wonderland of endgame magic is certain to reveal itself. Tomasz Markowski – Boris Gelfand Polanica Zdroj 2000

I spoke about the Rubinstein Memorial already in Positional Decision Making in Chess, page 12. In this game I was under a lot of pressure, but managed to save the draw. Later, Alexander Huzman analysed the game deeply and found that I had been lost and also when my position turned from difficult to desperate. Before we start analysing specific lines, it makes sense to review the key ideas from above and add four more. The first is about widely separated pawns. If there are only two ranks between two passed pawns, say b- and e-pawns, it is frequently possible to set up a fortress for the defending side. But if the pawns are further apart, it rarely is. This is relevant here, as a scenario where Black gives up all his pawns in order to control the passed pawns and free his bishop is not going to work. He will need to fight against one passed pawn only, or to generate counterplay. Then there is the principle of one diagonal. A bishop is more effective if it is able to do all of its jobs on one diagonal. In this game, this diagonal is from b1 to f5. Finally, there is the breakthrough. Here presented in its most basic form. White plays b6 and once taken, pushes the a-pawn rather than recapturing. 56.b5 The natural move. White is threatening to play b5-b6 in order to meet ...axb6 with a5-a6!. After 56.Kb5? Bc3! Black employs the standard defensive method of attacking the pawns from the 622

back. 56...Ke7? As we shall see, this move should have lost the game. For this reason, Black had to find a counterintuitive move: 56...Bb8!!

Of course, I only know about this option because Alex told me it was necessary. 57.b6 The only try. 57.Kc5 Bc7! would allow Black to set up a fortress. If White has a b-pawn, Black will put the king on d8 and White can never promote the b-pawn. 57...axb6 58.a6 Ke5! 58...Ba7? 59.f4! would leave Black entirely paralysed. And lost. 58...Ke7 59.Kb5 Kd6 loses most simply to 60.Kxb6 Kxd5 61.Kb7 and the pawn promotes. 59.Bxf7

623

59...f4 This is not the only drawing move, but it is the clearest one. The idea is simple. White is not allowed to advance his pawn to f5, which would be winning. 60.Kb5 Ba7 61.Kc6 b5! 62.Kb7 Be3 63.a7 Bxa7 64.Kxa7 Kd4 65.Kb6 Ke3 66.Bd5 b4 With a draw. 57.b6 57.Kc5? is met by 57...Kd7! when the black king will, in collaboration with the bishop, be able to block the passed pawn. 57...axb6 58.a6 58.axb6? would make it easy for Black to hold. The king will block the pawn. Even if White managed to get the king to a7, Black would always have ...Bd4, making it impossible for White to progress. 58...Bb8 59.Kb5 Ba7

624

Twenty years ago, I wrote: “Now Black would be forced to give up his bishop for the a-pawn, but it turns out that White will also lose the pawn on f3.” Later on, Alex found a wonderful win, as we shall see below. 60.Kc6? It would be possible to add a lot of sidelines to explain why Alex’s discovery works. But I fear that this may cloud the clarity of the solution. So, let’s take it here with a minimal amount of sidelines. 60.f4! Fixing the pawn on f5. 60...f6 Black needs this pawn for later. 61.Kc6 b5

625

Solving this position could be done by schematic thinking. Black is close to paralysed. If you look at the white bishop and see where it belongs, the rest will come easily. 62.Bg8!! This fantastic move is the beginning of a beautiful manoeuvre that wins the f5-pawn. This is needed in order to win the game. 62.Kb7 Be3 63.Bb3 Kd6 64.Bc2 Ke6 65.a7? draws as in the game. 65...Bxa7 66.Kxa7 Kd5 67.Bxf5 Kd4 68.Bb1 f5! 62...Be3 62...b4 63.Kb7 would not allow the black king to enter the game and thus White wins easily. 63.Bh7 Ke6 64.Bg6! b4 65.Be8 b3 66.Bd7† Ke7 67.Bxf5

626

White wins. He will put the bishop on b1 and the pawn on f5, then win the black bishop. 60...b5 61.Kb7 Be3 62.f4 62.a7 Bxa7 63.Kxa7 Kd6 64.Bxf7 Ke5 is an elementary draw. 62...Kf6 63.a7

Markowski offered a draw, having seen the same variation as I: 63...Bxa7 64.Kxa7 Kg6 65.Bf3 b4 66.Kb6 b3 67.Bd1 b2 68.Bc2 Kh5 69.Bxf5 Kh4 70.Bb1 f5!! and Black will succeed in eliminating the final white pawn. ½–½ The next game is the most complex in this chapter. It draws on the themes discussed already. We are well covered on these themes; but of course, applying the knowledge will take a lot of practising. The biggest difference from the previous example is that the breakthrough is for real this time. In the previous game it was a mini-breakthrough, giving up a single pawn to get a passed pawn. In this example we shall really see the main point of endings with opposite-coloured bishops: that passed pawns are far more important than blocked pawns. Boris Gelfand – Alexei Shirov Bazna 2009

627

We enter the game where the draw seems likely. Black is a pawn down, but the opposite-coloured bishops give him good drawing chances. Still, the weakness of the pawns on f7 and h7 gives White something to play for. 47.f4 Bc2? Shirov missed his chance to defend by actively fighting for the dark squares. 47...e5! This is a thematic move and no doubt Alexei considered it, but must have found it to be unnecessary. Sometimes we do not realize we are in danger until it is too late. I win a lot of games that way. 48.fxe5 Also after 48.f5 Bg4 Black is holding. Now the simplest way for Black to play is 48...Kc6 to establish a fortress. But the most instructive line is connected to a different type of fortress. 48...Bb3!? 49.Kd4 Kb5 50.e6 Without this move White cannot make progress. 50...Bxe6 51.Ke5 Ka6 52.Kf6 Kb7 53.Kg7 Ba2 54.Kxh7 Bb3 55.Kg7 Ka6 56.Kf6 Kb7

628

Black can wait for as long as White wants to shuffle around. But in the end, the black pawns will claim a pawn each and the bishop will take the last one, leaving White with only an unpromotable apawn.

629

With Najer, Shirov and Tomashevsky at the Nutcracker tournament, Moscow 2020

48.Kd4 Kc6 It is too late for alternative paths. Active counterplay with the king does not work: 48...Kb5 49.Ke5 Kc4 50.Kf6 Kd4 51.Kxf7 Kxe4 52.Bd6 Kf5 53.Kg7 Kg4 54.Kxh7 Kxh4 55.a4 Kh5 56.a5 Bd3 57.Kg7 Black is in zugzwang and will lose the g6-pawn, and with it the game. 49.Ke5 Kd7

50.Kf6!? Very concrete play, giving Black chances to make White’s path to victory narrower. A simpler way would have been 50.Bb4 Ke8 51.Bc5 Kd7 52.Kf6 Ke8 53.e5, reaching the ending from the game without allowing Black the extra defensive tries noted below. 50...Ke8 Because of the concrete nature of my last move, Shirov had the chance to go for more active defences. Both are very interesting to analyse; but clearly, from the outcome of the game, we can see that his choice was not a wrong one. 50...Bxe4 would not have been very difficult to refute. White can of course play slowly, rearranging his pieces optimally and only then considering how to break down the kingside defence. But more likely I would have found the direct win. 51.Kxf7 Bc2 52.Bf6 Bd1 53.Kg7 Ke8 54.Kxh7 and White wins. After 54...Bc2 the easiest is just 55.h5! gxh5† 56.g6 and Black will be in zugzwang very quickly, and lost. Also 54...Kf7 55.Kh6 is hopeless:

630

The key point to a lot of these lines was described above. The bishop is not good at performing jobs on two diagonals at once. Either you can deflect it, as the b-pawn was able to do in the Markowski game. Or it will end up in zugzwang, as here. White will, with patience, be allowed to play h4-h5 and later g5-g6. Black can only prevent this by letting the a-pawn run for it. But sooner or later he will have to block its progress and then the defence of the kingside will break down. It was too late for the sacrifice of the e-pawn: 50...e5!? 51.fxe5 Bb3 51...Ke8 52.e6 fxe6 53.e5 is thematically identical to the line above. It makes no difference if the white pawn is on f4 or e5.

Still. it was an interesting practical try – the win is not that simple. 631

52.Bb4 Giving up the a-pawn works only sometimes. The following line is quite typical. 52.a4? Bxa4 53.Kxf7 Bb3† 54.Kf6 Ke8 55.e6 Bc2 56.e5 Bb3 57.Ba3 Bc4 58.e7 Bb3 59.Kg7 Ba2 60.Kxh7 Bb1 61.Kg7 Be4 62.Kf6 Bc2 63.Bb4 Bd3 64.Be1 Bc2 65.e6 Bd3 66.h5 gxh5 67.g6 h4 68.g7 Bh7 69.Bxh4 Bg8 70.Kg6

70...Bh7†! and White cannot make progress. 52...Ke8 53.Bc5 53.Kg7 would be met with 53...Kd7!, threatening ...Ke6, forcing White to repeat with 54.Kf6 and then find the correct path. 53...Kd7 53...Bc2 54.e6 fxe6 55.e5 leads to a lot of lines we have seen already. White takes the h7-pawn and places the king on h6, and Black will be in zugzwang not long thereafter. 54.a4! This is the exception. This time around, sacrificing the a-pawn works. 54...Bxa4 55.Kxf7 Bb3†

632

56.e6†!! This diversionary move is necessary in order to break the fortress. This kind of diversion is a common theme when dealing with fortresses. Here the bishop has jobs on both e6 and g6, on adjacent diagonals. Because the white pawn is exactly on e4, there is no easy transfer between the two diagonals and Black loses a tempo. 56.Kg7 does not throw away the win, but only because 56...Bc2 can be met with 57.Kf7!, repeating the position, as 57.Kxh7 Bxe4 58.Kg7 Ke6 allowed Black to set up a blockade. 56...Bxe6† 56...Kc6 is obviously not going to hold. But I still like to include this line: 57.Kg7 Bxe6 58.Kxh7 Bf7 59.Kg7 Be8 60.Kf8 Bd7 61.Kf7!? Kxc5 62.Kxg6 and White wins. 57.Kg7 Bg4 57...Bb3 58.Kxh7 Bc2 59.Kxg6! and White wins. This time there are zero lines between the passed pawns, as they are connected and different observations apply: the pawns will work together to ensure promotion. 58.Kxh7 Bh5

633

59.Kg7 Again, White will put the bishop on f6 and Black will have to choose between exchanging the gpawn for the white e-pawn, or allowing Kf7 followed by e5-e6-e7-e8=Q. Both clearly hopeless options. 51.e5 Bb3 The position has stabilized. White will win a second pawn, while Black is just waiting. 52.Bb4 Bc2 53.Kg7 Ba4 54.Kxh7 Bc2 55.Kg7 Bb3 56.Kf6 Bc2

57.Bd6 634

During the game I considered a direct breakthrough with multiple pawn sacrifices, starting with: 57.Be7 Bb3 58.f5 exf5 59.e6 Bxe6 60.h5 gxh5 61.g6 fxg6

62.Kxe6 Winning the bishop. Unfortunately, the black f-pawn is going too fast. And I realized that going back to catch it after 62...f4 63.a4 f3 64.Bh4 would allow 64...f2 65.Bxf2 Kd8 when the black king runs all the way to a8, saving the day. Certainly, seeing this line obstructed my chances of finding the win on the next move. 57...Bb3

58.a4? At this point I missed the chance to create a masterpiece. 635

I saw the variation 58.f5!! exf5 59.e6 Bxe6 60.h5 gxh5 61.g6 fxg6 62.Kxe6 Kd8 and I believed it to be a draw once again.

But this time around White is winning after: 63.Kd5!! f4 (63...g5 64.a4 f4 65.a5 f3 66.a6 f2 67.a7 f1=Q 68.a8=Q† with mate) 64.Bxf4 g5 65.Bd6 g4 66.a4 h4 67.a5 g3 68.a6 g2 69.a7 g1=Q 70.a8=Q† Kd7 71.Qb7† Kd8 72.Qc7† Ke8 73.Qe7# But this was not even the only way to win the game. The young Belarusian player Viacheslav Zarubitsky suggested another breakthrough, which leads to an arguably even more beautiful win: 58.Bc5 This is not strictly necessary; but since Black can do nothing but wait, it feels right to place the bishop on an optimal square. 58...Bc2 59.Kg7 Bb3

636

60.a4! White will always have to give up a pawn. This time it is the a-pawn’s turn. 60...Bxa4 61.f5 Bc2 Black ends up in a deadly zugzwang after 61...exf5 62.e6 Bb3 63.exf7† Bxf7 64.Bd6. For example: 64...Kd7 (64...f4 65.Bxf4 Ke7 66.Bd6† Ke6 67.Kf8! and the zugzwang is rather pretty) 65.Kxf7 Kxd6 66.h5! f4 67.h6 f3 68.h7 f2 69.h8=Q f1=Q† 70.Qf6† and White wins. 62.h5!!

A beautiful image of all four white pawns on the same rank. 62...gxh5 62...gxf5 63.g6 fxg6 64.h6 is pretty too.

637

63.g6 fxg6 64.f6! White wins. 58...Bxa4 59.f5 Bc2! I had missed this move. Shirov manages to set up a blockade on the light squares. As mentioned above, the passed pawns are way too near to each other. Setting up a fortress is easy for Black. Alex Baburin mentioned a nice variation that I was hoping would happen in the game59...exf5? 60.e6 Bb3 61.exf7† Bxf7 62.Kg7 and we have returned to the Zarubitsky lines above. But Alexei is not so collaborative. 60.fxe6 fxe6 61.Kxe6 Bb3† 62.Kf6 Bc2 63.e6 Bd3 64.Bg3 Bc2 65.h5 gxh5 66.g6 Bd3 67.Bh4 Kf8 68.e7† Ke8 69.Kg7 Bc2 70.Kh6

70...Bb3! We can see here that White can never win. If it was possible to put the pawn on g7 and king on h8, it would win. But no such option exists. 71.Kxh5 With a draw offer. The fortress after 71.g7 Bg8 72.Kg6 Kd7 is impenetrable. ½–½ As a final theme of opposite-coloured bishops, let’s go in a different direction. It is important to know your basic endgame theory. At various points in this book we have referred to theoretical positions. We even analysed one of them deeply and found an improvement (see 638

Capablanca – Yates on page 100 of Decision Making in Major Piece Endings). As we have talked a lot about themes in opposite-coloured bishop endings, I think the most important theme of all is to know the basics. For example, it is valuable to know this theoretical position: Example

This version of the theoretical position is constructed for this book to give Black only one drawing move. The principle is simple. 1...Bh5! The game ends in a draw. If White plays 2.Ke3, Black draws with 2...Bf7! and White will not benefit from 2.e6† on account of 2...Kd6 and ...Kxd5. Notice that this fortress only holds because Black can shuffle the bishop with ...Bg8-f7-g8. If everything was moved one rank up the board, Black would lose immediately. Rasmus Svane – Erik van den Doel Batumi 2019

639

Knowing this kind of information will make it possible for us to make decisions quickly and accurately. Although most grandmasters, myself included, have made mistakes in positions with properties of theoretical positions, or in the transposition to theoretical positions, this happens rarely when you know the basics. We shall not speculate if Van den Doel knew what he had to aim for and made a careless slip or did not know the ideal set-up and the dangers to avoid. 60...Bh3? There is nothing wrong with the idea of waiting, but Black needed to play 60...Bf5 if he wanted to wait. However, I think it would be much more human to play: 60...b4! Eventually Black will need to play this, preventing the white bishop controlling the d6-square from a3. 61.Kxb4 Bh5! Aiming for ...Bf7. The key point is that after 62.e6† Kd6 63.Kc4 Bf3 64.Ba3† Ke5 Black makes a draw. He has won the fight for the light squares. 61.Kxb5 Black is lost. He is not in time to get to the g8-d5 diagonal. 61...Bg4 62.Kc5 Bf5 62...Bh5 63.e6† Ke7 64.Kd4 and White wins. 63.Kd4 Bg4 64.Ba3

640

64...Bf3 Desperation. Black has no defence against Ke3-f4-g5-f6, when Black cannot prevent the advance of the pawns any longer. 65.e6† Ke8 66.d6 Bc6 67.Kc5 Ba4 68.Bb2 Kd8 69.Bc3 Kc8 70.Bf6 Preparing to manoeuvre the king to e7. Black is defenceless. 70...Be8 71.Kd5 Bb5 72.Ke4 Bc6† 73.Kf5 Bd7 74.Bc3 Be8 75.Kf6 Kd8 76.Bd4 Bb5 77.Kf7 Be8† 78.Kf8 Bd7

641

79.Bf6† Kc8 80.Ke7 1–0 Endgames with opposite-coloured bishops are both interesting and difficult. As you will have noticed, only Pervakov and Svane managed to play with true accuracy in this chapter...

642

Name Index A Adams 130, 131, 219 Afek 199 Alekhine 32 Alex 5, 91, 183, 202, 204, 219, 220, 221, 222, 223, 234, 300, 308, 309, 315 Almasi 162 Anand 15, 110, 137, 197, 297 Aronian 15, 34, 110, 130, 183, 213, 214, 218, 292, 297 Atlas 219 Avrukh 97, 161 B Bacrot 199, 200, 201, 281 Bakh 97 Balashov 202, 205 Baramidze 62 Bartel 95 Benko 267 Botvinnik 15 C Capablanca 32, 168, 169, 170 Carlsen 15, 32, 33, 61, 108, 109, 130, 170, 297 Caruana 31, 133, 134, 159, 161, 162, 164, 167, 179 Cohn 15 D Dai 10 De la Villa 8 Dominguez 109, 281 Dubov 237, 269 Dvoretsky 8, 15, 269 E Eljanov 184 Euwe 32 F Fedoseev 199, 288 Fischer 131, 237 643

G Gasanov 186 Gelfand, Nella 29 Georgiev 68, 98, 179 Giri 161, 215, 237 Glauser 135 Grachev 109, 110, 111, 115 Grischuk 61, 131, 183 H Harikrishna 77, 79, 81, 82, 85, 86, 87, 91, 92 Hort 183 Howell 161 Huzman 5, 91, 183, 202, 204, 219, 220, 221, 222, 223, 234, 300, 308, 309, 315 I Inarkiev 238 Ivanchuk 29, 31, 32, 33, 34, 35, 36, 38, 162, 287 Ivanov 186 J Jakovenko 205, 209, 211, 237 Jumabayev 97, 99, 101, 103, 106 K Kapengut 29 Karjakin 137, 181, 183, 185, 187, 188, 189, 191, 192, 193, 195, 287 Karpov 108, 131, 137, 232, 237 Kasimdzhanov 33 Kasparov 15, 137, 281 Khanin 10 Koblencs 271 Konoval 8 Korchnoi 7, 15, 61, 62, 64, 130, 131, 161 Korobov 97 Kramnik 15, 33, 110, 197, 219, 220, 232, 233, 234 Kveinys 162, 163 L Lautier 297 Leko 80, 98, 303 Lenic 110 644

M Malakhov 172, 288 Mamedyarov 271, 272, 273, 275, 276, 278, 279, 281, 287 Marin 15 Markowski 308, 310, 312 Matocha 129 Miroshnichenko 301, 302 Morokhovsky 265 Morozevich 110 Movsesian 184 Mueller 8 N Najdorf 202, 272 Najer 95, 312 Nakamura 61, 161 Navara 4, 129, 134, 138, 139, 140, 142, 144, 146 Nielsen 62 Nimzowitsch 15 O Onischuk 281 Oparin 237 P Pelletier 61, 62, 64, 65, 67, 69, 71, 72, 74 Pervakov 299, 301, 317 Petrosian 15, 93, 137 Pleshkov 279 Polgar 281 Polugaevsky 62, 97, 129, 135 Ponkratov 199 Ponomariov 297 Psakhis 9, 10 R Rapport 287 Reti 15, 16, 17, 22, 46, 267 Rodshtein 183 Rubinstein 7, 15, 16, 17, 18, 19, 27, 61, 308 Rublevsky 265, 279 S Sadhwani 190 645

Sakaev 287, 288 Shankland 79, 81, 82, 85, 86, 87, 91, 92, 190, 215, 241, 305 Shaw 100 Shereshevsky 8 Shirov 137, 287, 288, 301, 311, 312, 315 Smirin 265 Smyslov 267 So 161 Spassky 237 Stern 110 Svane 316, 317 Svidler 161 T Tal 271, 272 Teichmann 61 Timman 267 Tkachenko 265, 269 Tomashevsky 288, 312 Topalov 130 V Vallejo Pons 63, 97 Van den Doel 316 Van Oosterom 161 Vidit 237, 238, 243, 245, 250, 252, 259, 261 W Wade 7 Wang Yue 39, 42, 43, 45, 284, 285 Z Zarubitsky 315 Zelkind 7 Zinar 267

646

Game Index Robert Wade – Viktor Korchnoi, Buenos Aires 1960 7 Semen Khanin – Changren Dai, Shanghai 2019 10 Richard Reti – Akiba Rubinstein, Gothenburg 1920 15 Vasyl Ivanchuk – Boris Gelfand, Wijk aan Zee 2012 31 Kasimdzhanov – Kramnik, Tromso (ol) 2014 33 Boris Gelfand – Wang Yue, Sochi 2008 39 Boris Gelfand – Yannick Pelletier, Biel 2001 61 P.H Nielsen – Baramidze, Plovdiv 2008 62 Gelfand – Vallejo Pons, Monte Carlo 2004 63 Boris Gelfand – Pentala Harikrishna, Wijk aan Zee 2014 79 Boris Gelfand – Rinat Jumabayev, Moscow 2016 97 Ki. Georgiev – Leko, Cacak 1996 98 Boris Gelfand – Boris Grachev, Moscow 2016 109 Gelfand – Kramnik, Internet 2020 110 Aronian – Anand, Moscow Candidates 2016 110 David Navara – Boris Gelfand, Prague (1) 2006 129 Gelfand – Topalov, Nice (rapid) 2008 130 Sargissian – Adams, Merida 2008 130 Korchnoi – Karpov, Moscow (21) 1974 131 Grischuk – Adams, Calvia 2007 131 Fabiano Caruana – Boris Gelfand, Amsterdam 2010 161 Caruana – Almasi, Reggio Emilia 2009 162 Boris Gelfand – Kiril Georgiev, Plovdiv 2010 179 Boris Gelfand – Sergey Karjakin, Nalchik 2009 183 Eljanov – Movsesian, Sochi 2012 184 Pavel Ponkratov – Etienne Bacrot, Berlin (rapid) 2015 199 Boris Gelfand – Yuri Balashov, Minsk 1986 202 Dmitry Jakovenko – Boris Gelfand, Khanty-Mansiysk 2015 205 Levon Aronian – Boris Gelfand, Nice (blindfold) 2008 213 Boris Gelfand – Vladimir Kramnik, Sanghi Nagar (6) 1994 219 Vidit Gujrathi – Boris Gelfand, Poikovsky 2018 237 Oparin – Dubov, Moscow 2018 237 Inarkiev – Gelfand, Magas (rapid) 2016 238 Boris Gelfand & Mikhail Zinar, Memorial Tournament for Mark Dvoretsky 2017 267 Sergey Tkachenko & Boris Gelfand, Memorial Tournament for Mark Dvoretsky 2017 269 Boris Gelfand – Shakhriyar Mamedyarov, Nice (rapid) 2008 273 Shakhriyar Mamedyarov – Boris Gelfand, Pamplona 2004 281 647

Alexander Onischuk – Leinier Dominguez Perez, Biel 2008 281 Boris Gelfand – Wang Yue, Dagomys 2010 284 Konstantin Sakaev – Boris Gelfand, Jurmala 2015 288 Boris Gelfand – Levon Aronian, Moscow 2009 292 Miroshnichenko & Pervakov, 2016 301 Peter Leko – Boris Gelfand, Dortmund 1996 303 Tomasz Markowski – Boris Gelfand, Polanica Zdroj 2000 308 Boris Gelfand – Alexei Shirov, Bazna 2009 311 Rasmus Svane – Erik van den Doel, Batumi 2019 316

648

Table of Contents Title Page Key to Symbols used & Bibliography 4 Publisher’s Foreword 5 Introduction 7 1 Akiba Showing the Way 13 2 Turning Points 29 3 Passive or Active Defence? 59 4 A Bad Plan is Better than No Plan 77 5 Long Games with Increment 95 6 When is the Right Time to Run? 127 7 Choosing the Right Transformations 159 8 Karjakin 181 9 Stalemate 197 10 Stalemated 235 11 The Relevance of Endgame Studies 265 12 Geometry 279 13 Endings with Opposite Coloured Bishops 299 Name Index 318 Game Index 320

649

3 5 7 9 16 49 111 145 178 241 308 352 385 469 534 562 603 643 647