• Author / Uploaded
• forbidden_city7414

• Categories
• Tendencia del mercado
• Burbuja económica
• Contratos de futuros
• Petróleo
• Inversiones

Citation preview

Trade Like a Casino Find Your Edge, Manage Risk, and Win Like the House

RICHARD L. WEISSMAN

@

WILEY

John Wiley &Sons, Inc.

xl

Preface

Acknowledgments

PART I

xvii

The Casino Paradigm

CHAPTER I Developing Positive Expectancy Models Why Technical Analysis Helps The Inefficient Market IfItFeels Good, Don’t Do It “Just Make the Money” Final Thoughts CHAPTER 2 Price Risk Management Methodologies

23

One Sure Thing

23

Base ofPyramid

26

Middle ofPyramid

37

Apex ofPyramid

47

Pros and Cons ofthe Risk Management Pyramid

49

Putting ItAll Together: ACase Study Final Thoughts

49

CHAPTER 3 Maintaining linwavering Discipline

53 53

Defining Discipline Discipline and the Positive Expectancy Model Types ofTraders Discipline and Price Risk Management

SI

55 6] 64 VII

\III

CONTENTS

Patience and Discipline

67

Final Thoughts

70

I'\li'l‘ ll

Trading Tools and Techniques

7|

W

(le “'1‘”! 4 (Zatpllalllnlng on the Cyclical \ulurc of \olalilll) The Only Constant Defining Volatility with Technical Indicators

73 73 76

Building Positive Expectancy Models with Volatility Indicators Final Thoughts (.‘ll “'1‘”! 3 Trading the “aria-ts and \M the “on”

Ten Thousand Dollars is a Lot of Money! Baby Needs a New Pair of Shoes

89 94

93 95 99

Trading with Scared Money

l00

Time Is Money

l0! 104

Final Thoughts (.‘llti'l‘lllt b‘ \linlnlizlng 'I‘rader Regret

")3

The Softer Side of Discipline

105

Issues for Trend Followers

i06

Issues for Mean Reversion Traders

H3

Final Thoughts

123

(Ill \PTH! 7 Timeframe \nalpsis

I23

Traditional Timeframe Analysis

iZS

Timeframe Confirmation Trading

131

Timeframe Divergence Trading

131

Final Thoughts

i42

(IIHP'I‘IIR 8 lion lo li'it' 'l‘radlng \Iodt‘ls

I43

Mechanical Trading Systems

I43

Nonmechanical Models

l62

Equity Trading Models

170

Final Thoughts

i75

ix

Contents

CH-\PTER 9 \nlit‘ipaling the Signal Always Trade Value. Never Trade Price Support (and Resistance) Were Made to Be Broken

I77

Don’t Anticipate, Just Participate

18]

Final Thoughts

I87

P‘RT III

Trader Psychology

I77 179

“39

CHAPTER II) Transcending Common Trading Pitfalls

I!”

Obstacle Makers to Growth as a Trader

19] 195

Final Thoughts

202

Characteristics ofMarket Behavior

CHAPTER II ;\nal\.1.ing Performance ADue Diligence Questionnaire

205 205 225 236

Trading Journal Final Thoughts CHAPTER i2 Becoming an Ewn-Tempcred Trader

24I

The “IDon’t Care” Guy

24]

The Master Trader

244

Reprogramming the Trader

247

Flexibility and Creativity

248

Meditation

250

Visualization

25]

Somatic Exercises

253

Final Thoughts

254

r\otets Bibliography “mat the .\uthor lnde.\'

257 263

263 266

Preface You cannot beat a roulette table unless you steal moneyfrom it. —Albert Einstein

A year ago, Iwas talking with astruggling trader about the profession of speculative tradin'g. He asked a question that ultimately culminated 1n‘thepublication ofthis book. That question was “Can someone really eam alivingas aspeculator?” This person was putting his life’s savm'gs on the line every day and yet did not know for certain whether anyone could actually earn a liv1n'g through speculative trading. Then, on thm'km’g back tomy own start 111'tlu’s business, itoccurred tome that Ihad done thesame thing. Ifyou have picked upthisbook and have been asking yourself that same question, there isgood news and bad news. The good news isthatthe answer tothe question is“Yes.” Yes,professional speculative trading isavalid career path. Yes, not only can itbedone, but ithas been, andcontinues tobeaccomplished bymany professional' traders. Itis not amatter ofluck orchance. The bad news isthat itisone ofthe. most difficult careers known to humankind. Itisdifficult because itrequires us to consistently dothat which ispsychologically uncomfortable an'd unnatural (werevisit whytrading isso difficult ingreat detail throughout the course ofthis book). Sohow do we transform the dicey game ofspeculative trading into a valid career path? We donot start from scratch. No need to reinvent the wheel. Noneed forluck, chance, oreven prayers. Instead, what isrequired istheadaptation ofan existing successful business model tothe career of speculation. That model isthe casino paradigm.1 How do casinos make. money? Although each and every spin ofa roulette wheel israndom, the casm'o remain's unconcerned because. probability isintheir favor. Intrading, we call this the development ofpositive expectancy trading models. Positive expectancy means that after deducting for liquidity risk—for ex— ample, the ns'k ofprice differences between our model’s hypothetical entryor exit price and the actual entry or exit price—and commissions, our model isprofitable. xl

\Il

PREFACE

But what ifsome multibillionaire walks into the casm'o with acashier’s check forabillion dollarsm"? She finds the cashier quite happy to change her check into chips . . . no questions asked. But when she walks her wheelbarrow ofchips over to the roulette wheel an'd tells the croupier, “Put it all on red.“ she ispolitely told that there isa maximum table limit bet srze' of $l0.00() per spin ofthe roulette wheel. Why does the casino need table Inn'-

its ifprobability is.skewed intheir favor? Because they know that despite the odds being in their favor, on any particular spin of the wheel it could come up red. ar'td ifitdid, our multibillionaire would own their casino. By using table limits, they force the player to limit her bet size. thereby ensuringthat aS' they keep playing, the casrn'o’s probability edge will eventually swallow up the entire billion dollars. In speculative tradtn'g we call table limits price risk managmnmzt. The final' prerequisite tothecasino model was actually un'plicitly stated in both ofthe preceding paragraphs. The specific sentence that ad'dressed, this third prerequt'site most clearly was “...the casino remains unconcemed because they have probability intheir favor." Casino owners do not become despondent or close the casino when players vn'n. Instead, they continue playing the probabilities .u'td manag'ing the ns‘k. They adhere to this par'adignt 24hours aday, seven days aweek, an'd 365 daysayear'. They never abandon the paradigm irrestx‘ctive ofhow good or how bad their reStilts ar-e on any given day. week, or month. Intrading. we. call unwavenng‘ adherence to positive expectancy trading models and price n'sk management Imder discipline. Of course, the model for successful speculative trading is' more complex than the casu'io paradigm and throughout this book we explore these van'ous complexities in‘ great detail. Nevertheless. now the book’s title makes more sense. Successful traders can walk under ladders, have tradingaccounts ending in the number 13. you name it. . .it makes absolutely no difference because successful speculation has nothing whatsoever to dowith luck. Luck iswhat thegamblers hopefor. Bycontrast, professional speculators consistently playtheprobabilities andmanage the risk. This book progresses in' a linear fashion from basic, rudimentary concepts to those ofgreater complexity. Chapter 1explores the casrn'o paradigm of trading with respect to the development of positive expectancy models in'exhaustive detail. F1rs't, we look atwhy techm'cal analysrs'helpsinthe development ofpositive expectancy trading models as well as the flaws in' fundamental analysis as a standalone methodology for the development ofpositive expectancy models. Then we examm'e the linu'tations of technical' ar'talysis and how fundamental analysis can be used to minin'uz‘e these limitations. Chapter 2exan'u'nes thecasino para-digm oftrading as itrelates toprice ns'k management. This chapter specifically introduces the reader to what I

Preface

XIII

call the risk management pyramid. The base ofthe risk management pyramid m'cludes traditional tools ofprice risk management such as stop loss placement and volumetric position sizing. Within the middle tier ofthe pyramid are tools used bythe portfolio school ofrisk management, valueat-rls'k and stress testing. At the pyramids apex isqualitative analysis by experienced risk managers that Icall management discretion. Chapter 3concludes our introduction to the casm'o paradigm with an m‘—depth exposition oftrader discipline. Itbegins bydefiru'ng discipline as itrelates to speculative trading and explaining why adherence to a disci— plin'ed approach is difficult. Then we see how discipline relates to developm‘g, 1m'plement1n‘g, and adhering to positive expectancy tradm’g models andprice n'sk management. Next isan examination ofhow the lack ofdisciplm'ecan undermm'e apositive expectancy tradingmodel. Nomatter how robust amodel 1's,there are times when the odds do not favor that model’s employment. Standing aside during such periods requires patience and disciplm'e, specrfi'cally the discipline not totrade until the market again displays the kind ofbehavior m'which the odds are inour favor. The chapter concludes bylookm’g atvarious types ofmarket action that traders can exploit, as well as pitfalls to avoid inattempting to capitalize upon that type ofaction. Chapter 4explores the best-kept secret intrading, the cyclical nature ofvolatih'ty. No one can guarantee whether markets will trend, revert to the mean, go up, or go down. The only guarantee is that they will cycle from low volatility to high volatility and vice versa. This chapter examinesallofthe commonly employed tools formeasuring volatility aswell as showing how to incorporate them into acomprehensive variety ofpositive expectancy trading models. Chapter 5looks at a problem that can undermine even the most ro— bust ofpositive expectancy trading models. Icall ittrading the money. Inexpen'enced traders are always thinking about the money. In2008, when crude oil dropped from $147 a barrel to $135 a barrel, that was a $12, or $12,000, move per contract. Traders who were thinkingabout the money took profits and then watched from the sidelines as the market moved another $100,000 per contract over the course ofacouple ofmonths. Trading themarket and not the money means forcing the dynamics oftheprice action to dictate decisions to close out trades m‘stead ofmaking emotional decisions based on how much money you are making orrisking. Chapter 6focuses on different techniques to minimize emotions ofregret. The greatest feelings of regret occur when we allow significant unrealized profits to turn into sigmfi'cant realized losses. We minimize these feelings ofregret bynot allowing unrealized profits to turn into realized losses and bytaking partial profits at logical technical support or resis— tance levels. The other major source ofregret forthetrader istaking small

\rlv

PREFACE

profits only to see the market make huge moves. We minimize this feelingofregret bytaking partial' profits at logical support or resistai'ice levels and allowing the remainder ofthe position to be held through the use of trailing stops. (.‘hapter 7discusses the, importance oftimeframe analysis. First, we look at the traditional' approach to this analysis, namely, the simultaneous exmnination ofmultiple timeframes tobetter understand the. market's trend, as well as multiple levels oftechnical support an'd resistance. Next isan introduction to one ofthe. most valuable tools used byprofessional' speculators, which Icall tt‘mefmme divergence. Timeframe divergence. occurs when shorter-term timeframes are out ofsync with longer-term timefrai'nes, and itenables traders to enjoy a low risk—high reward entry point in the direction ofthe longer-term trend. This chapter helps readers use technical' analysis so they can' better identify these trading opportunities. (,‘hapter 8 examines a wide array of positive expectancy trendfollowing and mean reversion trading models. Ital'so explores hybrid models that combine mean' reversion technical indicators with longer-term trend-following tools, so that traders can enjoy low n'sk—high reward entry points taken inthedirection ofthe longer-term trend. Chapter 9introduces the reader to another psychological trap that can derail positive expectancy trading models. Icall itanticipating thesignal. Anticipating .the signal occurs because traders tend to focus on selling at a high price—or buying at a low price——as opposed to selling only after there is evidence that a market top isin place (orbuying only after there. isevidence that. abottom ism'place). Incontrast to anticipating thesignal, this chapter shows thebenefits ofwaiting forevidence that itistime tosell or time to buy and explores some sun'ple technical tools to help traders avoid this costly mistake. Chapter 10 examines common tradin‘g pitfalls and how to transcend them. Byexploring characteristics ofmarket behavior, the chapter offers traders techniques to aid m' systematically stripping away delusional beliefs that can derail or impede performance. Then it explores van'ous emotional states that can subvert 0r lum"t success in trading, and helps speculators develop a wide array of techniques to overcome van'ous irrational trading biases. Chapter 11 offers a wide variety oftechniques for analyzing and miproving trader performance. The chapter begins with a comprehensive questionnaire to aid in highlighting strengths and weaknesses ofspeculators inareas such as trading edge identm“cation, perfomiance record analysis, trading methodologies, risk management methodologies, and trade execution considerations as well as research dl'ld development. Then I present one ofthe most powerful and underused tools forimproving trader performance, thecreation and maintenance ofa tradingjoumal'.

7"? preface

XV

Chapter 1.3" explores the psychological mindset required to succeed with a positive expectancy model. Ical'l iteven-mimicdrwss. .S‘uccessful traders shouldn't care about the result ofany specific trade liccaus-w they cons15t'ently employ positive expectancy models combined with robust risk management techniques. Since that isthe cas'c, ifthey do care, then they (a) haven't done enough research to be certain that it. is a positive expectancy trading' methodology. (b)we not mzmaging the risk, (c)are letung'pren'ous negative tradm'g experiences sabotage their edge, or (d) are. addicted tothegamblers mentality ofneeding to win as opposed to knowmg'thattheyWill'succeed. In this' final chapter. we look at van'ous tools zuid techniques to get traders ofl’the emotional euphoria—despondency roller coaster. Richar'd L. Weissman

E l l g l

Acknowledgments

I believe that allofan individual‘s accomplishments are integrally lm'ked to the totality ofhis or her life experiences. As such, all acknowledgments necessarily fall short oftheir goal. Having said this, Iwould like tothank family, friends, and colleagues for then’ support and encouragement inthewriting ofthis book. Inaddition, Iwould like to thank my wife, Pamela Nations-Weis'sman; Richard Hom, who continues to act as an unparalleled soundm'g board for many ofmy trading ideas; my friends and colleagues Dr. Alexander Elder, Konchog Tharchin, and James W. Shelton III; Stan Yabroff and Doug Janson atCQG; Stephen Gloyd, J.Scott Susich, Dominick Chirichella, and Salvatore Umek of the Energy Management Institute, who are tremendous advocates and supporters ofmywork; and myeditorial team atJohn Wiley&Sons, Kevin Commm's and MegFreebom. Ialso wish to acknowledge my indebtedness to all the authors h'sted 1n'this book’s reference list. Ifthis book has added anythm'g to the fields oftradm'g system development, trader psychology, risk management, and technical analysis, itisadirect result oftheir work. Finally, Iwould liketo acknowledge the depth ofmy gratitude to my friend and teacher, Drupon Thiley Ningpo Rm’poche, and his teacher, His Holm’ess Drikung Kyabgon Chetsang Rinpoche, whose works have m'splr'ed and transformed my work and mylife.

E

if

,1.3

R.L.W.

.“fl

The (Maine Paradigm

:0evel0ping f Positive Expectancy Models In the ease of(m eurt/uproke hitting Las Vegas, be sure In go .slr'u,igh,l, to Um keno {Imnge Nothing ever gels MI, ’III’,I"I’,. —-An anonymous casino boss

There are some prerequisite elements that are common to all suceessl'ul trading programs. This and the next two chapters that follow will cover such elements: This chapter is on developing positive expt't'lam'y trading models, the second on implementing robust risk manageim-nt methmlologies, and the third on trwler discipline. Let’s getstarted.

MWIIY TECHNICAL ANALYSIS HELPS 'l'erlmieal analysis is perhaps the single most valuable tool used in the development of positive expeetmicy trading models. According to techniclans, the reason that, teehnieal analysis helps in the development ofsuch models isdue to the notion that “pn’ee has" memory.” What, does this mean'? Itmeans that when crude oil traded at $40a barrel in 1990, this linear, horizontal resistam-e area would ag'ain w-t as resistance when retested in2003 (see lt'igure l.l ).'l‘his reality drives economists crazy because, according toeeonoinie theory, itmakes absolutely no sense forcrude oilto sell offat $40 a barrel in 2003, since the purchasing power ofthe US. dollar in2003 isdifferent from its purchasing power in 1990. Nevertheless, according to teelmical analysis, the .selloff at M0 a barrel in 2003 made perfect sense

3

THE CASINO PARADIGM

$40Ibbl resistance

FIGURE 1.I Rolling Front-Month Quarterly CME Group Crude Oil Futures Showing $40 a Barrel Horizontal Resistance

Source: CQG, Inc. ©2010. All rights reserved worldwide.

because price has memory. Price has memory means that traders experienced pain, pleasure, and regret associated with the linear price level of $40 abarrel. Let’s look atthis ingreater detail. Price has memory because back in1990 agroup oftraders bought oilat $40 a barrel. They had all sorts ofreasons fortheir purchase: Saddam Hussein had invaded Kuwait, global demand for oil and products was strong, and so on. However, ifthese buyers were honest with themselves, as oil prices tumbled, all these reasons evaporated and were replaced with one thought and one thought only—usually expressed inprayer form—~“Please, God, let itgo back to $40 a barrel and Iswear I’ll never trade crude oil again.” When itdoes rally back to$40 a barrel, that linear price represents the termination ofthe painful experience ofloss for such traders. And so they create selling pressure at this linear, $40-a—barrel price level. There isanother group oftraders that are also interested incrude oil at the linear price level of$40abarrel. This isthegroup that sold futures contracts to the first group. Because they sold the top ofthe resistance area, no matter where they covered their short positions, they took profits and so have a pleasurable experience associated with the linear $40-a-barrel

Developing Positive Expectancy Models

5

price. Consequently, when crude oil againrises to $40 abarrel in'2003, they seek a repetition ofthat pleasurable experience associated with the lm'ear $40-a—barrel price and they, too, create selling pressure. But ofcourse. most traders neither sold nor bought at $40 abarrel in 1990. Instead, they stood on the sidelines regretting that they missed the sale of the decade. The beauty of the markets isthat if'you wait around long enough, eventually you will probably gettosee thesame prices twice. When this happened in2003, this third and largest group oftraders got to minimize the. painf'ul feeling ofregret byselling the linear resistance level price of$40 a barrel. This is why technical analysis helps, because most humans seek to avoid pam’ and seek pleasure ins'tead. Inthe markets, pam’ and pleasure play themselves out at price levels such as $40 a barrel in crude oil. However, inApril 2011, when Iwrote these words, crude oilwas tradm‘g at $108 a barrel. Obviously, somethin‘g changed. In fact, th1n'gs constantly change inthe markets. As Chapter 4shows ingreat detail, change and the cyclical nature ofprice action are among the few things that are in fact guaranteed inthe markets. What changed was that during 2004, crude oilexperienced aphenomenon known as aparadigm shift. Aparadigm shift isan m'temiediate to long-term shift' inthe perception ofan asset’s value. Many fundamental factors led to this paradigm shift. The most im'portant one perhaps was unprecedented demand for hydrocarbons from China, India, and other emergln'g market economies. The interesting part about technical analysis, and more spec1fi‘cally about price having memory, was that when this paradigm shift occurred, we did not simply leave $40 abarrel on theash heaps ofmarket history. Instead, dun‘ng May 2004 when oil broke above $40 abarrel, the psychology ofthe market shifted and everyone who sold crude oil at $40 abarrel was wrong and everyone who bought at $40 a barrel was right. Consequently, when in December 2004 the market retested $40 a barrel, those who sold had achance to alleviate the. painful experience ofloss, those who bought $40 a barrel in May had a chance to repeat the pleasurable experience of profit, and those who regretted missing the opportunity to buy at $40 a barrel had the chance to minimize that feeling ofregret bybuying at that price. The old resistance priceof$40abarrel had become themarket’s new support level (see, Figure 1.2). Next, fast-forward the clocks to September 15,2008. Lehman Brothers isin bankruptcy, credit. markets are frozen, and itisobvious that crude oil—along with almost every other physical commodity—is in the throes ofa bear market. In fact, crude 011' prices have dropped from $147.27 a barrel to $95.71 a barrel. On that day, as on various prior and subsequent days when teaching tradln'g courses to speculators and hedgers, someone asked, “Where doyou think the bottom isincrude 011'?” Myanswer seemed

THE CASlNO PARADIGM

Para-p ahl'ft

“Olbbl Mama support

FlGliRl‘.‘ l.2 Rolling Front—Month Weekly CME Group Crude Oil Futures Showing Breakout Above and Retest of $40 a Barrel as Support Source: CQC, Inc. ©2010. All rights reserved worldwide.

incredible tothe roomful ofyoung energy traders: “Forty dollars a barrel." Ofcourse my prediction proved too optimistic as crude oil eventually bottomed out at $32.48 abarr'el (see Figure 1.3). Nevertheless, the mar‘ket had proven over the course ofthe decades that $40 abarrel was" a level atwhich price had and continues to have memory inthe. crude oil market.

'I'IIE INEFFICIENT MARKET Incredibly, academics and economists with strong science backgrounds have put forth a theory ofan efficient market without any statistical evidence ofmarket efficiency, despite much evidence to the contrary. The markets have always been m'efficient, have always cycled from panic to bubble to panic again, and will always continue to doso. In fact, as stated earlier, this cyclical nature ofmarket behavior isone ofthe few tlu'ngs we as traders can actually count on. Ludicrous as itsounds, accordm'g to efficient market hypothesis there can be no such thm'g as a bubble because markets are always trading at

Developing Positive Expectancy Mode/s

$40lbbl as on port

durvng Great access-on

FIGURE [.3 Rolling Front-Month Monthly CME Group Crude Oil Futures Showing $40 aBarrel as Support during Great Recession Source: CQG, Inc. ©2010. All rights reserved worldwide.

thelr‘ correct, or efficient, price levels. In other words, according to these theorists, atulip m’ Holland that was correctly priced at 2,500 gull'ders on February 2, 1637, was also correctly priced at 2guil‘ders on February 3, 1637.1 Icall this an example ofthe “Napoleon Analogy.” The Napoleon Analogy occurs when we enter amental m'stitution m' which one charismatic patient has thoroughly convm'ced him'self as well as other patients thatheisNapoleon. Nomatter how many psychiatm‘ts struggleto assure these patients that he isnot Napoleon, neither the deluded patient nor his loyal admirers can be convm'ced. One day, our delusional patient escapes from the mental institution and discovers not asm'gle soul who believes him' to be Napoleon. This ofcourse isbecause he never was Napoleon. He was merely deluded and had conv1n‘ced others ofhis delusional belief. Perhaps he will never beconvm'ced that heisnot Napoleon. Perhaps there are still people who remain' convm'ced that synthetic Collat— eralized Debt Obligations (CD05) on pools ofsubprun'e mortgages circa 2005 should still be trading at par value. Despite their conw‘ction to the contrary, those synthetic CDOs are still worthless. Furthermore, much like our deluded, Napoleon—impersonating mental patient, despite the

THE CASINO PARADIGM

temporary delusional valuation ofthese synthetic. CDOs at par byvarious financial institutions during the 2005. housing mar-ket bubble, the synthetic CD05 were, infact, always worthless. Nevertheless, just because the minority are delusional and prices are temporarily out ofsync with value, this book isfor traders, not long-term investors, an'd traders must wait for evidence that. our mental patient has escaped from thehospital before trading against irrational, bubble-induced price levels. We wait forevidence inthe form oflower prices because irrationally priced markets tend tobecome. even more. irrationally priced—this is the nature ofan inefficient, fat-tailed market—before crashing, and no one can know where the top isuntil after that tophas been proved through the printing oflower prices. As John Maynard Keynes said, “Markets can remain irrational a lot longer than you or Ican remain solvent,” or as I like to say, “Don’t anticipate, just participate.” Wait for the evidence of a top to start selling and wait for evidence of a bottom to start buying. The history ofmarkets is littered with graves ofthose who were prematurely right. Being right over the long run isfatal for traders. Speculators need to be. right on the markets inthe right season. For example, around January 2009, SemGroup started shorting crude oil around $100 a barrel. They correctly surmised that oil prices were unsustainable at such levels and were out of sync with the asset’s long-term value. Nevertheless, on July 16,2008, SemGroup announced that theyhad “liquidity problems” and sold their CME Group trading account to Barclays. On December 12,2008, January 2009 crude oil futures on the CME Group bottomed at$32.48 (see Figure 1.4). Ofcourse, this was no help to SemGroup since they had filed for bankruptcy on July22,2008.2 But why does the inefficiency ofmarkets matter to us as traders? It is this inefficiency that allows us to develop positive expectancy trading models. This inefficient behavior ofmarkets leads towhat statisticians call a leptokurtic as opposed to a normal—distribution of asset prices (see Figure 1.5). This means that prices display a greater propensity toward mean reversion than would occur ifmarkets were efficient, and, when they are not in this mean reverting mode, they have a greater propensity totrending action (statisticians call this propensity for trending action the fattail ofthe distribution). Itis because markets display this leptokurtic price distribution that positive expectancy trading models tend to fall into two categories:

I. Countertrend models that capitalize on the market’s propensity toward reversion to the mean.

2. Trend-following models that take advantage ofthose times when markets undergo afattail event.

Developing Positive Expectancy Models

sell: My": (runny FSMroup account to hrclays

SuGroup sluts shorting

CrWo OII drops lo $32.48Ibbl

FIGURE L4 Rolling Front-Month Weekly CME Group Crude Oil Futures Showing SemGroup‘s Failure Despite Correct Assessment ofAsset's Long-Term Value Source: CQC, Inc. ©2010. All rights reserved worldwide.

It,isno coincidence that two ofthe three major types oftechnical indicators are oscillators that signal when markets are—at least temporarily— overbought or oversold and trend-followm‘g indicators like moving aver— ages, moving average convergence divergence, Ichimoku clouds, and so on, which signal when markets are displaying bullish— or beans'h—trendm'g behavior. You might be asking yourself, “If markets can do only two thm‘gs— trend or trade inarange—why are there three major categories oftechnical indicators?” The third major category is the volatility indicators and they Normal

Leptokunic

l»1

H

FIGURE l.5 Leptokurtic versus Normal Distribution ofAsset Prices Source: www.risk.glossary.com.

‘u. w v7 -\

I0

THE CASINO PARADiCM

clue us into when markets shift‘ from their memi mwrting mode to trending action and Vice versa In fact. it is this thu‘t‘i category of indicators that proves most useful in the development of positive e.\'pect;uicy tmd~ u'igmodels and Ihave conseq:uently devoted Chapter 4to theyzm'ous types ofvolatility indicators. how they can beused. midtheir‘ lu'nitations.

IF IT FEELS GOOD, DON’T DO IT Well, speculative trading sounds simple enough. Markets can do only two things, either trade inarange or trend, and volatility indicators can be used‘ to clue you mtowhich km'd ofbehan'or the market iscurrently exlu'biting. Why then do almost all speculators lose money"? They lose because successful speculation requires that we consistently do that which ispsycho— logically uncomfortable and unnaturz . Why are mean reversion trading models psychologitn'illy uncomfortable to implement? In Plgtrire 1.6 (see: Figure 1.6) we see that on Friday. March 6. 2009. the E-Mini MP ot’lO futures are not only in a clemly de fined bear trend. but that they have once ag'au"i made new contract lows.

wthOI below trondlino

FlGl RE 1.6 March 2009 E-Mini S&P 500 Futures Contract Makes New Lows WI-th Relative Strength lndex Oscillator at Oversold Levels Source' CQC. Inc. @2010. All rights reserved worldwide.

Developing Positive Expectancy Mode/s What the chart cannot Show ishow overwhelmingly bean'sh market sentiment was on that day. On Fridays, after finishing mymarket analysis forthe day, Iturn offthecomputer and turn on the financial news, as itisusually entertainin"g. Ontlu's particular Friday, the market hadjustclosed and they were interviewm'g two market pundits. They will typically have one interviewee advocating the bear argument while their counterpart is bullish. Our first analyst’s forecast was 5,000 on the Dow Jones Industrial Average and 500 m' the S&P 500 Index. As soon as the words “five hundred” left his' lips. the other interrupted, “You are out ofyour mind.” Ithought, “Ah, here’s the bullish argument." The other analyst then proceeded to berate our beans‘h forecaster bytelling him he was out ofhis mind because the Dow was gom‘g to 2,000 and the S&P 500 to 200. Iglanced atthe bottom ofthe screen just to make certain' that Ihad not lost my nu'nd . . . no, the E-Mm‘i S&P futures had in fact closed at 687.75. Next thought, “When the market isat 687.75 and the bullish analyst is calling for itto drop to 500, this'hasgotto be the bottom.” Sure enough, the 2009 stock market bottom occurred on Friday, March 6,2009 (seeFigure 1.7).The trader usm'gamean reversion model has to consrs'tently buy 111'to that type ofoverwhelmingly

Closing below low! Dollars”! Blfld_

FlGl RR 1.7 Rolling Front-Month Weekly E-Mini S&P 500 Futures Contract Show—

ing Close Below Lower Bollinger Band and Oversold Reading on Relative Strength Index Source: CQG. Inc. ©2010. All rights reserved worldwide.

THE CASINO PARADIGM

12

bearish sentiment or sell intoa 16303-era tulip—or 2005 housing—bubble like bullis'h environment. Executing atrend-following model is even more psychologically challenging. The mar'ket breaks to 1068, all-time new highs. Itell you that the prudent play isto buythese all-time. new highs. You glance at achart'. an'd notice that only 12weeks ago itwas trading at 775. You place a limit order to buy 775, figuring you will buy cheaper, experience less risk, and enjoy more. reward. Byplacing the order at 775 you are trading the asset’s pn'ce, irrespective ofvalue (formore details on trading price. irrespective ofvalue. see Chapters 5and 10).On November 3,1982, the Dow Jones Industrial Average hitan al'l-time new high of1068.1 (see Figure 1.8). Since that time we experienced market crashes, the bursting of the dot—com bubble, terroristattacks, the worst credit crisrs' since the 1930s, and the Great Recession, and as ofthe.writingofthisbook in2011, we still have not traded anywhere close to 1068 (see Figure 1.9). For both mean reversion as well as" trend-following traders, the profitable trade is the one that is almost impossible to execute. Or as Ilike to say, “If it feels good, don’t do it." Ifit feels awful, like a guaranteed loss—more often than' an'yone could imagine—that is the profitable trade.

Do- broaks old high.

Willi

will.1‘

Wmm” hillullluillllllfiW will"

11 ll

FIGI RE LB Quarterly Cash Dow Jones Industrial Average Chart Breaks to AllTlme New Highs in 1982

Source: CQC, Inc. ©2010. All rights reserved worldwide.

Developing Positive Expectancy Mode/s

13

FIGURE 1.9 Yearly Cash DowJones Industrial Average Chart from 1982 Break of Old Highs to July 2010

Source: CQG, Inc. ©2010. All rights reserved worldwide.

If,on theother hand, the trade feels like easy money. . . run the other way. Weare allhuman bein'gs, experiencing greed and fearatthesame moment; ifitfeels easy for us, itfeels easy for everyone else and isalmost guaran— teed to be a losing proposition. If, bycontrast, itfeels almost impossible forus, then few others can take the trade, and bydom‘g that which ispsychologically uncomfortable—by taking the difficult trade—you make the money beinglost bythe other 90percent ofallspeculators. Although the reader now knows why90percent ofall speculators fail, we can learn more about how to succeed and how to develop positive expectancy models as well as risk management byexamining the psychologicalbiases that lead to failure forthe majority ofspeculators. In 1979, two social scientists, Daniel Kahneman and Amos Wersky, developed an al— temative to the dominant efficient market hypothesis ofmarket behavior. As opposed to assuming rationality ofmarket participants and our preference for choices with the greatest risk—adjusted utility, Kahneman and Tversky posed various questions regardln'g risk and reward. The results oftheir research became known as Prospect Theory and the Reflection Effect. Their' work proved that people were 1rr'ational and biased in their' decision-making processes.

THE CASINO PARADIGM

l4

They asked people .to make speclfi'c choices between various altema. tives. Kahneman and 'I‘versky first had participants choose between one of the. two gambles, or prospects: Gamble A:A100 percent. chance oflosm'g $3.000. Gamble B: An 80 percent. chance of losm’g $4,000, and a 20 percent chance oflosm'g nothing. Next, you must choose between: Gamble C:A100 percent chance ofreceivm‘g $3,000. Gamble D:An 80percent chance ofreceiving $4,000, and a 20percent chance ofreceiving nothing. Kahneman and 'IVersky found that of the first groupln'g, 92 percent chose B.0fthe. second grouping, 20percent ofpeople chose D. What the reflection effect proved was that people were risk-averse regarding choices involvu’ig prospects of gains and risk-seeking over prospects involving losses.3 This means that Virt'ually all human beings— including successful speculative traders—are wired the same way: We are all programmed to take smal'l profits and large losses (see Figure 1.10). What then separates successful traders from the rest of the speculative community? Successful traders have developed and employ rule-based positive expectancy models that force them to overcome then” um'ate bias toward small profits and large losses. They have learned to accept small losses quickly and to let large profits grow larger. Or, as Ilike to tell my students, “You need to contln‘uously ask yourself, ‘How can Ireduce the risk? How can Iincrease the reward?” The positive expectancy models

value

outcome

fmsses Gains

Reference point FIGI'RE L“)

Prospect Theory

7' Developing Positive Expectancy Models

I3

force us to do that which ispsytthologically Inmatural and unctnnl‘omible. They force us to sueeeed despite our biases and they do so byexploiting the irrationality an'd biases ofother market participzmts,

“JUST MAKE THE MONEY” Traders will often ask me why Ithink a particulm' market isgoing to go down, why Iam long some other market when the inventory numbers just came out decidedly bearish, and so on. [H have the time, imight give them areason or two, although Iwill more often simply respond with, “lio you wan't to understan'd all the intricate reasons behind the moves or do you want to make the money". Nobody can know all the reasons. li‘orget the reasons, just make the money." The problem or linu't'ation with l'umlamental analysis —-»~as well as the problem with (.‘lassical teeluu’ml imiicators such as a trendline isitssubjectivity. Development ofpositive expectancy models ismuch tougher with fundamental an'alysis because we. are. trying to develop models with disciplined rules to help us get away from our natural tendemty to trade with a bias toward big losses and small profits. lteimtmlwr, you can always find fundan'iental argunumts for s(.-lling or buying at any given price, oth— erwise no one would bewilling to buyor sell atthat price. Also, these arguments can actually prevent you from acting on the high-prolmbility move 0r~even worse~—from marmging the risk. Intrmling, we call this[)(H"(l,/,I_’/.S'l,".8' from analysis. Consemiently, most positive expmttaney models are based upon objective, matlu-.*matical tuthm‘cal imlicators such as oscillators or moving averages. We can never know all the, reasons why the. market rose, on bearish inventory numbers or why it fell despite a decrease in unemployment, but we can develop various rules for entry, exit, and risk management based upon objective, matiuimalhtally derivml technical formulas". Does this mean that fundanu.*ntal analysis is useless for speculative traders“? Notatall. Instead Iam trying toestablish arealistic umlerstanding ofits limitations before our examination ofits utility. So how can we augment our positive expectancy models with l'undanu-ntaI armlysis". The. way Iteach fundamental analysis to traders isthrough old Wall Street cliches. First cliché: “Buy the rumor, sell the news.” Ifthe rumor is that the unemployment report isgoing to show adecline inunemploynuent and therefore astrengthening economy, one might buy the stock market. Once the reportcomes out showing the anticipated improvement injobs, sell the. market. Why? Because the reason for the rally has" come to fruition and there istherefore no longer any reason to own (.‘(mities However, there is one caveat tothis cliche’, and itisan'other Wall Street cliche: “The market hates surprises.” This means that ifthe market was rallying before the release of

Hi

THE CASINO PARADIGM

the unemployment report based on the rumor ofthejobless rate falhn'g to 9.7 percent, and the rate actually falls to 9.1 percent, equities should probably still be bought because the news was a bullish surprise beyond the expectations ofmarket participants. Another valuable. way ofincorporatln’g fundamental news into our pos itive expectancy trading models isto capitalize on t1rn'es when the market reacts in the opposite manner from what would be expected based upon the release ofa bearish or bullish fundamental news item. For example, on April 29, 2009, the US. Energy Information Administration released its weekly inventories report, which showed that crude oil stockpiles in'creased bytwice the expected amount.4 Despite this“ bearish news, the011' market rallied (see Figure 1.11). This rally on bearish news was the most bullish infomiation the market could offer. Itsuggested buyers were waiting for bearish news to establish or add to their existing long positions; consequently, the market could not drop despite the release of negative fundamental news. Oras my fn'end Richard Hom likes tosay, “Ifthey can't sell offon this news. what“ they do when the bullish news hits?”

.441me - Bearish Inventory ropoti

I’ll-IRE l.ll June 2009 Daily CME Group Crude Oil Futures Contract Rallies Despite Bearish inventories Report Source: CQC, Inc. ©2010. All rights reserved worldwide.

'7 I7

Developing Positive Expectancy Models

Perhaps the most invaluable way ofincorporating fundamental analysis into our positive expectancy model is its ability to help us distinguish lll'lWl‘l‘ll price shock events dI'ld paradigm shifts. We have already defined a pm'mligin shift during our examination of the crude oil market and its shill ol'long~term value from below to above $40 a barrel. You may recall lhnl this shill. in the perception ofval'ue ofcrude oil occurred because of a combination of fumlamental supply an'd deman'd factors. Bycontrast, a priceshock isa headlirm—driven event that temporarily spikes the price of an asset beyond itsvalue. The easiest way to distinguish between the two isbylooking atsome historical exzunples. Figure 1.12 clearly illustrates a long-term shift in the perception ofvalue forhigh-grade copper. Before 2005, the $1.60 area acted as resistance to higher prices throughout the contract’s history. In 2005, the perception ofvalue ofcopper underwent aparadigm shift and as ofthe writing ofthis book in2011, the $1.60 area represents along-term support. level fortheasset. One ofmy favorite examples ofaprice shock event was the capture ofSaddam Hussein on Saturday, December 13,2003, bycoalilion forces during the second Gulf War. Hussein’s capture occurred over

.

.~..w .v: ,, ' s; ‘

ll‘

lum‘i‘ M “llcjfil'llmm Ill hot“ l

i “unbound

R “m- "lion"

1997

'm

I’ll-‘lllllu' I.l2 Quarterly Continuation Chart ofCME Group Copper Futures ShowIng 2005 Paradigm Shift Source: CQC, Inc. ©20l 0. All rights reserved worldwide.

THE CASINO PARADIGM

l

on m... ..\..\\ “MMHWO Mario!

Ialulna to

FlGl=‘Rl‘.‘ l.l3 2003 Hourly Cash Eurocurrency—U.S. Dollar Chart Showing Price Shock Event of Hussein's Capture Source: CQC, Inc. ©2010. All rights reserved worldwide.

theweekend and when the cash foreign exchange markets opened on Sunday, December 14,the US. dollar' rallied sharply against the eurocurrency. However, over the course of the next 24 trading hours, currency traders realized that the capture ofHussein had no lasting impact on the value. of the U.S. dollar against the eurocurrency and the asset returned to its pre— headline value area (see Figure 1.13). Why isthe ability to identify a par'adigm shift essential' to our implementation ofa positive expectwicy trading model? Because these models tend tobedriven byrules generated from mathematically derived technical indicators like moving averages. Bollinger Bands, aJ'ld so on. Ifwe blindly ignore the paradigm shift, itispossible that these technical tools will tell us the wrong story regarding price behavior and asset value, especially if we are using mean reversion models. Ifour mathematically derived rule—based system 1'sa seasonal pattern recognition model, we must prepare for the occurrence of an anomaly year. Anomaly years are well illustrated by examining the unleaded gasoline—heating oil spread. Historically, unleaded gasoline. had always traded at a premium to heating oil during the spring, typically peaking

Developing Positive Expectancy Mode/s

l9

FIGURE l.l4 1995—2007 Monthly Continuation Chart of CME Group Unleaded Gasoline—Heating Oil Spread Showing Pattern ofSeasonal Strength in May Source: CQG, Inc. ©2010. All rights reserved worldwide.

against the winter fuel dunn'g the calendar month of May in anticipation ofsummer drivm'g season (see Figure 1.14). However, in 2008, the market experienced an anomaly year in which petroleum product prices moved counter to this historical relationship. Increasing demand for middle distillates like heating oil from developing world nations drove the price up against unleaded gasolln'e because the latter was not used as the pn'mary transportation fuel In those countries (see Figure 1.15). For those who blin’dly followed their technical models to the exclusion of fundamental news, itseemed like easy money to buythe undervalued unleaded gasoline and sell the overvalued heating oil. By contrast, those with one eye on the fundamentals tempered their technically driven models in light ofthis shift inthevalue ofpetroleum products. Regarding price shock events, Ihave. often heard traders dismiss such events as completely random and therefore a50-50 chan‘ce. Inother words. they do not concern themselves with price shock events and rationalize away their occurrence through the delusional belief that over the long run they will end up on the winning side ofthe shock 50 percent ofthe time. Having done the research, Ican assure you that price shock events are

THE CASINO PARADIGM

I’ll-‘1 RE LIS 2005-2010 Monthly Continuation Chart of CME Group Unleaded Gasoline-Heating Oil Spread Showing 2008 Anomaly Year Source: CQC. Inc. "CC20l0. All rights reserved worldwide.

not 50-50 propositions. Instead, you have a greater probability ofbeing on the right side of the event ifyou are trading in the direction of the longtenu—one to six mont,hs—~'trend, an'd a greater likelihood ofbeing on the wrong side ifyou employ a meal reversion model (see Figure 1.16). Now that we have eanined the strengths ofpositive expectancy models derived from mathematical' technical imlicatois as well as their weak— nesses an'd tools to offset such wwknesses. we will briefly review turning these models into mechanical trading systems. Isay, “Briefly review," because for those interested in an in—depth study ofthe topic, Irefer you to my first book, 1 eclianit'al Tradiug qulents: Pairing Trader Psychology with Technical Analysis. Instead ofrehashing maten'al's presented inthat book. Imerely point out here that mechan'ical trading systems ba5'ed on mathematical technical indicators help us determine the following.0 Does this model enjoy positive expectancy? 0 What kinds ofweaknesses—maximum consecutive. losses, worst peakto-valley equity drawdowns, percentage of winning trades, average trad'e duration, and so forth—did this model experience inthe past?

Developing Positive Expectancy Models

2|

Nio203 arcan'ocLuou4 Inc“ 4.6 rolilexflru.. loudne- «we minus-uno5! Dram-ml .1004! orconlvmnon 24 voTououluilDuIOuutvonI ‘ xConucIim - mohrconugo wort-Down 53 rCoMoclins I a l a u 2: v. «a 12 I. 20

FIGURE 2.8 Daily Chart of Coogle with 9- and 26-Day Simple Moving Average Crossover System Note: Trade summary includes data from January 1, 2000. to December 3l. 2009. assumes position size of 100 shares, and includes $l0 round-turn deductions for slippage and commissions. Source: CQG, Inc. (92010. All rights reserved worldwide.

traders using some ofthese types ofindicator-den'ved stops cannot definitively state when the stop loss will be triggered beforelumd :uul so have no crisp, predetermined stop loss exit order placement before trade entry. Stops based on violation ofsupport dJ'ld resistzmce leVels are especially attractive because they ar'e attuned to “price having memory" as stated in Chapter 1. These stops are typically placed to trigger at violation of the highest high or lowest low ofa pzuticular number ofprevious trading days (see Figure 2.9). Monetarv'v or percentage stops are also quite popular' because they force us to quan'tify risk in relationship to rewar'd pn'or to trade entry. zmd in this way, ensure that we are. consistently adhering to one of the cardi— nalrules ofpositive expectancy trend-following trading models: large profits and small losses. Figure 2.10 shows a countertrend system that enters when both Bollinger Bands and a relative strength index signal extreme overbought or oversold levels. Exit occurs with profit. ifthe mar'ket returns

33

Price Risk Management Methodologies

10600

Highest high ofprewou's 3days

101M

1m 10500 “MOO i VOW

a 3

10200

Lowesn low ofprevrou's 3days

@ 3-day low

5 l V t I ? \ l l

my

‘osmlPlof-lTelevuooCouu-silnm Pvtwtiong Arr-gemvu tr«anew-.6" '. nr or" re

5 >°lllNflPm'—ri

.. ..n-m 2M7 venomous .luau ‘twl- 5' Dim! Im'lilm' oat-n! M ' - “Cease:- ».5 ~ ‘'23

. .. “m. '7 I remain“:.. relitYMr.s., whitooshl'oMT‘J.’ rmlirmu '9 I v (an-cirr‘ “tuning. Mir-Dov! 7

i2“ 1

' 29 ‘ 2. 0.6? 2 5) 34 63 -,"66 1‘

2‘5” 4 g.

:2!

NM

I.

FlGl'RE 2.9 Equalized Active Daily Continuation Chart for CME Group Soybean Futures Contract with RSI Trend System Where Stop IsPlaced atLowest Low or Highest High ofPrevious Three Trading Days Note: Trade summary includes data from January 1,2000, to December 31. 2009, and includes SlO round-turn deductions for slippage and commissions. Source: CQC, Inc. ©2010. All rights reserved worldwide.

tothe previous day’s 20—day sm‘lple movm'g average or a stop loss order istriggered when the asset n‘olates a $1,000 monetary stop loss level (see Figure 2.10). Aword ofcaution on monetary orpercentage stops: They should never be constrained solely byassets under management irr'espective ofan assets volatility. Inother words. Ifyou cannot risk more than $1,000 without becomln‘g overleveraged. theanswer isnot to blin'dlyplace $1000 stoploss orders in'all available markets. but ins'tead to lm'u't one’s trading to lower volatility assets in which such stop orders can be placed m’thout bein'g triggered bymeaningless market fluctuations (see Chapter 5for a more comprehensive explan‘ation). Finally, an examination of stops would be incomplete m‘thout disCussing time as a stop loss em't criterion tool. In fact. usm'g the calendar— oreven theclockmfor stopplacement was implied insome oftools already discussed, such as our pn'ce—driven stop (see figure 2.9). in'which trall'm'g Stops were. set at the highest high or lowest low ofthe preceding three

THE CASINO PARADIGM

LOGO ()0 lup Inn! Iu mound

Exit with profits @. previous day's 20-day sun'ple MA

0

. Opanl‘aultoh Polclnllohg Ant-Murrow Ely-rayon!"1| . on I'M -

2!

to“ atonuciouio'i- H" "w ‘«or imconuotalulIOI minimal)!“-

in; “mm run yang-I on ll lllll'.0II

m ll

to:

If’lluOflDl" ImnAnovnl IDIIIDUIIHM Ilromuwinl uttenculim -

ho

[Ir

lot 172 If!

'

Yolnmotlmlit LImomulrmi lIloulluuotmvnh

roliltonlllloorconlmnnvn onou'oMululInfernal-go 1 urou_r_w9wn -

In

" 13116!”“W’Nloo'mm'

FIGURE 2J0 Daily Chart of Powershares QQQ Trust ETF with Bollinger Band Countertrend System Using $1.000 Monetary Stop Loss Note: Trade summary includes data from January 1, 2000. to December 31, 2009, assumes position size of 1,000 shares, and includes $10 round-turn deductions for slippage and commissions. Source: CQG. Inc. (C) 2010. All rights reserved worldwide.

trading days. Beyond price and indicator-derived stops, which were somehow tirrw-dependrant, the clock itself can also be the primary method of forcing trade exit. Acommon example ofthis is a trading system that forces an exit if mark—to-market settlement on the position shows an unrealized loss after three trading days. My only caution on using time as a stop isto restrict the forcing ofexits to trades that are showing unrealized losses. Exitm'g of trades showing unrealized profits because you have been inthe profitable trade too my days isnot only counterintuitive, but also extremely eountemroduetive, as" will be shown ingreat detail throughout Chapter 5.

Volurmalrie Position Sizing Although stop loss placement isthe most rudimentary and indispensable fomt of price risk management, itis not robust enough as a standalone

V "WWV" V .;w

7’ Pric1Risk Management

Methodologies

SB

toensure success as atrader. and must be combined’ mth‘ m" p0 5mm 51mg. Whereas Stop loss“ orders answer the questm‘ ‘Whem m 1 mtnus position in order to prewn'e capital.” volumetm" perm" m’ answers the. question ‘llow many units ofthis asset can Itrade um becoming overleveraged'." The most robust n-stmme‘ to this my 15' [he mse-‘_guy answer: “small enough to eist'un‘ that a positn‘e expectancy model Will' not blow up. while still producu‘lg retiu'ns in excess of the risk-free rate." Although twhnical'ly accurate. this arsrmer 5'too argue to be useful to n'sk managers or traders. We need u‘tst‘ead to exam the per trade. percentage n'sk ofassets under nwragement m‘ relanm' to tk worst. peak-to—valley drawdown expen'enced by our posim‘e expectant! tradm'g system. m‘rem} Thegeneral guideline regarding vohunetn‘c infltior‘i ship to assets under management st" the lpercent rule." This rule m that traders should risk no more than 1percent of.:L'\‘\8L\“ under manage ment on any sm'gle trade. The Idea here isthat thevast man'u'ity ofpoems?“ expectan.cy trading models will berobust enough tosun-I've petite-win drawdowns inequity ifwe risk only 1percent ofaswts under W on anysm‘gle trade. Returning to our back-tested trading‘ results in Figure ll fora 1110!!) baseload currency ofthe Bn'tish poundagainst the [15. dollar mum l.2000, to December 31. 2009. the reader will recall that \k‘fi‘ltk“ mm an overall profit of$5,974. the model expen'enced u wont: peakhuflh-s drawdown in account equity of310%". To better understand the m“ cance ofthis drawdown in relation to the 1percent rule. we need to but atour models performance m'relation to itsexpected wont: pertrade {can Our RSI extremes trading system for cash‘ tbreigt'l exchange M" pound—US. dollar used a$3-KpO' stop loss. mid we assumed a$10per mm!turn cost forslippage andcommissions. This means that auxmhmr' tod» 1 percent rule. we need to trade the system without beu‘tgmerinee aged. Italso means that our worst [X‘dk“[0-\dl“l€y dramiown m‘ seem-i “ll-11W 0f$2.90'5 represented avery mar-iageable wont: txulwtowfley draw down of 6.96 percent. The bad news isthat our 10-year total net lama} of 33-1974 translated into an annualized average total net profit of$597.40‘ or a 1.66 percent annual rate ofretum. Ifthe 1.66 percent rate ofreturn swms unattracthe thesmtpkfi sols» tion is”torisk 2percent ofasu“ts under mai‘iagement on aper trade kw“ The800d news is'that bydoing this. our (u-uiualizul mte ot‘n‘tum “‘M‘ m3.32 percent. but the bad news is that we now have to (“hire a “N peak‘m—V'cllley equity drawdown of 13.9.." percent ofgusts under um "lent Ifa3.32 percent annual'ized return on investment is>11.“ too am“ forUS. We might betempted tons'k 10percent oftt'N‘IS under “WIN 0“everytrade. Inso doingwe could enjoyarobust ltio‘pereent stratum

"w. V1011 . .

36

THE CASINO PARADE,"

rate ofretum. but would have to risk enduring the near fatal worst peak. to—valley drawdown ofoars percent. Why isa peak-to-valley drawdown in equity of 00.58 percent consid ered to be “near fatal"? Because ifitoccurs at the outset ofour trading— which we always have to .‘issume as a distinct possiln'lity—- we would needa return on investment in excess of225 percent to regain our initial asset un. der mzmagement imestmeut. Also. this e.\'tr'.iordinary rate ofreturn would have to be accomplished with a stake ofless than one‘third ofour initial‘ assets under management. While achieving such a rate ofretum with this dimim'shed equity stake isnot mathematically impossible, Iwould prefer betting against. as opposed to in favor of. such :ui occurrence. Now that we have proven the robustness ofthe 1percent nile forvolumetric position sizing. how can we s.'ifelv_' increase our position size with out increasing n'sk as our account equity increases‘.’ Also, how do we determine when it isnecessumg to decrease our volumetric position siLe' as our account equity decreases"? Although there are numerous methods of luuidling the adjustment ofvolumetric position size as assets under management change. one ofthe best-known teclmiques is Ralph Vince's fixed fractional position sizing.T The most cousenative way ofusing fixed frt'ictional position sizm‘g is' to look at the worst peak-to— 'alley drawdown over the lmckvtested pen'od. Retuming to a volumetrically modified version ofour RSI extremes bac'k test ofcash British pound—US. dollar (see Figure 2.] l),we cai'i see that bytrading 100,000 base currency, our worst drawdown was $223,100 and that this represented a worst pmk-to-valltw drawdown of0.43 percent of assets under mzuiagement ifwe adhered to the 1percent rule. This being the (‘380-—-b215(‘d on our historical back-test ed perfoniuuu'e—i'f We started with $360,000 in assets under iuzmagemeut, we could safely increase our position size from 100.000 base currency to 110,000 base currency without exceeding this 6.5percent worst pmk-to-v.'il|ey equity drawdown when as sets under mzmagement increased to around $.‘t.0".),000_ Bycontrast. ifas sets under management decreased to around $.‘V..‘5,t)()(), we would need to decrease our volumetric position size to 00,000 base currency inorder to maintain the fixed fractional worst pn'ik-to-valley drawdown in equity of around 6.5 percent. While the 1percent rule isa robust solution for the 'ast umon'ty of trading models. in rare instances itissometimes a suboptimal solution for certain shorter-term trading models. Imagine a model that eiu'oys 0.)" percent winning trades, but the average win isaround one—fifth the size ofits average loss. For these models, adherence to the 1percent rule is counten’ntuitive since after experiencing a loss, the odds of endun’ng a swcond consecutive loss are astrouomimlly low. In such insuuices. since the aV' erage profit per trade isso small when compared to the :m-rage loss and

ma Risk Management Methodologies

etlllbl w H IMIPIOIII-

37

e.

Exfmv—mum'CMOIIOI' mil "MIionitaulu io-

FlGl’RF. 2.” Daily Chart of Spot British Pound-U5. Dollar Chart with RSI Extremes Trading System Note: Trade summary includes data from January 1, 2000, to December 31, 2009, and assumes $10 round-turn deductions for slippage and commissions.

Source: CQG, Inc. © 2010. All rights reserved worldwide.

because the model’s sufferln'g oflosses so rare, an argument can bemade forusm'g a stop loss of 6or '7percent ofassets under management. 'Ihs‘ stated, until you can prove the robustness ofthese short-term model results. itis.always safer to stick with the 1percent rule.

MIDDLE 0F PYRAMID Quantitative tools inthe middle level ofthens'kmanagement pyramidoffer robust solutions to issues, m'cludm'g correlations between assets held in'a portfolio as well as thevolatilities ofthose assets.

\‘alue-al-Rlsk Amore, robust answer regarding stop loss placement is‘that our stop levelsshould beattuned tothe current volatzh'ty ofthe asset traded Inother

m: CASINO mum words. in higher‘ volatility”' ennro'nments. we will need to place our stops further from our entry pn‘ce so we can avoid being needlessly stopped, gm of trades that would eventually result 111‘ profit, while in lower volatility markets we can place our stop levels much closer to entry without get. tmg'§opped out on false countennoves. Tlu's relationship between v-olam'. m‘;and Sop level placement is'the reason we never look at stop losses“ in a vacuum but inst'ead examine them in' conjunction with volumetn'c post; 000' sung," In other words. when the volatility of the asset is'higher, we place our Sop further from the entry pn‘ce level but we could potentiall'y trade fewer contracts. whereas when the volatility is lower, we place the stop closer to our entry pn'ce and could therefore potentially trade a larger number ofcontracts m'thout violatmg' rules ofprudent n'sk management. “its relationship" between stop loss placement level, volumetric post Don' sang." and the volatility of the asset—or assets—traded transitions 15 to the middle” trer' of our risk management pyramid and specrfi‘cally to \‘alueat-Rxsk.‘ or VaR. VaR adds two m'dis'pensable elements to our ns'k management models: volatili'ty and correlations. \"aR examines the histo‘rial‘ \olanlrt\",' ofassets held In' a trading' portfolio an'd the. correlations be tween those msets so as to make our stop loss placement and volumetric Inaddition' to mcorpo‘ration oflust'on‘cal volatility ofassets traded git"mg' is a more roots answer as to where to place our stops and what our posmon" sme‘ m’the market should be. volatili'ty is'a natural complement to stop placement and position 512mg" because it automatically attunes us to changes in"met nsk' due to shifts' tn'theassets value. For example. m‘ December 193.1. Procter &Gamble traded at $10 per share; a 5030' stop loss therefore represented a significant 5percent move In"the stock Bycontrmt. in‘Marc‘h 1999. P&G was trading at 500'per share Ifwe continued to blmdl'y set our stop at the static 5090' level‘ we were now only ns‘lung‘ 1 percent of the stocks value and could be stopped out Lawa rumor fluctuation (see Figure 2.12). Switclun'g from static dollar amount stop loss placements to stops attuned to shifts' in asset value—and has im‘plications for position srzm"g. In 1991. when P&G traded at $10 per share. a 5000' per trade ns'k ceilin"g translated into trading~ 1.0:!) shares with a 5030' per share. or 5percent stop loss. On the other hand in’ LUQQQ when the company traded at 3500' per share. traders needmg' to retam' the stop loss of5percent ofthe stock's value as well as the Snail mk‘ ceiling" not onlychanged the per share stop loss level to 5290'. but were abo forced to reduce their volumetric position size to 200 shares. In addition to attumng.' us to his‘ton'cal volatility. Value-at-Ris‘k makes WW" demons more robust by analyzing historical correlations among assets held in‘ our portfolio. Tlu’s is‘ an important consideration because. as a standalone. volumetric position srz'in'g gives a static—and

..'V_

ma Risk Management Methodologies

II"

12191 . o no no FIGURE 2J2 during 19905

Monthly Procter 8:Gamble Chart Showing Change In Stor k's Value

Source: CQC, Inc. ’92010. All rights reserved worldwide.

therefore suboptimal—answer regarding how many units ofa particular mset we can trade without being overleveraged. For example, a 10-year back test of a (.l- and 26-day moving average crossover trading system for a ((‘ME Group front month-del'erred month crude 011' calendar spread expen‘enced a worst pezik-toyalley drawdown of$7.490 per contract. Ifwe. have $1 million in assets under management andare ml'hn‘g toendure a worst peak-to-valley drawdown of7.49 percent, our volumetric position limit forthe calendar spreml would therefore be It) contracts. Ifwe then blindly applied this static volumetric position-sixing formula of10contracts toan outn'ght, position in(IMlu‘ (iroup crude oil, we would have endured a 31.15 percent peak-to-valley equity drawdown usingthesame mechanical trading system over the same mick—tested period. This example illustrates how the strong positive correlation between the (alendar months inthespread translated into lower risk than outright long orshort positions in'the commodity (see Figure 2.13). Table 2.1 shows a first quarter of2010 correlation study of various as— sets. in‘cludin'g cash foreign exchange instmments like the eurocurreucy 383ml the US. dollar, the Australian dollar against the [1.8. dollar, (.‘ME

40

THE CASINO PARADlCM

F—l—o o6'°W.' "Mo K

Yo§.'.

.1"‘6'

-

I . -

Yolalmwroht ClosoauolProiilTolnltudommWas” :00 PorcuMLong vougobunnonvar-gown“! Aver «In . 1

xDrlIOUuIionxConncvms mans-chm -

H705 xnlulmn 1405 - vougoLou 91 lulu-Lou 1 11010101101" 51 xornA-ouni 2 xonvouruloo101 XConncIm: 061 urConucIIM r. .

- xcomocLouu- 1 urConncLonosI'IolllexDrm I (culminatio-- orconulmnn: novfloflcutrllimorconugo -

-

»

xConucLoesu«room-cunnn vanituuxnnw-- rulllLouRuio-- orconrvlmnen nanny-7011001111unePorconugo 4 112‘ ~ y.

FIGURE 2.13 Equalized Active Daily Continuation Charts for CME Group Crude Oil Futures Contract and Crude Oil Calendar Spread with 9- and 26-day Moving Av— erage Crossover System Note: Trade summary includes data from January 1. 2000, to December 3], 2009. and includes $10 round-turn deductions for slippage and commissions.

Source: CQG, inc. ©2010. All rights reserved worldwide.

TABLE 2.I

AUDUSD EURUSD Corn E-Mini sap 500 UST-Notes Crude Oil Cold

First-Quarter 2010 Correlation Studies E-Mini

US

Crude Oil

AUDUSD

EURUSD

Corn

S&P 500

T-Notes

1.0 .32 .32 .86

.32 1.0 .73 —.08

.32 .73 1.0 .01

.86 —.08 .01 1.0

—.36 —.67 —.65 —.1 6

.82 .05 .34 .85

.73 .39 .53 .Sl

—.36 .82 .73

—.67 .05 .39

—.65 .34 .53

—.16 .85 .51

1.0 —.28 —.34

—.28 1.0 .73

—.34 .73 1.0

All futures data are equalized active daily continuation. Data shown are from January 4,20l 0,through March 3i, 2010. Source: CQG, Inc. ©2010. All rights reserved worldwide.

Gold

7' 4|

Price Risk Management Methodologies

[M 8.2

Third-Quarter 2010 Correlation Studies

W

AUDUSD EURUSD Corn {Mini S&P

E-Mlni S&P 500

Crude Oil Gold

Crude Oil

AUDUSD

EURUSD

1.0 .80 .86 .89

.80 1.0 .51 .84

.86 .51 1.0 .59

.89 .84 .59 1.0

.49 .33 .59 .17

.92 .40 ~34 .44

.74 .41 .93 .41

.49 .92 .74

.33 .40 .41

.59 —.34 .93

.17 .44 .41

1.0 —.33 .62

—.33 1.0 ~50

.62 —.50 1.0

Corn

500 U.S. T-Notes

U.S. T-Notes

Gold

Allfutures data are equalized active daily continuation. Data shown are from July 1, 2010. through September 30, 2010.

Source: CQC, Inc. (c: 2010. All rights reserved worldwide.

Group com, E-Mini S&P 500 futures, U 10-year Treasury note futures, (‘ME Group crude oil, an'd (.‘ME Group gold futures (see Table 2.1). Notice how certain strong positive correlations like the +0.82 correlation between the Australian dollar and crude oil are exactly as expected, while others like the low correlation of0.05 between the euro and crude 011' d1f’— ferdramatically from assumptions. This iswhy itis"always safer to perform correlation studies instead ofblindly assumm'g that hls'torically strong. stable correlations will hold up indefinitely despite the ever-changm'g nature ofmarkets. Although correlation anomalies shown in Table 2.1 were somewhat sumn'sing. a bigger problem is revealed by comparm'g Table 2.1 to Table 2.2 (which shows correlations of the same assets during the tlur'd quarter of2010). In compan’ng the tables, although we find some correlations such as the Australian dollar an'd the E-Mini S&P 000' remam'ed stable. mar-1y correlations changed dramatically from the first to the third quarter of3010 (see Table 2.2). Such shifts in histon‘cal correlations are prec1se'ly why Vail should always be augmented bystress testing (which allows for correlation breakdowns). Now that we have shown how VaR makes our stop loss placement and position sizing more robust, let us examine VaR as a risk metric and see how itincomorat es histon'cal volatilities and correlations in'its attempt to measure future portfolio risk. Although an in—depth presentation of\‘aR is' beyond the scope of this volume, my book Mechanical Trading Systems offers readers agood overview explanation ofthe topic:

Value—at-risk methodologies attempt to quantify the standard demotion (or historical 'tvolati'lity) ofa trading asset or portfolio of

42

THE CASINO PARADIGM

assets and the historical correlations between these assets inorder to answer the question: “What is the likelihood ofour losing Xdollars or more over a specfieid time horizon under normal market conditions?” Forexample, aparticular hedgefund might have adaily VaR of$30 million at the 95percent corfiidence level. This would translate into there being a 95percent probability ofthe portfolio not erperiencing a loss inexcess of$30 million over the next twentyf-our hours.8 So a basic question regarding historical volatility as well as historical correlations as inputs forour VaR models is“What isour lookback pen’od?” Remember, “it‘s not magic; it‘sjustmath.” Inother words, there isno perfect lookback period. The advantage ofshorter lookback pen'ods—such as 90 trading days—is that itgives greater emphasis to recent readings ofvolatility and correlations. The problem with shorter lookbacks isthey can give aview ofcorrelations and volatilities that are distorted byshort~ temi trends in‘the market. Bycontrast, longer lookback pen'ods—such as 250 days—give a broader View of correlations and volatility but can dampen the current risk suggested by those. inputs. Also remember that correlations and volatilities of many physical commodities shift because ofseasonal factors each year; risk managers consequently should augment ordinary 90- and 250-day studies with three-to—five-year seasonal lookbacks (see. Figure 2.14). Risk man'agers attempt to address these problems oflookback pen'ods ina wide variety ofways, including—~but not limited to~the use ofexpo nentially weighted moving averages so they can' give greater weight to the most, recent volatilities an'd correlations. But here again, there isno perfect answer as to how much weighting istoo much and how much istoo little. Instead of giving a suboptimal answer to these questions, my suggestion to risk man'agers ai'id traders isto explore amultitude oflookback periods along with a multitude ofdata weightings so you ('an' gai'n a robust and accurate feel forcurrent as well as future trends ofvolatility ar'id correlations. Another problem regarding histon‘cal volatilities and correlations is that of throwu‘ig away data. There are two ways in which historical data can be discarded: intentional exclusion an'd scenari'b roll-off. Intentional exclusion is fairly straightforward and self—explanatory. It occurs when risk managers intentionally exclude segments ofdata byconvincrn'g themselves—and theorganization theywork for—of the prudence indeletm'g a segment ofdata history because it represents a histon'cal' anomaly. Some risk managers in the natural gas industry did this after the 2005 hum'cane season, reasoning that data derived from that season—which included Hurricane Katrina—represented an outlier event and therefore should be omitted from volatility and correlation studies (see Figure 2.15)-

43

" J

Price Risk Management Methodologies

5/05

5/“ \

i+\“°lll \,‘;5. 5/01

SIM

5/04

Wsloz ' '\

5/017

Ill l"‘\l 'l|'\ l“ “Ml illull"|' Ill

l

FlGl‘RE 2.14 1995—2007 Monthly Continuation Chart of CME Group Unleaded Gasoline—Heating Oil Spread Showing Pattern ofSeasonal Strength in May Source: CQG, Inc. © 2010. All rights reserved worldwide.

2005 hurrl cane suaon as bull:or event?

'

Pro-2005’ hi’ ho I'n Natura| as

\

'"Gl'ltlu‘ 2.15 Monthly CME Group Natural Gas Futures Continuation Chart Show‘W 2005 Hurricane Season

Source: CQG, Inc. © 2010. All rights reserved worldwide.

THE CASINO PARADIGM

2005 Minoan. at outta-r own”

zoos rovultallm of unit-r level. In volatility

10 2005 M o I. Mun-l 8::

FIG! RH 2J6 Monthly CME Group Natural Gas Futures Continuation Chart, Including Data after 2005 Hurricane Season

Source: CQG, inc. CL2010. All rights reserved worldwide.

Although it might seem logical to exclude data that represents a histon— cal'ly unprecedented event, thepractice isextremely dai'igerous since thest“ volatilities and correlations are part ofa real' historical data segment, aid in deletu'tg them we we assuming away levels of risk that were actual'ly endured bymar‘ket participants inthe past. Worse still, because price has memory, the fact that such data history occurred in the past suggests the distinct possibility oftheir reoccurrence inthe future (see Figure 2.16). Bycontrast. scenan'o roll-off occurs when risk mai'iagers unintentionally exclude a segment ofdata history due to the model moving forwart'l 1n'time and thereby assuming away levels ofvolatility experienced outside theselected lookback period. Forexample, risk numagers usingaone—year lookback for their volatility and correlation zuialysis would have assumed away the possibility ofd1‘10th0r9/ll occurring on September 12.2002. simplybecause September 11,2001, had suddenly rolled offfrom their historical lookback window. Solutions for problems of data exclnsion—both intentional' and imintentional—are inclusion ofthe data in question. Issues of how much or how little weighting to give to outlier events that would otherwise be

7’ riceRisk Management P

Methodologies

M

our data history ispart art. and part scit‘ncv, and risk mun excludcd from .lgcrsshould migmcnt any DUN‘ly (mantitativc tools for data nwamnmnmt O{Outliers ai'id data othcrwisc suliicct to roll—ol'l' with thvir knowledge of asst).ishold inthcportfolio. (

Stress Testing Assuggcsted in our discussion of VaR, it is thc risk nnmagm-‘s nttwnpt togain a robust. and accuratc fool for currcnt as wcll as l'uturv timid-s of volatility ar'id correlations that rcprcscnts hisgroan-st challvngc. 'l‘lw t-lml lenge isparticularly acute ho nurse of two prohloms: first, tlw cyclical on. ture ofvolatility and sccond, correlation breakdowns. 'l'lu- cyclical naturiofvolatility suggests that even a robust mcasurc ol'Iiistorlcal volatilily nor essan'ly falls short because periods of high volatility lcad to low volatility andmore dan'gerously, periods oflow volatility rcsolvc tlwmsclvcs in high volatility (see Figure 2.17). The risk manager should ('onscuucnlly always erronthe side ofcaution in his estimation offuturc volatility trcnds.

' -

M. W. 1.13 1020 "29/

10‘!"

1mm

IOMI

Mien Va ati llly

I'll!"

HI‘ voFIQI‘II'!1

I'll"!

7000

k

MIYh v0 allla'ly

"M

l700

"'00

Im vol-llll'ly

"00

vol-(I'li'ty volatility

‘Iu.

,5 r

.‘.‘.v Jul

Jun

..‘7 Jul

Jan

-.‘nr 0

Jun

"'.4 Jul

Jon

0

M

ul

Flour, 2.17 Monthly ICE Number 11 World Sugar Continuation Chart Showing CYcliczil Nature ofVolatility

Sou’cef CQC, Inc. ©2010. All rights reserved worldwide.

t'oi‘riilution lirvnkilown is also a signinmi‘it pl‘Ublt‘lll inherent in Val: iiioilvl ltflflllllllillullfi' Historical vorri‘latiom muting amass hi‘hl In a with» No lllt' not only sulijc-i-t to tin-liking down in tlw hiturt', hut thNI‘ lu‘mkv ilouu (Hills to owur whvn wv llt‘t‘d thvm to hold up the moat. “hm um lu-l \olnlllity iiii'i‘st-i-~..'. This is oxrluplitlvd bythe i'orn-lation hl‘euluhmu lu-im-i-n gold uml vquity imlvx t‘uturvs (llll‘lllg Iht‘ ('l‘t‘dll t‘l‘lnl‘h' of mm In Humv .‘3,lh‘, wv saw that for much ot'tlw tlmt halfofflll‘i‘. (ME ih‘oup gold and slot-k iiulvx l'ului'vs displaiyi'd u signitluu'tt cu'ul stabh‘ Iu‘gum v nuw lution unit-«ling (is. lly contrzL‘st, throughout [ht‘ ('l’t‘dll i'i‘im‘s ot' .‘Nltl’x‘ t'oiuwi‘h' sluhlo uvguliu- vorrvlations hrokv down prwiwh “hi-u trailers uvmlml Ilioso most to dzunpm portfolio risk in(11‘! ("llVll‘Olllllt‘lll oflllt‘M'V' luglll.‘ll'kt‘l Volatility (500 Figun' 2.18) l‘t'ohlvms of tho vyclival nuturv of volatility as well as i‘orwlau'm lirmikilomm (along with M“2L\‘0ltd‘l volatility shi‘fis dl'td seam‘uw wiouial'm ot' pliys-i'vzil vommoilitiosum‘0 l-‘igun‘s 2.14 mid 2.15) illustrate Wm \‘aR‘ mmlvling is not robust as a .s'uuul‘u'onv I‘Ls‘k managenwnt metric mid mu

.9'

l"‘.-V!’ *ll It.I.‘ 'i".VI'IVI.‘.'i"

iiililf

HHHH!

THE CASINO PARAGON!

MI

JP“ crisis WIDV. .0 colrclntion with oquily mdicoa

emulation biggies. Quin. "can an“)

MU

M

FIG! RE 2.“! Weekly CME Group Gold Continuation Chart Showing erw."o“ Breakdown with Stock Index Futures during Credit Crisis of 2008 Note Correlation study is a 20~week lookback period. Source: CQC. Inc. \C‘2010. All rights reserved worldwide.

price Risk Management Methodologies

47

itshould always be augmented bystress testing. VaR incorporates histon'calvolatilities and correlations inan attempt to quantify the likelihood of aportfolio‘s breaching ofaparticular loss threshold over a specified time horizon, but says nothing regarding the severity ofaparticular loss. Stress testing attempts todetermine how bad this low probability event could be» come. In addition, itallows for correlation breakdown as well as attemptu'igto model forthe, cyclical nature ofvolatility. Just as an in-depth examination ofVaR 1's beyond the scope of ms book. so too our presentation ofstress testing isnecessarily cursory and is limited to its overall value as one segment within our risk management pyranru‘d. One of the most commonly employed types of stress tests is known as scenario analysis. Inscenan‘o analysis, the risk manager applies her current portfolio holdings to either a hypothetical scenario such as a 100 basis point rise in interest rates, or an actual historical scenan'o like the credit crisis of2008. The purpose in running either ofthese types of scenarios is identification ofexcessive n'sk levels in our current portfolio holdings. and where appropn‘ate, implementation ofcorrective risk reduction measures including exiting ofpositions, purchasing ofoptions, and so forth.

APEX 0F PYRAMID When some traders first hear the temi management discretion m'relation tothe risk management pyramid, a spark ofhope reignites in their gambhn'g hearts. Let me therefore extln‘guish that spark from the outset by remu‘idu'ig readers that this isa risk management pyranu'd. and therefore any managerial discretion could be used only to augment and strengthen purely quantitative tools at the pyranu'd’s lower rungs and never to relln“ qu1s'h those tools infavor ofdiscretionary risk-increasing behan’or. Bydefinition, theterm discretion suggests tools that defy purely quantitative mathematical modeling. Itis consequently \irtually impossible to pron‘de an exhaustive list ofall the possible. ways in which mar-lagement discretion can supplement aquantitative risk management model. Instead. letme outline a scenario in which management experience and discretion could be used to complement such quantitative n‘sk models. On se‘ptember 11. 2001, acts ofterrorism ar'e shifting markets to heightened levels 0fpanic. Ahedge fund’s risk riimuiger checks portfolio exposures. agzu'nst VaRlimits, even runs astress test to determine ifthe fund‘s trading book is enduring excessive levels ofrisk. Despite the fact that all her quzmtitative models suggest exposure is within nonnal' tolerances. she cal'ls the fund's headtrader. suggesting areduction ofportfolio exposures.

48

THE CASINO PARADIGM

Another example of management discretion is especially ins'tructive as itSimultaneously illustrates how manager experience can be used to augment quantitative ris'k tools of our pyrarru'd’s lower rungs while highlighting instances in'which we might ignore entry signals generated bymecharu'cal trading models. Regarding this' second pom’t, when people askme ifmy own ns'k management is' 100 percent mechanical, Ianswer thatitis 99 percent mecham'cal and 1percent discretionary. When they roll thelr' eyes at the seenrun‘gly arbitrariness ofthis' answer, Ielaborate, explaimn"g the discretionary element can only reduce and never increase the risk endured. Then Ishow them spec1fi’c examples such as the September 2010 CME Group wheat futures contract (see Figure 2.19). On August 5,2010, wheat futures closed locked hnu"t up. The followingday, August 6,2010, ittraded up almost the 60-cent daily lim'it, only to turn around and settle locked lmu"t down on the day. The followm‘g trading day, August 9,2010, saw some good follow-through selhn'g In'the market, which resulted in'the triggerm'g ofasell signal for one ofmy countertrend trading models. Despite the fact that Icould have sold September wheat futures without violating volumetric position-Siz'ln'g lmu"ts (or any other

rfrm: “.55 1/4 hrghor Mt elm: up w.w «I'ly ll'll't ‘\

Wntertrem sell aI' nal for mn of Mgust 1 th

FlGl'RE 2.I9 DailyChart ofSeptember 2010 CME Group Wheat Showing Extraordinary Levels ofVolatility Source: CQG, Inc. CC2010. All rights reserved worldwide.

Pme Risk Management Methodologies

49

purely qti.'mt,i't..'it.i've risk criteria), Iused discretionary n'sk management as an oVerIay of those purely quantitative tools and chose to ignore. the sell signal for wheat. generated bymy mechanical trading system.

FINDS AND CIDNS (IF THE RISK MALMI‘l‘i‘Ml‘INT PYRAMID 'I'hc pyrtunid discussed in this chapter is a comprehensive model for traders and risk managers combining a diverse array of quantitative tools such as stop losses, volumtdn’c position sizing, volatilities.., and correlat.ions——'represented byits lower rungs—‘which are augmented bya discretiotmry overlay at its apex. Also, the weaknesses ofeach tier ofthe pyrtunid are offset. byother tiers. Despite the robustness ofthe pyramid as a risk management model, it isnot the ultimate solution to pn'ce n'sk man'agement, but instead should function as a solid foundation upon which leaders in the field can build. Although the model augments purely quantitative. tools with managen'al discretion, Iintentionally avoided thetemptation offormulating some sub— optimal, quantitative—volatility-based—rules as towhen this discretionary overlay should be introduced. Ioffered instead some obvious examples where the introduction ofmanagerial discretion proved prudent 1n' hopes ofshowcasing robust management discretion.

PUTTING IT ALL 'l‘lNfilu‘Tlllu‘R: ACASE STUDY Aspeculative trader decides to fund a futures account with $100,000. He iscomfortable with risking 2percent ofassets under management on any single, trade and wants to simultaneously trade gold, corn, and the E-Mini .WI’ 500 futures. This means he can trade only when their strategy’s initial stoploss levels are $2,000per contract or less. Ifhisback-tested correlation study ofgold, com, and the E-Mini S&P 500 suggests a low and stable correlation, hecould potentially have as much as $2,000 at n‘sk in each ofthe three d.S".S('.*iS traded. This would not account for the possibility. however. ofcorrelation breakdown. Heconsequently decides that despite the historically low correlations between the assets, he will not commit more than 32,50), or 2.5 percent, oftotal assets under management in all the assets traded simultaneously. In October 2010, his strategies have simultaneously generated two trmling signals: aFibonacci retracement buy signal inDecember 2010 corn

50

THE CASINO PARADICM

Buy .61 “ lair-mat 4“ 3/4 IIlll

.lh‘ll It? 0 444 1/ oupwrl

I’ll-“IRE 2.20 Daily Chart ofDecember 2010 CME Group Corn Showing Limit Buy level as Well as initial Sell Stop Level Sauna CQC, Inc. (c,2010. All rights reserved worldwide.

futures and abreakout signal inthe E—Mini S&P 500 futures. The strategy in corn isto buyat the limit price of15.4.5875. The initial sell stop loss order is" at$4.445 and represents 14.25 cents, or $712.50 per contract, so he can' buy two contracts without violating his 2percent rule. His limit order tobuyat' $45875 W('LS'(‘,X(E(,'U1)C(1 on October 4,2010 (see Figure 2.20). S.'innilizmeously, hehas been waitingforabreakout from anarrow trar'lingrange in the December E—Mini S&P 500 futures. Hewants to place abuy stop at the resistance area of11".)3.50 an'd asell stop atthe support mm of 1127.25. llis protective stop loss order would bethe other side ofthe sideways channel, which represents a $1,312..,")0 per contract risk level, which means that he could buyone contract without violating his 2percent rule. (in October 4,2010, however, when his order to buytwo corn contracts is tilled, itrepresents an initial risk ofaround $1,425, excluding commissions an'd slimmge. Heshould therefore cancel resting orders inthe E—Mini 8&3? 500 futures, otherwise he risks getting stopped out on both corn as well as the stock index futures an'd enduring a loss of $2,737.50, or 2.74 percent. ofassets under man'ag'ement, which isbeyond his stated risk tolerances (seeFigure2.21).

>

3|

price Risk Management Methodologies

1127 25k DWI.

"x. \‘ 1014 M Com, moo-I E-fluu ordonl

l-‘IGlRH 2.2l Daily Chart ofDecember 2m 0CME Group E-Mini S&P 500 Showing Buyand Sell Stop Orders Source. CQC, Inc. g 2010. All rights reserved worldwide.

Despite its sun‘plicity, this case study used clearly defined stop loss levels. volumetric position Slz'in'g, and correlations between assets traded. Also. although not explicitly stated m' the example, sm'ce stop loss levels were based on support and reSIS'tance of the assets traded (as opposed to monetary stops irr'espective of volat1h"ty—see Chapter 5for more details). we were adjustm'g our position srz'e based on his’torical volatth'h; of theassets.

Fl.\‘_-\L TIIOIJGII'I‘S Before monn‘g on from the topic ofpn'ce ns'k management. Iwanted to share some final thoughts based upon personal trading expen‘ences and extensive rese.arch. As stated earlier. successful speculative tradin'g demands doing that Which Isuncomfortable and unnatural. For this" chapter. that specrfi'cally means exitin'g losin'g trades quickly. As the cliche goes. "Big vnnn‘ers

imam

52

THE CASINO PARADIGM

and small losers,” or as Itell my students, “Any idiot can take a profit. Professionals know how to take losses.” As a general rule ofthumb, the quicker you can identify losing trades and kick them out ofyour trading book, the better. Ofcourse Iam not talking about putting on a trade and immediately getting out With a loss merely to prove you have discipline. Instead, the quicker you can identify' and exit trades that will ultimately become losses, themore successful you will be. Finally, Ihave often heard traders link their‘ appetite forrisk with their' masculinity, busm’ess acumen, as well as a wide variety ofother irrational associations. Sure, traders want to live on the edge, but do they want to die out there, too? Risk management is a no foohn'g, no second chances proposition. Sinceyou only need to getrisk management wrong once, there isabsolutely no room forbravado, ego, or irr'ationality when itcomes tothe busm'ess ofmanaging risk. To paraphrase Larry Hite, “Ifyou don’t bet, you can’t win. Ifyou lose allyour chips, you can’t bet.”9

(I ll \I''I'IIR 3

Maintaining Unwavering Discipline It’s not thework that's hard, it's the discipline. —Anonymous

Positiveexpectancy tradin'g models fail' because speculators abandon prudent ris'k management methodologies or they deviate from the models themselves. This' chapter completes our introduction to the casrn'o paradigm by exploring why traders abandon positive expectancy models or price ns'k management. Particular emphasis is'on development and use of various psychological tools to aid in' ma1n'tarmn"g trader disc'iplin'e.

DEFINV'ING DISCIPLINE “hat makes successful tradin'g so challenging is’ that itis'possible to deVEIOD apositive expectancy tradin'g model and still lose overall even ifyou are properly capitahz'ed and employ prudent rules ofpn'ce n'sk manage ment. The problem is best illustrated bythe analogy of the opaque urn.l A“Opaque urn containing 100 marbles isplaced inthe center ofthe room. Ififty-seven ofthose marbles are green and 43 are. red. Now Iaskyou tobet on the color ofthe marble you will pull from the um. and you pick green. Ont comes a red marble. Iagam' ask you to pick the color ofthe marble. am you choose green, and agarn' you pull a red mar'ble. Third tm‘ie: You choose green, and out comes a red marble. Fourth tu'ue. you again choose green and again pull out a red marble. After the fourth loss. you begin to

33

s‘

THE CASINO PARADlC“

54

o

3351133} i

'2 H

......i..........,...s....n. .‘i,mm"or"... s.

m "'mmmwaumuwmmwmn "(mummum-ra.

a;

doubt. Maybe there are more red marbles than green. And so you either stop betting altogether or worse still, you bet on the red marble. Now look at Figure 3.1. Ichose to reexamine this particular' grar'ih an'd back-tested model specifically because itwas alrear'ly shown to you twim in'the previous chapter (see Figures 2.1 and 11).Here we look at' it inthe context oftrader discipline ( see Figure 3.1). Inpam'cular, look atthe number next to the heading marked “MaxConse'(,-I/)sses_,.,” which stan'ds for maxrm'um number ofconsecutive losses. As you can see, this [xm'itive expectancy model experienced four consecutive losses, despite producing an' overall profit of$64,420. This means that ifyou are either unai')le—beum-'se' ofoverleveraging on anyparticular trade—or unm'lling—lwcxause ofalack of confidence in' the robustness ofthe model—to take the fifth trade after four consecutive losses, you do not enjoy the profit of $64',420. You lose the maximum drawdown amount of$23,160 instead. Worse.’ yet, lf’a!ter four consecutive losses and a drawdown of$23,160, you decided not only to abandon the model, but instead to fade2 it (that is, bet on the red marble), your next trade would have resulted in a loss of$6,320 and total

3>x 06‘

Younmmvu ImmrProhiTouv‘l‘r-accoum'- nPosiiion !arconiLow ‘ vorageDururongvougomohl V . or m n I,io

cu 644 5

A

xi-u-nn "rogcunt KINDLO‘E xClosNDIDI I xDrnAI-oo'n I xcrnOuraixonlZJ Conieclms 474 urConsecu‘ms u -7 u 1

-231 4, a

rob :LOSIRII -oreel-.lvinnen We‘lmrraiH . i-Puconxag: ulna-Own v 5 x. ,

33 v

FlGl'RE 3.] Daily Chart ofSpot British Pound—U5. Dollar Chart with RSI Extremes Trading System Note: Trade summary includes data from January 1, 2000, to December 3i. 2009. and assumes S10 round-turn deductions for slippage and commissions. Source: CQG, Inc. ©2010. All rights reserved worldwide.

1‘!

psychological demoralization. So it is not enough to have a positive expectancymodel; itisnot even enough tosucc BHI(@,Sim,20,2.00) (—1] Short Entry: Close(@) [—l]< BLO(@,Sim,20,2.00) [-1] Long Exit and Short Exit: MA(@,Sim,20) I-l]

Now compare. both results to athird trend-following model that Icall RSITrend. The system issomewhat eounterintuitive, as most traders think ofWilder’s Relative. Strength Index as a mean reversion indicator. Nevertheless, byentering longpositions when the nine-bar RSI sigmtls a slightly overbought reading ofgreater than (5.") (or short positions when it gives aslightly oversold reading of less than 3.)") mid combining it with a tight I'iSk man‘agement criterion for exit (stops at prew‘ous three-bar low for lOIlg positions an'd previous three-bar high for short positions), it offers aSomewhat respectable positive expectancy model. More important ly,for

THE CASlNO PARADlCM

bolo—J

9'i-.I-I‘D’mr'TI"-"”— :osmrrroi iI-

.‘I u

IDII'7*|DI-' (on (touml io-

FIGIIRE 3.3 Equalized Active Daily Continuation Chart for ICE Brent Crude Oil Futures with Bollinger Band Breakout System Note: Trade summary includes data from january 1, 2000, to December 3i, 2009. and assumes $10 round-turn deductions for slippage and commissions.

Source: CQC, Inc. © 2010. All rights reserved worldwide. our purposes, the tight risk management exit means that our average trar'ie duration is now reduced to an even more attrar'ttive Six trading days (see Figure 3.4). Using CQG, the programming code for RS] Trend iswn'tten this way: Long Entry.RSI(@,9)[-1]

> 65

Short Entry~. RSI(@,9)[-l]

< 35

Long Ex1‘t.LoLevel(@,3)[—l] Short Exit:

H1'Level(@,3) [-1]

V

...|.._.....A.~...-».-.”. ...u.... .. .

Maintaining Unwavering Discipline

_.__ _...__. '-'-’ .' it"s—"Wm "11-" foam—[mon- *1 -'- VUIIQOLO‘I remnant.“ T n gin-Lou ~00! intimationI '0 701 whitest”: :01 92. - 100‘qu 1mm! 4“ tern-mun 37 36 141 mulch-inni6 17 lanthanum57 xConucIIr-o inflate-Mm 37 50 32! ulConIocIIm - snarl-own — -u H” . ._ .. ._., i 1 -

[V v.-. ; otium io'TiT'; i lumwiohl;olaflmooCovnI,Iw Motlleon i imnuong a \NIOOOUIIIIM' ungoholu .l s in I I'. A 1

FlGl‘iRE 3.4 Equalized Active Daily Continuation Chart for ICE Brent Crude Oil Futures with RSI Trend System Note: Trade summary includes data from January 1, 2000, to December BL 2009, and assumes $10 round-turn deductions for slippage and commissions.

Source: CQG, Inc. ©2010. All rights reserved worldwide.

Finally, byusing the same RSI Trend model and sun'ply tightemn'g the exit criterion from stops atthe previous three-bar low (orhigh)tothe previous bars’ low (orhigh), our average trade duration is'cut from Six'tothree trading days (see figure 3.5). Usm‘g CQG, the programmm'g code forshort-term RSI trend is“written thisway: Long Entry: RsI(@,9)[—1] > 65 Short Entry: RSI(@,9)[-l] < 35 Long Exit.LOLeve1A-M«_\ZI’

v _V

r—‘iz—“uofi“"fiL“To"

FIGURE) 4.I2 Daily Chart ofMay 2006 CME Group Soybeans with Bollinger Band Difference Showing $0.90 as High Volatility and $0.27 as Low Volatility Source: CQC, Inc. (92010. All rights reserved worldwide.

83

Current ht‘gflh

Prev: ous close

Prev: w: ficloae

True tango

\ True range

\

Current

Im

FlGl'RF. 4.13 Daily Chart of October 2010 Pit-Traded CME Group Live Cattle Futures Showing True Range

Source: CQG, Inc. ©2010. All rights reserved worldwide.

Now that we have defined the asset’s true range, we can examine Wilder’s volatility indicators. We willstart with the simpler volatility calculation, average true range. Theaverage true range. takes themoving average oftheasset’s true range over a spec1fi'ed period. Typically, the average true range, or ATR, iscalculated on a l4-period simple moving average. ’l‘hat stated, Ihave commonly seen itcalculated on a lO-period simple moving average as well as a l4—pen‘od exponentially weighted moving average. As noted earlier, it’snot magic;it’sjustmath. Different periods andweightings 0fthemoving averages tellslightly different stories regar'ding thehistorical volatility ofan asset (see Figure 4.14). The main limitations mentioned earlier regarding Bollinger Band difference also apply toaverage true range. Torecap, the problem with both ofthese mathematically derived volatility indicators isthat the numbers generated bythe indicator have little or no significance to other assets, nor do they have sigmfi'cance for the same aS'set for either a longer or shorter tim'eframe, or even for the same asset over different periods of a 24-hour trading day. Wilder’s other historical volatility indicator, average

1M

TRADING TOOLS AND TECHNIQUES

2 l l E E l i;

14-day ailplo M, vol-tilily fall: later In mnlh

IA"_V\~N_1O—day oanontlll M, ' ‘tvolatility lull: lrm puk quick.-

FIGI RE 4J4 Daily 20l 0Chart ofApple Computer Comparing 10-Day Exponentially Weighted ATR with 14-Day Simple Weighted ATR Source: CQC. Inc. ©2010. All rights reserved worldwide.

directional movement index, or ADX, solves these problems by giving us an indicator that isbounded, meaning that itisapercentage oscillator that cannot go below zero or above 100. Consequently, its volatility readings are applicable. to all assets and on any timeframe. This is illustrated by Figure 4.15, in which high and low volatility in'cash eurocurrency isobjectively identified with 4—day ADX readin'gs above 75 and below 25. The average directional movement index is' a moving average— commonly set to 10periods—0f directional movement index, or DMI. DMI isa momentum indicator that compares the current pn'ce with theprevious price range. Spec1fi'cally, DMI measures positive net directional movement or +DI vis-a‘-vis negative net directional movement or —DI over aspecified lookback period (commonly set to 10 or 14 pen‘ods). Wilder defines positive. net directional movement as an interval of time in which the majority ofdirectional movement is higher than the previous time period. By contrast, negative net directional movement isan interval oftime in‘which the maj'ority ofdirectional movement islower than the. previous time. period.

on the Cyclical Nature ofVolatility 85 Emitmum

200

160

'

500

no

"Lou‘"93!".1—"61"}T

’

‘

in

” High volatility

16

i

--

12

IO

23

:i

-

'0

25

I.

FIGURE 4J5 Daily 2010 Cash Eurocurrency—U.S. Dollar Chart with 4-Day ATR. Bollinger Band Difference, and ADX Source' CQG, Inc. ©2010. All rights reserved worldwide.

Forpositive pn'ce intervals, the. formula is: +Dl = (+1)] :—TR) x 100 Where +Dl = positive Directional Movement an'd TR = True Range Fornegative pn‘ee intervals, the formula is:

—DI _— (—Dl .—'TR) x 100 where —D1 = negative Directional Movement and TR = True Range Wecan then calculate DIdiff and DIsum as follows:

DIdif'f = l((+DI) — (—DIDI and

DIsum = ((+DI) + (—DI))

TRADING TOOLS AND TECHNIQUE

86

Directional Movement Index, or DX, iscalculated as: DX = (Didiff .—' Disum) x 100 Fimtlly, we calculate ADX as follows: ADX : SUM[(+DI — (—DI)) —.' (+DI +(—DI)). N]:— N

-i-......i .. “Jan”;

“A-

where N= the number ofperiods used in'the calculation. Although these objective, mathematically derived volatility indicators are extremely valuable in providing us with indicators that can augment both trend-following and eountertrend systems, classical techm‘cal analysis is also useful in giving us a sense. of where we are in the volatility cycle. For example, Figure 4.16 shows a prolonged, multi-month pen’od oflow volatility in ("ME wheat futures with a classical rectangular formation ofsideways, hon‘zi'outal support, and resistance, which ultimately resolves itself in a breakout to the upside an'd high volatility. By contrast, Figure 4.17 offers ashorter-tenu View ofthe cyclical nature ofvolatili"ty in' which a two-day verti'eal' flagpole—of high volatility—leads to a 14~day,

*W

Buy breakout , trending act Ion

Narrow, side'ays action Low volatility

r

y“

7

M

Jul

77

h

FIGURE 4. I6 Weekly 2010 Front-Month Continuation Chart ofCME Group Wheat Futures Showing Breakout from Low Volatility Source: CQC. Inc. (cf) 2010. All rights reserved worldwide.

r (finalizing on the Cyclical Nature ofVolatility

87

Ponnunl (lw volatility)

Jilin». 2~da (hug

llng to vol.rllnly)

HG! Rl.‘ 4J7 August 2008 Daily Chart ofCME Group Crude Oil Futures Showing Cyclical Nature ofVolatility Source CQC, Inc. to 2010. All rights reserved worldwide.

low-volatility consolidation period in August 2008 CME Group crude oil futures Among the more popular of the objective, malhematically den'ved \'olatilit\_' indicators are historical and implied volatility ofthe traded asset. Regarding lllSlUl'it'dJ' \I'olatility. it isimportant to remember that Bollinger Band difference. ATR. and ADX are in fact all calculations of historical Volatility of the asset. In addition to these at})rementioned indicators, I Al's) use astatistical titethuremt‘nt of historical volatility cal'culated as the stzuulard deviation from the mean ofthe most recent 20 bars. The main limitation to all measures oflu'storieal volatility isthat they are lagging indicators. in other words, because they ar'e den'ved from historical data, theyare always telling us more about the volatility ofthe asset in the past as opposed to its current volatility. lmplied volatility addresses this problem bycult-mating volatility of asset based on current option premiums. ltconsequently tends to respond to changes in the volatility ofthe asset more quickly. Although there we numerous formulas for calculating im— Dliedvolatility. the inputs 1use are premiums of both puts and calls struck dI‘-the-money as“ well as three strikes above and below the at-the—money

TRADlNG TOOLS AND12cm ’. -..,i

.

"'9." ..

so." look lol [our I. _’ n.‘~ loll fr“ W"9 M W; *‘ ‘ _V‘a

V

r?

V

ltl‘fi|‘yljlr‘l|ll

9 I 9 I.

:1,We

loll (rm IV pout la volalrlnly cyclo

’-

rune

Iqliod vol-tility lull: iro- pout

historical volnl I Ifly

lull: Ire. pool.

w./ I. I- in historical volatility

20.nok low in inpliod volatility

I’M-IRE 4.18

Weekly CME Group Natural Gas Futures Showing Implied and

Historical Volatility Source' CQG. Inc. © 2010. All rights reserved worldwide.

options forthe three sen‘es ofmonths closest to expira'tion.‘ Figure 4.1815“ pan'icularly useful because itnot only gives us a chart ofhis'ton'cal volatil ity, but perhaps more un'portantly, itcharts the history of—and therefore the trend of—implied volatility. A final, cautionary note regarding the cyclical nature of volatilin; as it relates to implied as' opposed. to thton'cal volatility indicators: When markets an'ticipate the release of major fundamental reports (for example, quart-erly earn'ings, central bank policy statements. crop reports. and so on), implied and histon’cal volatility indicators tend to diverge. Hts ton'cal volatility tends to decrease ahead of such reports as cash mrd futures traders stand aside because of market uncertainty. By contrast‘. options traders will purchase calls and puts—thereby increasing implied volatility—as protection agams't post-report spikes involatility irrespective ofmarket direction. After the news event‘s release, market participants assrm'ilate new u’r formation regarding the asset’s value, and his'torical volatility tends to 111'crease. On the other hand, the behavior ofimplied volatility is much IPS“ predictable after the release ofmarket-moving news. Ifthe news divenlf‘d

r

r 3

Capitalizing on the Cyclical Nature ofVolatility

30 AM ES! 48 values at 05' Ewtmnt

Rowrl \\‘

I

#Prorowrl Poattawrl IInger Band volatility totonoo 5‘» Mn

“GI Ill'.‘ 4.") One-Minute Cash Euro—U.S. Dollar Chart Showing increase in Historical Volatility after 8:30 AM. Release of Monthly U.S. Employment Report on October 8.2010 Source: CQC, Inc. (9' 2010. All rights reserved worldwide.

drmnatically from pre—release consensus opinion, implied volatility will increase as options traders seek to panicipate in the embryonic phases of the fat tail event. On the other han'd, ifthe news was within the rmige of mar'ket expectations, option premiums—‘as well as implied volatility-will decrease despite increases in historical volatility (see Figure 4.19).

BlII.IDI\(-‘ I'IDSI'I‘“ I‘.‘ l‘.‘\l'l‘.‘(."l‘\\(2\ “0|!_I‘.'I._S_ “I'I_'_II“ \0|. \'I'II.I'I‘\ I\I)I(li\'l‘0ll8 As stated earlier. we can use volatility indicators in two ways: first, to identity‘ periods oflow volatility dJ'ld employ trend-following indicators to take advantage of when the market breaks out of its low—volatility cycle; and second, to identify periods ofhigh volatility and use countertrend tools to pmicipate inthe asset’s reversion to the mean. Although there isa vast multitude ofways in which to combine trendfollowing tools with volatility indicators so as to participate in instances

TRADlNC TOOLS AND ltt’HNlttUtS

90

in which the market breaks out of low volatility, Iintroduce the render to a basic example here. First we need a ninthetnaticutly objective tool lot detennining thatWe we ina low-volatility environment Although we mnld use (my of the tools mentioned earlier in this chapter, Wt‘ will ltn‘ttn on AUX since its definition of low volatility is universally applicable to any asset an'd all trading tiniefran'ies. While there is nothing lttttgh‘d‘l about it ltl-period AUX, we will stick with itsimply because of its popultn‘tty. We will also define [me volatility us times during which the It)period Alix gives areading below 20. Next‘ we need a mathematically Objective criterion to detine u ltll‘itlt out from this low~volatility environment. Again, there are innnnn-mhle ways ofdefining thisbreakout, but forsiinplicity’s sake Itit-tine nasatclose above the. Z‘O-pen'od upper Bollinger Band or below the 20 period lower Bollinger Band. Finally, we. need a matheinatically objectiw risk manage ment rule along with a criterion for exiting with protits. For the sulw ot simplicity, after entry we, will set a trailing stop for long positions art the lowest low ofthe previous three trading days and for short positions at the highest high ofthe previous three, trading days (see Figure 4.30). Using (IQG, the programming code for this simple trend-following sys tem iswritten this way: Long Entry'. Close(@) [—1]

> BHI(@.Sim,20,2.00) [wll AND ADX(@,10)1421‘ 20

Long Exit, set "Price"

field to:

LoLeve1(@,3)[—1] Short Entry.Close(@)l-1]< BLO(@,sim,20,2.00)[—1] Short Exit,

set 'Price'

AND ADX(@,10)[~2]

c

20

field to:

HiLevel(@,3)l-1l

For countertrend systems that use volatility indicators to help identify setups with a high probability ofsuccess, we need evidence oftrend ex haustion. Although this does not guarai'itee protection against a false trend reversal signal, at least it decreases the odds of such an occurretn-e In other words. despite the 10—period ADX signaling high volatility bygiving a reading greater than 50, the asset could generate such reznlings day at ter dayas we contm‘ue to fight apersistently trending market. llyadding it

Capitalizing on the Cyclical Nature ofVolatility

e

W‘JIJ-L1+"

l-‘ll-‘l RE 4.20 Equalized Active Daily Continuation Chan for ICE Sugar Futures Contract with ADX Low-Volatility Breakout System. Data Show Results from January 1.2000. to December 3]. 2009 Note: Trade summary includes 5l0round-turn trade deduction for slippage andcommissions. Source: CQC. Inc. ©2010. All rights reserved worldwide.

cn'ten'on that ahigh-volatility market thatistrending lower needs toViolate thepren‘ous day's highs before signaling abuy (orViolating the pren'ous day’s lows before signaling a sell), itreduces this risk. We m'll ren‘sit the dangers ofanticipating the signal ingreat detail throughout (‘hapter 9.but fornow it is essential to avoid developing high—volatility trading modesL that tryto pick tops and bottoms merely because the volatility indicator signals a high volatility reading. In addition to requiring that ADX signast ahighvolatility reading, Iadded apercentage oscillator (Relative Strength Index) to ensure that the asset has infact become overbought or oversold. Since we we fighting the trend. our stop loss exit isset to thepren‘ous bar’s low forlong positions dl‘id prew'ous bar‘s high for short positions. This isatighter risk management criterion than the three-bar high or low used inour low-volatility breakout model because counted rend signals are' generated in a high-volatility—and therefore high-risk—etm‘rotunent. l-‘t‘nal'ly. ifthe asset. reverts tothe mean. our model exits with profits at thepren'ous bar's 20—period simple moving average (see figure 4.21).

,‘

TRADING TOOLS AND TECHNIQUES

0 70 day W8.” Maple IA

7 plutonium: .. IMmIWDI-l'

{I II II l I 47 non-noun" “(mama-sol' '("‘l‘--IOI-‘ u . It . . I t l ‘ I‘ I v . t -o 0‘ I J: 4

r

I"I(-‘l'RI-.‘ 4.2! Cash Daily US. Dollar—Swiss Franc Chart with ADX High-Volatility Countertrend System Data Show Results from January 1. 2000. to December 31. 2009 Note: Trade summary includes Sl0round—turn trade deduction for slippage and commn’ssr’ons. Source CQG, Inc. (C 2010. All rights reserved worldwide.

Using CQG, the progmmmm’g code forthis lu'gh—volat111"ty countertrend system is written this way: Long Entry'.

mX(@,lO)[-1] > 50 m H1'gh(@)[-1] l@,9)l-l] < 35

> H1'gh(@)[-2]

m RSI

Long Ex1't.LoLevel(@,l)[—1]

OR ml@,51'm,20)[—1]

short Entry.-

ADX(@,10)[-l] > 50 AND Low(@)[-1] (@,9)[-1] > 65

< Low(@)[—2] AND RSI

Short Ex1‘t.' H1'Level(@,l)[—l]

OR M(@,Sl’m,20)[-1]

93

Capitalizing on the Cyclical Nature ofVolatility

Afinal' word of caution regar'ding the incorporation of volatility m'dicators into cmmtertrend trading models: Be. careful. Despite the protective {ml-safe critm‘ia that were built into the high-volatility mean reversion system, because these we countertrend systems, they are by definition subject to fat tail event risk. In other words, when a market shift's—without Winning—from a normal' trending environment to a parabolic trending environment, these systems are subject to risks, which could potentially overshadow rewards, sometimes dramatically. For example, on January 8,1980, March 1980 (U‘omex Silver traded below the previous day’s low, the 9-day RS] reading was adramatically overbought 92.82, and the 10-day ADX was" at82.27. The criteria required to sell the market were met so our coun— tenrend system would have sold the January 9 open at $33.50. Although our protective buy stop forJan'uary 10 would have been set at the January 9th high of $33.50, silver, unfortunately, opened locked lm‘u’t up that day. Worse still, the mar‘ket traded locked limit up through January 21. Finally, on January 22,our sell stop would have been elected at $40.50, for a loss of $35,000 per contract (see Figure 4.22). Admittedly, itispossible for trend-following traders to endure a fat tail event. risk scenario, but it is much less likely than for those consistently

nooo

kaod llllt up buyatop $33 50 no oxoou edl

-.\

.

Sold 0 333.50

+' 4..

+

lBuy slowp {£28310

MIN \0 II 14 I5 18 I7 II 2! 22 3

EFlGliRRIt.‘4.22 March 1980 Daily Chart of Comex Silver Futures Showing Fat Tail vent isk Source: CQG. Inc. ©2010. All rights reserved worldwide.

94

TRADING TOOLS AND TECHNIQUES

using countertrend trading models. Despite the fact that bydefinition 0pt.ion premiums will be most expensive when markets signal highvolatility, buying options remains the safest antidote to problems of fat tail event risk forcountertrend traders. Oras Iliketosay, everyone wants tosurfthe wave. but no one wants to surfduring aCategory 5hurricane.

FIN-AL THOUGHTS Volatility studies are indispensable in defining current ns'k and reward as well as helping to model future risk and reward through analysis of where the asset is'in'regard to its historical volatility cycle. Also, although it is possible to develop robust positive expectancy trend-followm‘g and countertrend trading models without the volatility indicators showcased throughout this chapter, whenever asked how to make a model more ro bust, Ialways begin byexamining these tools.

I‘llill'lll ' ‘F11’i

(I II \I"l‘l'.' R 5

Trading the

Markets and Not the Money

What does a man do when he sets out to make the stock market payfora sudden need? Why, hemerely hopes. He gambles. He therefore runs much greater ns'ks than he would tf'he were speculating intelligently. —Edw1n' Lefe‘vre

Why dospeculators enduptradln‘g the money instead ofthemarket’s dynamics? This chapter explores various ways speculators trade the money, including superimposing artificial monetary profit targets onto the market. irrespective of conditions, exiting profitable trades because of monetary—as opposed to market-derived—targets, and placingstops too close to entry price because of fear of loss irrespective of market volatility.

TEN THOUSAND DOLLARS ISA LOT 0F MONEY! Bernard Baruch once stated, “Nobody ever lost money taking aprofit.” To which Irespond, “No, they lost all their money on the next three trades.” If allwe do istake small profits, then we have nothing to offset the inevitable losses. Although many books advise traders to cut losses and let profits mn, the problem is' we. always think about the money instead of market dynamics, and so when we have significant unrealized profits we translate

95

96

TRADING TOOLS AND TECHNIQUE

them into monetary terms and end up em’tin'g trades prematurely. Chap ter 6outlines specrfi'c methods for enabling traders to simultaneously take money offthe table while letting part of our position run. Bycontrast, the goal of this chapter is showing the pitfalls for intermediate to long-term trend traders mnot all-owing profits to run. It is only natural for speculators to translate unrealized profits m’to monetary terms su‘ice we. all entered the busm’ess of tradln'g to make money. But that which is psychologically natural and comfortable leads to failure. That which is tmnatural and uncomfortable leads to success. Stated more simply. tlu‘nkm'g about the money is poison. Trade the market, not the money. The markets offer examples every year oflarge trends morphing 1n'to monster trends. yet year after year inexperienced traders contln'ue to think about the money and settle for small gains' despite huge profit opportunities. One of my favorite examples of large unrealiz'ed profits turning into obscenely large realized profits is the 2008—2009 CME Group natural gas futures market. On July 8,2008, the market broke significant. support and many technical traders sold at the $12628 level. Within aweek, themarket broke. the $11.53 level. wlu'ch represented an unrealized profit of$10,980 per contract. Ten thousand nine hundred and eighty dollars is a lot of money to many of us, especially when generated over five trading days (see Figure 5.1). But the market does not know what “a lot of money" means; it only knows that itis'gom‘g from $12628 down to $3.155 over the next nine months and itisgom'g to offer long—term trend-followm'g traders well over $80,000 per contract. The question is,are you gom'g to take what the market is offenn'g byallowm'g its own internal dynamics to detennm’e when you Will exit with profits, or are you gorn‘g to superimpose an artificial ceiling on profits based on an irrational in'ternal psychological bias regarding how much money over What period oftime istoo much, too fast (see Figure 5.2)? Whydo traders cut unrealized profits short instead ofletting them run? Fear of leavm'g money on the table, or worse, fear of allowing a significant unrealized profit to turn Into asignificant realized loss. Fear ofleann‘g money on the table is quite debilitating to the psyche oftraders because you had the opportum‘ty tocapture larger profit levels and failed tocapital— ize on it. Many times as the market falls from its highs, traders m'll place their exit orders at these old lu'ghs. At this point, one of two things can happen: First, the market can plow through those previous highs, enablm'g us to minimize our regret byselling the highs. This typically happens be cause. the market is destined for significantly higher levels and we soon regret our premature exit. The other possibility isthat the. market fails to retest its old highs and our rigid focus on exiting at the old high prevents us from protectin'g sigmfi'cant unrealized profits bymoving stops tological

97

Trading the Markets and Not the Money

hrkol brats in rt 0 $17 628 'w

l l g l

2‘!“

l

Harkol omit: tn 53,5 an UNIOIIIIM prof-t ’t at 810,980 00 g ‘ 3

FIGURE SJ August 2008 Daily CME Group Natural Gas Futures Move Over Sl0,000 in Five Trading Days Source: CQC. Inc. to20l 0. All rights reserved worldwide.

technical support levels. Itisthis second scenario, in which our focus on exiting at a specrfi'c profit level blinds us to risk of reversals that can result inallowing significant unreal'ized profits to turn into significant realizedlosses. Ineither case, we are focusing on the asset's price u'r'espective ofvalue. The solution isto realize from the outset that we will almost never sell the highs or buy the. lows and instead offocusm’g on the elusive perfect entry or exit price, shift our focus onto tradm‘g based on the market's dynamics (asdefined bysupport, resistance, and volatility). Oras Iliketosay. “Always trade value; never trade price.” Since fear- ofleasing money on the table leads to our premature exit ofprofitable positions. our positive expectancy model should include exit rules based on the dwizum'cs ofmarket action instead ofartificial monetary price tar'gets. Iffear' ofal'lovnn'g aSignificant unrealized profit to tum into asignificant real'ized losS‘ forces us to exit profitable positions prematurely, then incomorating rules for mon'ng Stops to breakeven as soon as the market moves significmttly in our favor should be built into our positive expectancy models.

TRADING TOOLS AND TECHNIQUES

krkol hrukn $12 628 supmrl

._._Wim_c-

0.10.9.0 w

-lh‘Ivkol brooks $11.53 suprrl

1m

now

8!» ed 0 $444 groil't of 2 040 w

M- 7-14.91 ll’.;';"'ufl 'Ir u

v

y

iwi

a

an -'

o: my"

NV

y

_§.-

J

—

Q

FIGURE 5.2 Rolling Front-Month Weekly CME Group Natural Gas Futures Move Over $80,000 before Achieving a Significant Technical Reversal Source' CQC, Inc. ©2010. All rights reserved worldwide.

Another argument often stated in'relation to trading the money is”that “There must be some price at which you would be willin'g to exit your position." IfIam short the asset and using an in'termediate- to long-term trend-followm'g model, then yes, the price Iwould be willing to exit would be zero. Otherms‘e, no, there is"absolutely no arbitrarily derived monetary profit level at wlu'ch Iwould be Willing to sacrifice my profitable trendfollowing positions. This isbecause Ihave trained myself to be absolutely flexible regarding market prices. Our natural bias as traders is set byre cent or current pn‘ces. Instead ofbiases related to current, recent. or even historical prices, Ihave train'ed myself to imagine the market tradm'g at any and every pn'ce. Itisbecause Iunequivocally accept that any price is possible that Inever hesitate in'placin'g stop loss orders after entenn'g apo sition, because Iacknowledge that certain possible prices would result in' my trading account blowing up. Ifwe admit that any pn‘ce ispossible and that. certain prices would result inthetermination ofour careers as traders. itis illogical for us to Simultaneously dismiss the possibility ofprices that could catapult our trading careers to new levels ofsuccess. Ofcourse. jllSl as our account would blow up long before the market reached its ultin'iatt>

r” Trading the Markets and Not the Money

99

highor low, bythe smile token we will never capture the ultimate high"or low when exiting on profitable positions... but as theysay. “You can make a heck ofa lot ofmoney bybeing less than perfect.”

BABY NEEDS ANEW PAIR OF SHOES The phrase “Baby needs. a new pair of shoes" on'ginated in the casm‘os; Specifically. itisthe mantra ofdice-tossing craps players. The fact that thsi.‘ phrase isattributed to those with the odds skewed against them is especially instructive to speculators wanting to trade like a casino. Focusing onaspecific monetan'; goal, whether itisgenen‘c. like a thousan‘d dollars aday, or specific. like a $3.500 mortgage payment. is equally pmso'notts to successful speculation because‘ we are sutwn‘mtmsing mi artificial goal upon market actin‘ty. The mar-ket does not know or care' whether we need tomake $300 or $3million. and our thinking about the market intheset mtificial temis blinds us from what the market islikely tooffer basel upon its intemal dynai'nics (volatility. suppon. wid resistance). To trade like a casino. we need to think about profits in tenns ofprobabilities instead of persosnal monetary needs. Imu extremely fan'iiliar with the “baby needs a new pair” ofshoes“ approach tospeculative tradmg' because when Ifirst bought aseat on theNew York Futures Exchange in 1987, thestated goal ofour hedging corporation was togenerate $2..500 per week from tradtn'g. We were tlu‘nkm'g about the money. not the markets. Needless to say. that monetary goal acted as an albatross around my neck, blinding me to market dynamics and opportunities. Itseemed like a logical' approach for a busm‘ess model. but tradm'g is'not a logical' enterpn’se where Xnumber of hours ofhard' work translates into Ysalary. Itisinstead a business ofattuning onese:lf to the emotions ofother market participants and flowing sean-ilessly with their waves ofgreed. fear. and boredom. Sometimes the market offers $2.000' a week: sometimes itoffers $500. sometimes $5millmniQVe are habituated to think‘ interms ofsteady monthly flows ofincome because our expenses come in‘ steady increments . . . but the market does not care about us. our bills. or it“ ourbabies need new pairsofshoes. Forget about monetary goals; just take what the market isgiving. Ofcourse. now more than 20year: later. Iknow from pcu-‘nful experience this truth oftrading the markets and not the money. Now; wheneyer lstart amonth with gangbuster profits. fn‘ends that do not understtuid the nature ofmar-kets will say, “Wow. it'sMay tiand you are alreautv up 9percent {orthe month? You should stop trading so you don‘t give itall back.‘ “"5" (00. is’trading the money dl’ld not the mar'kets. The markets do not

is...

"“fl

TRADING TOOLS AND TECHNIQUE

know that you are already up itpercent. haVe had tlve winners ina row. an. overdue for a losing trade. zutd so on. nor do they care. They are olTeting you the opponunity to participate in their game ofprobabilities .."-l houma day. every week. every month. Whether you personally s‘ttpt‘t'tmpose some artificial limitation upon these probabilities isentirely up to you. It hasat» solutely nothing to do with the markets and what they are offering. At the other end of the irrational psyclu)logical spectrum isthe elicho‘ that you can be more .‘tggressive or reckless after signiticant trading profits beczmse you are now pl.'tying with the house's money. This. too. istrading the money instead of the markets. Since the probabilities ofour positive expectancy trading models do not mirtu'ulous'ly' improve after profitable trades. We should not abzmdon our adherence to rules of risk management. As soon as you place the trade, even before exiting with re;tli1.‘etl profits. those lllll't‘illll.‘(‘(l gains are your money and need to be treated in the same casino pzmuligm manner as all monies in your trading account. It is probably even more common to trade the money after a loss or series oflosses. We suddenly abandon market dymuuics :uul probabilitim in hopes of regtu'ning the breakeven level in account equity. This part'icular irrational attitude toward trading isespecially dzmgerous to speculators because we now not only trade the money. but also abandon the goal of winning in favor oftrading not to lose. lnstezul. after a string oflosses: we should ask the following questions: “Are we adhering to a robust positive e.\'pectm1cy model?” dJ'ltl “Are we continuing to obey stringent rules ofm’k lllal'lagt‘llk‘lll?“ lfthe answer to these questions isyes, then we should continue trading the mar-kets in exactly the same manner as we would afiera profit or even a string ofprofitable trades.

'I‘RV\DI\G “I'I'Il Sllfllln‘ll \I0\’l‘.‘\ Whenever we place stops based upon monetary considerations instead of the dynamics ofmar'ket volatility, we are also trading the money md not the mar'ket. There isa caveat to this rule ofmonetary stop loss placement. namely that we should not risk more than 1to 2percent of assets under management on any particular trading idea. At first glance. the 1percent rule looks a lot like trading the money. The distinction is that we need to overlay the 1percent rule on top of our market aimlysis so that we out" execute trades in which market -derived stop levels are less than 1pert‘t‘l“ ofassets under management. In this way. we adhere to prudent risk lllcll‘l‘ agement parzunetets while simultmteottsly allowing the market :md not tht‘ money to determine where stops should be placed. Ifthis means tmditltl 200 shar'es of Microsoft at

a share instead of 200 shares of 60081“ “I

7" I01

Trading the Markets and Not the Money TABLE 5.]

Volatility-Derived Position-Sizing Limits

M

10-Day ATR

Max. Position Size

E-Mini S&P 500

$775.00 $1,137.50 $2,920.00 52,01 0.00 $968.75 $10.18 $0.48 Sl.840.00 $l,300.00

100 4,000 100.000 100.000

Corn Gold Crude Oil UST-Notes GOOG MSFT EURUSD AUDUSD

NOO—‘N

Asset

Ten-day ATR calculated on November 18, 2010, for a $100,000 account. using 2percent ofassets under management rule. Note: Position sizing shown does not account for correlations between assets. Source: CQC, Inc. ©2010. All rights reserved worldwide.

$600 a share, then so be it. Itis better to allow the probabilities of your positive expectancy model to play out in your favor than to superimpose an artificial monetary stop loss level onto a higher-volatility asset simply because it15'more “exciting” (seeTable 5.1). Placement ofstop loss orders at monetary thresholds irrespective of theasset’s volatility isknown intrading vernacular as “trading with scared money.” Because they are afraid of large losses, traders place stop loss orders too close to entry prices, therefore Virtually guaranteeing their' en— durance ofnumerous unnecessary small losses. We must instead be willing to accept the possibility oflosses in‘ order to enjoy the profitability ofpositive expectan'cy models. This not only means initial stop loss placement based upon market-derived volatility levels, but also moving ofstops only after market-derived indicators justify their adjustment.

TIME IS MONEY Ifyou art1fi'cially superin'lpose atime-based exit criterion onto your trading model, you are bydefinition no longer exiting exclusively based upon the dynamics ofthe asset. To me, this is eerily similar to trading the money ms'tead ofthe market. Although Ifreely admit that shorter—term traders Want to avoid overnight or weekend event risk, itmust simultaneously be acknowledged that itis counterintuitive and a wholly artifi'cially im'posed Constraint to exit trades with unrealized profits simply because aclock has ticked one second beyond 4RM. Eastern Standard Tim'e. Exiting profitable

TRADING TOOLS AND TECHNIQUES

I02

trades heennse ofthis zutitleiailly derived time enlist mint flies in the face of cutting [US$08 and letting pt‘ullls run. 'l‘hut slatted. 1, ton. have exited short term ineun reversion system trades heeause ot'vnrinus tinte related issues Hllt‘h its lzu-k ofovernight liquidity. weekend event risk. and so on. 'l‘herel‘ure, lhave no problem with exiting short term nieiui reveminn trades liet'nnse the (‘Im'k has ticked forward beyond it predes.‘igniited entnl‘l' point. lint Ido reeognize that using the elm'k in this fashion isit t'orni nl‘lritding the money instead ofthe "Wkets. lconsequently must ensure these tittie~~h.'t.s'ed exit models enjoy higher winning pereentziges.‘ than models that allow prullts to run in order tocompensxwte for l_\"pi(‘:|ll\.' inferior average prolit to :weruge loss ratios. ngre 5.2! shows it short-lertn trend‘l‘ullowing sysll‘ln in whieh an' exit is triggered when the market breaks the highest high or lowest low of the previous three trading days. Now look :it Figure 5.4, which is at eoitipiu‘ison of our original short-term lrend—following system to a modilied Version wherelw ti tiine-den'ved exit en'terion has been added

Mignon! high at provloul 3 day.

Ull‘.l'l¢ll‘ .

"moonwal-

1“

wit-Own unIfMIKI‘NI wtmurlim -

[Minimum-vvol I “MIMI to. ICORIIIWII -

mar-tuning. I “cutout-i

FIG! RR 5.3 Equalized Active Daily Continuation Chart for CME Group Soybean Futures Contract with RSITrend System Where Stop isPlaced at Lowest Low or Highest High ofPrevious Three Trading Days Note. Trade summary includes data from January 1. 2000. to December 31. 2009. and includes SlO round‘turn deductions for slippage and commissions. Source, CQC. Inc. to20l0. All rights reserved worldwide.

Trading the Markets and Not the Money

‘

. 1 LP . A. ctr-“#1911 Iona-unenl-

a.,___. um uncoun-

..

_’VnuIVrmnI. > Milan 1‘ lemma-Io

limo"

-

.

'- "'xt‘imm3chunu«(Cancun-u- roliltuuxnrn-

.

10°"qu normal DrmvnllonConuciim - - icontoclinl -

. -

-~-

ioHILnuultoorconllmnou canton-unuinf-luring- -

i—

-

‘ c‘tn‘l’ulf'voi‘il 5—7 m " immiliottl-

-

I03

.

Into-mun! lolmllllmIConIocIrnu . rcoouclyn.

$68 4743

mnlLouulioorconlilnnorl -

I’lGl RE 5.4 Equalized Active Daily Continuation Chart for CME Group Soybean Futures Contract Comparing Original RSI Trend System with Modified Version with Time-Based Exit Criterion Note: Trade summary includes data from January 1,2000. to December 31. 2009. and includes 510 round—turn deductions for slippage and commissions. Source: CQC, lnc. Co)2010. All rights reserved worldwide.

that forces m] exit after the. position has been held for more, than one. trading day. Using (,‘Q(}, the prograi'nming code for thistime-based exit criterion is written this way: Long Exit: BarsSinceEntry(@,O,All,ThisTradeOn1y) > 1 OR Price field set to:

"LoLevel(@,3) [-11"

Short Exit..- BarsSinceEntry (@, 0, All , 'I‘hisTradeOnly) > 1 OR Price field set to: "HiLevel(@,3)[-1]"

Looldng at the “PercentWinners” row, notice how the time-based exit improved our winning trade. percentage from 40.62 to 56.99 percent. on the other han‘d, looking at the “AvmgeWin” and “AverageLoss” columns, We see the time-den'ved exit criten'on simultaneously eroding the average profit toaverage loss ratio from 1.88:1 to 1.10:1 (see Figure 5.4).

"M

TRADING TOOLS AND TECHNIQUES

FINAL new!" Insummary, 1m-knnwk-dgc- thm van"ous speculators (-mplny dim-rent ham mum-9s ham-d upnn n‘mvfrmvs .m'd market. action they dr'v mus! ('nmfnn. al'm- oxplniting (for (-anplt', mean n-wminn), and lhvrvfurc- sumo oftho vivws (NpUUfiu‘d inthis ('hapN-r may not seem applicahlc- Inthem. Nt-w-nhe. lms, dmpitv tl'w matrn’dl' being most. pt-ninvm m l'nlvnnwlialr— mu! longlrrm trvnd trmlvrs, Idfl'l (-nnfldum all .s'ywrulamrs.‘ ('d!’l hem-m from undatstwuh'ng the pitMls intrwh’ng the money inswwl 0fthe mar'kvt, mpwially 43' itroldu'r-s. m mamgn’hg n’sk.

.vn-4 ‘ l.ll\lll.fl

‘~ D

Minimizing Trader Regret

Isee itallpmjfectly; there are two possible .91,"(mum/ts ——(me can (,n'ther do this or that. My fumes! optima/'4 andfn'mutly advice. is(M [)0 i!or (to not do iI/——you will regret both. -So‘ren Klt.‘l'k(,‘g,ddl"'d

an an'ything be done to make adherence to disciplined, rule-based trading easier? Since the most common reaso"n for ai'ian'donment of discipline is regret over losses or missed opportunities, tth" chag'iter offers van'ous techniques to counteract these self—destructive uendencies. Particular emphasis isplaced on specific techniques to minimize' regr;et for both trend-following as well as countenrend traders.

'I‘IIF. SOFTER SIDE OF DISCIPLINE Chapter 3examined unwavering discipline as aprerequis'ite forour adherence to positive expectancy models and robust risk man'ag'ement methodologies. Here we will augment that work with what Ilike. to call the wfter side ofdiscipline. These are techniques that make adherence topositive exli‘ectancy models and risk management more palatable. These te(_-hm'ques are consequently not necessarily intended to make positive exrmctancy models more robust, but merely to make sticking with them easier ms'tead. Because each ofthese tools tries to mu'iimize the emotion of regret that inevitably accompan'ies any and all trading decisions, my umbrella term f0rallthese techniques isregret vizi'm'mizatim1. Toexplain the technique.

“I?!

i

“)6

TRADING TOOLS AND TECHNIQUES

let its retum to the opaque urn introduced in Chapter 3. Remember that the urn contained 57 green marbles and 43 red marbles, and therefore the probability favored our betting on the green marble. The problem in con. sistently betting on the green marble was that occasionally we would pull out a number ofred marbles consecutively. But what ifwe could cut the green marbles in half? No. itwould not increase the probability ofdrawm‘g out a whole red marble, nor even the number ofwhole green marbles. It would instead increase the number oftirn'es we pulled out a green pieceas opposed to ared piece. This' is' what all the regret mimrnizat"'ion techniques try to accomplish and what was alluded to earlier. The other common denominator linking the techm'ques explored throughout this' chapter is'theassumption that traders are sufficiently capital'ized so that the trading ofmultiple contracts will not result in'their ns'king more than 1to 2percent ofassets under management This assump tion of adequate capitaliza'tion introduces the question of how much is' adequate—defined here byusm'g the 1to2percent rule—as well as the ad vantages and disadvantages in'respect to levels ofcapitalization.l Another disclaimer isthat none ofthese techniques is'suggested as the onlyway of han'dling regret mimm'izat'ion iss'ues. They are offered instead asjumpm‘g off points from which readers are encouraged to do further research. My final disclaimer applies only to trend followers, namely that the ability of trend followers to use regret miru"miza'tion techniques ispredicated on the mar'ket moving in our favor after entry. In instances in' wlu'ch we enter a trade and the market immediately moves against us, we will be stopped out with a loss: an'd will not have an opportunity to use the techniques de scn'hed further on.

ISSUES FOR TREND FOLLOWL‘RS Feelings of regret can' any." for trend-following traders in' a multitude of ways such as allowing sigm'ficant unrealized profits to turn into sigm'ficant realized losses, exiting trades with statistically significant profits only to watch from the sidelines as the market moves relentlessly in‘ the direc‘.tion ofour prematurely exited position. as well as lettin'g small manageable losses tum into large catastroplu‘c ones. Although there isno single magic. bullet toresolve all ofthese iss'ues. Ihope to offer some robust methodolo giesto aid inmm'imiz‘in'g the regre.t surrounding' these iss'ues.

Trade: ICE Brent Crude Oil Traditional trend-followm‘g rules renun'd us that the trend is' your friend and sm'iultan'eously warn that we should cut losses.. and let winners “1”

A

r" Minimizing Trader Regret

II7

How can we reconcll'e these two seemingly antithetical ideas.o Lookrn'g at Figure 6.1, we see that ICE July 2010 Brent Crude 011' 15' in' a hrst'on'cally low-volatility environment as defined byits 10day average true range. As discussed in Chapter 4.this favors implementing a trend-following breakoutmodel. so we place buy and sell stops at classical techm'cal honzo'ntal

Janka}; “L _y.... ‘

support an'd resistance levels in' hopes ofparticipatmg' in'a breakout from this low-volatility emirr'onment. As shown m' Figure 6.1. we placed a buy stop at$88.50 and asell stop at$84.98. On April 29‘. 2010. the market broke through itsresrs‘tance and we bought two contracts for $88.51. Our sellstop loss order could be placed at \anr'ous levels—depending' on our tradingpersonality andtrm'eframe as described m'Chapter -3—'1n'cludm'g the previous day's low, the three-day low. or even the support level of$84.98. The April 30, 2010. close at $88.46 was so close to our entry level that no protective (risk-reducrn'g) action could be. taken. Bycontrast. on May 2010. when the market settled at $80.87. this significant unrealized‘ profit of$1,360 per contract enabled us to take profits on 00" percent ofour position atthe next open as well as rarse' our sell stop tothe breakeven level of $88.51. Although there are numerous ways ofdefim'ng asignificant unreal— tz'edprofit level. here we use .30"percent ofthe market‘s 10-day average true range. Since the previous day’s average true range was $1.64. our unreaL iz‘edmark-to-market profit of3186' exceeded our cn'teria forem'tm'g 30-"percent ofthe position and raisrn'g the stop on the remam'der to breakeven. Atthis porn't we have realized a partial profit of$1.360 we are allowm‘gprofits on the second contract to run (thereby trading' the market and not the money, as described throughout Chapter 5)while sun'ultaneously preventing a significant unrealiz'ed profit—as defined bythe average true range—to turn mm a sigmfi'cant reahz'ed loss. Two things' could happen: Ifthe market experiences a powerful breakout. then it will' not be weak enough totrigger our breakeven stop on the other contract. allowtng' 15 to letthemnn‘er run. Ifitisamoderate breakout. itWill' trigger our stop loss. butat least we mimnuz"'ed feehn'gs ofregret bybooking anotable profit on halfofthe position. Tlu's second possibrlr'ty 15' what actually occurred. A moderate breakout ensued. \."evertheless. bybeing proactive and oppor~ tums'tic. we booked a decent profit on halfour position inst'ead ofallowrn'g asignificant unrealrz’ed profit to turn into asignificant realrz'ed loss. But after our breakeven stop triggered on May 4. 2010. we still’ remaln'ed in"alow-volatility environment (asdefined bythe market's average true range). Consequently. on May 5.2010. when our sell stop triggered. Wesold two contracts at $84.97. The settlement pn‘ce that daywas m‘flt". 50we exited one contract on the next open at for aprofit of$1.20 and lowered the buy stop for \.lay 6,2010. on our other contract to the lJreakeven pn’ce of$84.97. Because this" was asignificant breakout. our buy St0D was not triggered that day. allowing us to trail the stop based on our 95V.'cholog1‘cal temperament as defined 111‘Chapter 3.

TRADING TOOLS AND TECHNIQUES

Returning to Figure 6.1, notice. that Ihave shown multiple potential trcu'ling stop styles based on various trader personalities. For example, if' one feels the greatest regret watching sigmfi'cant unrealized profits evap orate before being stopped out, you could trail with a buystop set tothe previous day’s high. Ifyou chose this tight stop strategy, you exited the second contract on May 10, 2010, at $81.70 for a profit. of $3.270. Ifyour personality is more comfortable with letting unrealized profits evaporate inexchan'ge foroccasionally capturing bigger profits, you would keepyour buystop at the breakeven level until the lu'ghest high ofthe previous three days was below your short entry price. Ifyou chose to use this three—day trailing stop, it would have been stopped out on May 12, 2010, at $82.83. foraprofit $2,140. Finally, ifyour temperament iseven more acclimated to sacrificing unrealized gain's so as toenjoy an occasional home run myyou could use a five-day trailing stop. In this particular example, the five-day trat’ling stop proved well suited to the market action and would have exited this second contract. on May 27, 2010, at$74.23, fora profit of$10,740. The key point here isnot whether trailing the. market with athree-day stop is in'ferior or superior to various alternatives; it is instead that irrespective ofhow we chose to trail' the market, this regret nun‘m'u'zation strategy

i -r-‘w-."'~.:Mr')-Ar‘File—"vhfV'-.L‘~‘;.v Lu".'..'\" M .:. {39'

»

%Cloaod Q ATR alfnl'll‘canl pro! Il, :0 I M

Stopped on rml'nder .brukeven St

ad on rmn’nmr .83 a‘t

*i W84."W. .m.

Stopm on rmi'ndor $74.23 I'I -day h:9h tral I

I

Slomd on rmlnar g$81.70_it previous day I'gh trail

T

‘ I

+“In?

Wvolatt‘li‘ty

W_—_g1—_l—T—~h——_FL—Ji0

I’ll-IRE 6.1 July 2010 Daily ICE Brent Crude Oil Futures Contract Source: CQC. Inc. ©2010. All rights reserved worldwide.

#7 7

Trader Regret TAKE Kl

I09 Sold 2July Brent Crude Oil @ $84.97

Trailing Stop

Profit on 50%

Profit on Remainder

1-day

$1,220.00 $1,220.00 $1,220.00

$3,270.00 $2,140.00 $10,740.00

3-day 5-day

Note: Performance results shown excluding commissions and slippage. Source: CQC. lnc.

2010. All rights reserved worldwide.

prmemed a Sigrufi"cant unreahz‘ed profit from turning into asigmfi'cant realized' loss while sun'ultaneously allowm'g profits on at least part ofthe posru."on to run (see Figure 6.1).

Trade: Cash LS. Dollar—Canadian Dollar Emnext trade is'a false breakout signal incash U.S. dollar—Canadian dollar“.l'SD CAD). All of the criteria for entry are the same, namely our 10m’_\'average tme range signaled low volatility; we placed a buy stop at the honzo‘nral resrs‘tance level above the September 13, 2010, high of 1.0372 and below the Se”ptember 17. 2010, low at 1.0215. On September 22, 2010, our sellstop order at 1.0214 triggered ashort position of200,000 base cur— L'nfortunately. the breakout turned out to be false and we were enhe‘r stopped out when the market violated the previous day’s high at 1.0331 for a loss of$2400 or when it n’olated its three-day high at 1.0333 fora loss of$2.760 or when itbroke its longer-term resistan'ce on SeptemberZ3. 20'10. at 1.0373 for a losS‘ of $3,180. Although in this particular instance the tigh'test ris'k cn’ten’on represented the best decision, the main POIII'I is'that our disc'iplin‘ed placement of the stop loss order prevented amanageable loss from needlessly turning into a catastrophic one (see figure 6.2). Our July 2010 ICE Brent Crude Oil led to a continuation ofthe low_“013u11"ty ennro’nrnent and enabled us to participate in the real breakout mtheOpposite direc'tion. Bycontrast, the September 2010 cash BSD-CAD beans'h breakdown led to a continuation of the low volatility reading In average true range, but offered a potentially tradable downside breakr’m‘ I53." 'potentially" tradable because tlu's all depended on how tightly we managed ris'k. Bysettm'g our stop to the previous day‘s high, we enwe‘j‘m the smallest loss on the false dovmside breakout. However, when 9501d200.000 ['SD—CAD on October 1,2010, at 1.0190, ifwe set our buy 51""31 theprevious day’s tu'gh, we were stopped out fora second time on

g

“0

TRADING TOOLS AND TECHNlQUfi

Supper! . 1 0215

fill slop 1 0214 rl'ggorod

VA"! “Y'allo- vo utility

' “4 “NET” "“‘Iif“ ' fi’wr—a—T-‘u

I’ll-1|“) “.2 Daily Cash USO-CAD Chart Source: CQG. inc. (C; 2010. All rights reserved worldwide.

October 5,2010, at 1.0246, foraloss of$1,120. Bycontrast, those willingto take more risk bysetting their stops at either the three- or five-day highs were not stopped out for a loss on October 5,2010, and so on October 6. 2010, when the market settled at 1.0111, theyenjoyed astatistically significant unrealized gainas defined byexceeding 50percent ofthe 10-day average true range. Theyconsequently exited half their‘ position on the open of

TABLE 8.2 Sold 200.000 Cash USD—CAD @ 1.0214 Trailing Stop

Loss

l-day 3-day 5-day

($2,400.00) ($2,760.00) ($3,180.00)

Note: Performance results excluding slippage. Source: CQC, Inc. ©2010. All rights reserved worldwide.

r" Minimizing Trader Regret

trail W1—dny alwm . 1 0246 Sloppw on rmInGOI 0 1,01%

-/

' $10114 IOU 3- I 5-day [IDA 01:! m m noxl v.9976 mu 3 mvo slop on rm'undor to m brook-van am, goh't12

WAIR Ui?".ls ATR IITM'IV In vo ltl’li'ly lw vo utility

FIGURE 6.3 Daily Cash USD—CAD Chart Source: CQC, Inc. © 2010. All rights reserved worldwide.

October 7,2010, at 1.0112, for a profit of$780 and were able. to lower their buystop on the remainder to the breakeven entry price level of 1.0190, where theywere stopped out later that day. As an aside, because we were unwilling to allow a significan't unrealized profit to tum into a significant realized loss, we were, stopped out at breakeven 0n the remainder of our position an'd so (lid not, participate in the move to cycle lows on ()("lOlK‘r 14,2010, at the 0.9976 ar'ea (see Figure 6.3).

TABLE “.3 Sold 200.000 Cash USD-CAD @ LG] 90 Trailing Stop

Prom or Loss on 50%

Profit on Remainder

l-day 3-day 5-day

($1,120.00) $780.00 $780.00

Not Applicable $0.00 $0.00

Note: Performance results excluding slippage. Source: CQG. Inc. ©2010. All rights reserved worldwide.

,‘

“2

TRADING TOOLS AND TECHNIQQS

Trade: Google Our next regret minun'iza‘tion example shows how bytradlng' three um ofthe asset. we enjoy even greater freedom in' exploitmg' trend-follow“ opportunities. On September 16, 2010, Google generated an extreme to. volatility, as measured byits 10—day average true range. We c placed a sell stop below its honz'ontal support of$475.08 and a buystop above resistance at $484.75 (see Figure 6.4). On September 1?, 20'10. the stock broke to the upside, and we bought 300 shares at$484.76. That day's strongly higher close represented unreallz'ed profits greater than 00'perm-m ofaverage. true range, so we took profits on 100 shares at the openmg‘ on September 20, 2010, at $492.18 (aprofit of$742) and moved our stopto breakeven on our remaining 200 shares. At this pom‘t, we continue trailin'g our sell stop at the lowest low of the previous five trading days. Also, by lookm'g at the weekly chart {see Figure 6.5), we see sigrufi'cant rests'tance around the $600area and placea limit order to sell 100shares at that pn'ce. Our lum"t order to sell 100shares at $600pershare 15'filled on October 15.2010. for again' of311.524. Weare stopped out ofour remauu'ng 100 shares ofGoogle on November ll.2010.

9] n” “Mom 80M 100 okra: 300000 luminooval

i.

" H'l sJ

In It above “‘4 750’VOIIICMO.

I Stupped out 0!

I won. [.317 51 n.

!I/fl‘Jl'flmfi/ "

t

M \,

1" I , ). l

Sold 100 about on 090a 0 $492.18

.‘t‘.

“my6

ML“ volat I llly

'1

.m‘ jawwwvfi‘LW—r—W—WF—r—r—fiL—r—r‘u'um' Hm I") 6.4 Daily Chart ofGoogle Inc. Source: CQC, Inc. ,-C’- NI0. All rights reserved worldwide.

7*

Minimizing Trader Regret

II'! I

3600,00 ran-tone. uu

FIGURE 6.5 Weekly Chart ofCoogle Inc. Source: CQG. Inc. ©2010. All rights reserved worldwide.

when the stock takes out its five-day low at $617.51, generating a profit of$13,275.

ISSUES FOR MEAN REVERSION TRADERS There are many reasons for mean reversion traders to feel regret. bo‘me are easfly resolved bytechm'ques introduced in this chapter, whereas others are quite challenging. Bydefinition, mean reversion traders tend toexit when theasset reverts to themean, so there isa natural- tendency forthese methodologies to experience regret whenever positions they exit extend m'to significant trending opportunities. For example. Figure 6.6 shows a lipical mean reversion set-up inwhich the nine—day relative strength index ofcash euro—US. dollar' (EUR—USD) crosses below overbought levels on November 2010. Our mean reversion trader then sells the November 8, 2010, open at 1.4060, placm’g a protective buy stop at the previous day's high of 1.4248/ and a limit order to exit with profits at the previous day’s

TRADlNG TOOLS AND TECHNIQUES

RSI moon. bolo-

ovo thought lovol W/ \nxosut 014000 a“ opon

'y

20-day I loving aver-go

‘35“

Cycll lo,t 3441 six fading days later

RSf’c'rossés"bo'lo"' 7' m " WM" ovorbought lovol

FIGURE 6.6 Daily Cash EUR—USD chart Source: CQC, Inc. cg 2010. All rights reserved worldwide.

20-day simple moving average at 1.3955 (see Figure 6.6). Despite our mean reversion trader realizing this high—probability profit, of$1,050 per 100,000 base (-urrent'y that same trading day, over the following six trading days the market eontinued trending down to its eyele low of 1.3447. which represented the potential for signifieai'itly greater additional profits. Although itis possible to modify traditional mean reversion models so they i'ninimize regret over trending market action. in doing so we typically shift away from the shorter trat'le duration an'd higher winning percentage type of mar'ket behavior that these‘ traders were trying to capitalize on in the first place. (,‘onsequently. my answer for traders expen‘eneing these types of regret was introduced in (.‘hap'ter 3,where we discussed eating your own lunch. Another disc-lau'ner regarding mean reversion regret mimm'iza'tion techniques isthat they should not be confused with adding to a losing po sition, averaging down, or dollar—cost averaging, which are antithetical to sound n'sk management techniques. When adding to a losing position, we are trying to extricate. ourselves from a loss bylowen’ng our average entry level. Forexample, letusassume inJuly2007 we bought 1,000shares ofCitigroup Inc. at $50per share. ByOctober 2007, the stock had dropped to $40.

r—yfilt’ admittedly [Wm of a mmageabthougil realizmg' adof ms'te at $40 per share, may lower we decide to buy 2.000 shares to $43.33. In .N'member A‘flfl, shares 3,000 on all pn’ce even 510‘000' an unreallzed' loss.“ofnearly 341,111,, represe‘nun'g $30, to pdropped _ Wu 4,011} such a large lose, we to dC'Cept 'lling unvn were ause we

m 71,00 our breakeven pace on lowenn’g thereby share, at $30 per $20 per“.8. an m ares, Ber to r dropped stock the sh InMarch 2008. to buy

1055., we decide ares to $35.71. the acceptmg' of shlzed. a", 1055 of $109,970. Instead pne'e to 52733 on share per lowering our average shares at $20, re000al 2008. Citigroup was obkusb'v m'm'em 12, November By 8’ shares. ' govemment tosunn‘e (that will»! bythe bailout a require anbdw'oooould 2008). Its drop to 810 per .sl'ur'e 2‘3, November on was officially approved Assummg’ we comm’ued dus'

meant an unreahzed' ' cmastrophic ave

loss of211111051 $260,000. W 'e would have bought 16.000 By rage share price to $18.39 on all 31.000 shares. loss of dropped to $1 per share, for an unrealued’

also a“. commonly known as pymm ul‘z‘ng method The position-SIZ'm'g as or addm'g to a losmg' position.’ as well down, averagm’g both fers from

u.~..nn...... VRhL-xm-ysxs'lnwl-nol... .«lwxvlprhn‘lv.

."

n L“;

i '9‘

‘.000 out.-

""ét"/?!m,,lcause on

its.” on". pnco '- ‘03 33

l'_/'.

I

00 mm...

£133.00!) that” W m1... O use. 5 ‘35 II

/ :32, 1/ ‘

W

whose u»... 3m. - t2:n"""'

“I”

"a

“.m...

«

" 60"1'c-vcny"'5 furl / rum“: " I. j ’0 em 0 6F; L",r’_:0,.z",§.l i H

ml 16 000 um” 2.2 so...-,.-- m...

.

’5‘“

"'”~.'.:~‘,.=* ‘ I

"H" ‘ 6‘ Source-"C: -7 Weekly Chart ofCitigroup Inc. - QC. Inc. @g 2010. All rights reserved worldwide.

g

p..-

“fl

ll“

TRADING TOOLS AND TECHNlQUES

'I'\lllil'.' 6.4 Averaging Down with Citigroup Inc. Share Price

Shares Purchased

Avg Share Price

Unrealized Loss

$50.00 $40.00 $30.00 $20.00 “0.00 $l.00

I.000 2,000 4,000 8.000 16,000 Not Applicable

$50.00 $43.33 $35.7l $27.33 $18.39 $18.39

$0.00 (5l0,000.00) ($39,990.00) ($109,970.00) ($259,950.00) ($539,090.00)

Note: Performance results excluding commissions and slippage. Source: CQC. lnc. (C) 2010. All rights reserved worldwide.

mean reversion regret minimization techniques. First, unlike both addingto losing positions and regret minimization for mean reversion traders, pyra~ nu'ding should only be used in conjunction with trend-following tradm'g models. Successful pyrznuiding is inmar'iy ways the polar opposite toaveraging down in that pyraniiding begins the trade with the largest number of contracts, and adds fewer contracts as thetrade generates larger amounts ofunrwilized gains. Bycontrast, averaging down begins each trade with its smallest position size and adds larger numbers ofcontracts as unrealized" losses rise in an exponential fashion (see Table 6.4). Consequently, many successful long trend—following traders adhere to pyramidin'g and argue m' favor ofitas a prudent position-sizing strategy. Itisobviously true that averaging down isnot aprudent position-mg technique whereas pyramiding could be. The problem isthat most professional traders. agree that mar'kets onlytrend around 30percent oftheum’e. which isfine for trend-following traders because durm'g those lessfrequent periods when they we in this trendm’g mode, they offer larger profits than losses endured during the 70 percent of the time when revertm'g to the mean. For pyramiding to be successful, however, markets not onlyhaveto be trending, but trending in a near-parabolic manner. When they ms‘tead stair—step higher or lower, pyramiding is psychologically debilitating because it,forces us to allow significant unrealiz'ed profits to turn m'to sigmfi' cant. realized losses. The summer 2010 rally in December 2010 CME Group wheat futures was a near-paralmlic bull trend that was particularly well suited to pyraniiding. In determining appropriate position-sizing limits and levels forOur pyraruidingstrategy, we must ensure that our risk never exceeds 1to2‘percent oftotal assets under management. Assuming a tradm’g account m'th $0)".(),0()0 and a risk appetite of 2percent of total assets under man-age“ ment, we could place a buy stop at the resistance level of$5.20 per bush'r?l on seven contracts. This allows us to place our stop loss order to sell3‘

r Minimizing Trader Regret

+1?!

me cycle low of $4.93 per bushel. n‘presentlng a $0217 cent, or $1,850, "5kper contract times seven contracts for a total position risk of$13,050. which isjust under 2percent oftotal assets under nnnuigement for our $511000 account. On July (1, 2010. when the market rallies to $13.10 per bushel. we can buy four more contracts and raise our sell stop on all 1lcontracts to our initial entry pn’ce of$.»"...v"0 per bushel Notice that although we are adding toour existing position. our signitlczmt uin‘eali/xed gains enable us toadd to the position without Violating our .3‘percent risk rule. In fact, the addition Oltht‘M“ four contracts along with the nursing ofour stop loss to $5.30 rep resents a$4.000 n'sk on the 11 contracts. This isbecause our awmge price perbushel is now around $5..."72.i" on all Ilcontracts. which iswell under 1percent oftotal assets under inanagenn-nt. As long as Wt‘ are stopped out. there is little or no liquidity risk (for e.\".unple, a locked limit down more that might prevent our sell stop from being elected at or near the $3.20 level). OnJuly 8.2010, when the nun‘ket rallies to$5130 per bushel, we buy “No more contracts ai'id raise our stop on all 1:1 contracts to $13.40 per bushel. Now. ifour $5.40 sell stop Were triggered. we would be stopped out at a total profit of $0,000 on all 13 contracts beczmse our :werage price per bushel was $-.)"..‘3..".‘l.‘l cents per bushel. Finally on July ltl, 3010. the market hasrallied to$5.80 per bushel. when we buy our tlnal contract while raising ourstop on all 14 contracts to $5150 per bushel. ll'onr sell stop at $5.00 per bushel were tn’ggnered. itwould tnmslate into a protit of$17,000, since our average price per bushel on all 14 contr.'icts is now $751.67."). The wlient market rises instead in a near-paraln)lic fashion, :u'hieving our long-term pn'ce tar'get of $7.50 per bushel on August ‘1. 3010. when We sell out our position, realizing a protit of$150,000 (see Figure 0.8). But remember that the 2010 wheat market was a llt‘.‘|l'-|h‘ll‘ill)(Dll(' trendingmar'ket. Now let us apply this pyrauuiding strategv in an ordinary bull market like December 2010 IRS. 10-year 'l‘reasnw note futures. During .St‘ptember 2010, the ltiyear note l‘ntures displayed strong overhead resistan'ce at 1.."ti'0...".i'.'.3 and recent cycle lows Were established at 13".»‘05. ’l‘his represesnted an initial n'sk of$.0.'l7..")0 per contract. 'l‘raders with $1 million inassets under management who were willing to Use a 2percent risk ceilingcould place an order to buy20contracts at l.."b"0.‘l on a stop and would ben'sking $18,750 ifat'ter entry they were stopped out at 1.."5‘05. The good news was that the break ofthe resistance level of 120112.25 Was a real breakout in the context of a long—term bull market. (‘onseQuently. when on October (5,2010. the market rallied to 1.."7'00, our trader bOlltiht another 10contracts and raised stops to his initial entry price level 0112603 on all 30 contracts. l-‘iifoniitiziti-l\_'. although the 'l‘reasnries were trending higher. they were not trending higher in a near-parala)lic fc‘lfiillt)", “d.

TRADING TOOLS AND TECHNIQUE

‘80“! .14 0$7.50 largo!

Bought 1 . sseo', raisod stop on.14 lo 85‘“

wt 2 85m hind Hop.on 1'3 to $540

sou, M 4o 85. O, raised stop on 11 to 85.20

FlGl'RE 6.8 December 2010 Daily CME Group Wheat Futures Source: CQC. Inc. g2010. All rights reserved worldwide.

sO‘ on October 15.2010, he was stopped out ofall30contracts at 12603, for a loss of$4,031.25 (see Figure 6.9). Although tlus' loss was well withm‘ acceptable ns‘k tolerance levels, it may not have been psychologically palatable forour trader to watch asigmfi'cant unreahz‘ed profit of $9,062.50 turn mm a significant reallz'ed loss of $4\.-)"31.._‘75. Unless your tradn‘ig personality 15' well suited for endunng'

Entry Price

55.20 55.40 55.60 55.80

Pyramiding December 2010 Wheat Futures Contracts

Avg Price

Market Price

Unrealized P/l.

-'NJ-‘~\J

TABLE 6.5

55.20 55.2725 55.3233 55.3575 55.3575

55.20 55.40 55.60 55.80 57.50

$0.00 $7,000.00 Sl8,000.00 531,000.00 5]50,000.00

Note: Performance results shown exclude commissions and slippage. Source: CQG, Inc. ©2010. All rights reserved worldwide.

Minimizing Trader Regret

I19

0., ,s M

30th 10 g

Roan'sl-noo Q 126'02V5

,.

.4.—

Mu hl N 126902 5 .

ii i “ \

‘\

. il ? WW

5”;8'-o"2 5

cycle I". of 125'05

mcocn 'mxuwiumm “Whitman

“1MB.” unm-

FlGl-lllE 6.9 December 20] 0Daily CME Group lO-Year Treasury Note Futures Source: CQC. Inc. ©2010. All rights reserved worldwide.

this km‘d of regret, pyramiding is probably not the best strategy. Ifyou feelwatching such unrealized gains turn into significant losses debilitating, then the portion ofthis chapter on regret minimization fortrend followers isoffered as an attractive alternative. Now that we have clearly defined position sizing as itrelates to both averaging down as well as pyramiding, we can distinguish both of these from mean reversion and countertrend regret mm'im'ization techniques. Some ofthe major problems with most mean reversion and countertrend

TABLE 6.6 Pyramiding with December 2010 Ten-Year T-Note Futures Entry Price

Contracts

Avg Price

Market Price

Profit/Loss

126'03 l27’00

20 IO

126’03 l26’l 3 126'13

126'03 127'00 126'03

$0.00 $9,062.50 ($4,531.25)

Note: Performance results shown excluding commissions and slippage. Source: CQG. inc. ©2010. All rights reserved worldwide.

l20

TRADING TOOLS AND TECHNIQUES

models include nu'ssm'g our limit entry price target or endun’ng large draw. downs tobetterensure trade participation. The techniques introduced here minimize trader regret through the staggering oflimit. entry orders atvan“. ous support or resistance levels. At first glance, staggering ofentry orders looks somewhat similar to averaging dovm, which is'why Iclearly defined the latter earlier. The man distinction between averaging down and staggering isthat averaging down involves an increasing of position she as unrealized losses mount so as. to improve our average position pn‘ce. Also, averaging down is,general'ly speaking, a reactive—as opposed to premeditated—volumetn'c positionsizing tool introduced inhopes ofavoiding the pain ofexiting with losses. Bycontrast. the staggen'ng oflimit entry orders isused by mean reversion traders who plantoachieve specific. predefined position sizes, but are willingtoaccept smaller positions to help them minimize the regret ofmissing a positive expectancy trading opportunity.

Trade: IB“ In October 2007, IBM shares began pulling back from cycle highs at $121.46 per share. Measuring from July 2006 cycle lows of$72.73, memt reversion traders noticed a wide array Fibonacci retracement levels to place buy orders at 38.2 percent, or $102.85 per share, 50 percent, or $97.09 and 61.8 percent, or $91.34. One possible solution according to regret minimi7'.at.ion theory isnot choosing at all. Ilike to say. “Don‘t miticipate, just participate.” Instead ofguessing which ofthese support levels would he tested, we place buy orders at each level along with a sell stop below the. final 61.8 percent Fibonacci support ar'ea. Assuming a trading account has $52 million in assets under management and we are comfonable n'sking 2percent ofassets under ntzuuigenient., or $401100, on a per trade basis, we could safely place limit orders to buy 500 shares at each of the aforementioned Fibonacci retraceiuent levels along with a sell stop order on all 1,500 shares at the cycle low level of $88.76 per share. Looking at Figure (3.10. we Sis-e that IBM pulled back 50 percent byJanuary 2008 before the retesting ofits previous cycle high at $1...’-‘l.4b' per share in April 2008 (see Figure 6.10). Table (5.7 shows how the stagg'en'ng ofbuy orders enabled our pan'j('ipation in this trar'le setup instead' ofmissing the opportunity byplacing a single limit order at$91.34 per shar'e. Also notice that staggering buy orderS at $102.85 ai-id $97.09 offered supen‘or performan'ce—an'd less regret—than placing a single buy order at the $102.85 pn'ce level. Admittedly, buying 41" shares at $97.09 would have proved even more successful, but agm'n, our goal isminimizm’g regret ofnu‘ssed opportunities (18' opposed to captun'ng the elusive “perfect trade” (see Table 6.7).

"HI‘I WI “|

i i F

"V‘yu‘. ,

I i

l2!

Minimizing Trader Regret

~

“hop on t to: ‘8' 76 cycio ’0'

mu; 8J0 Weekly Chart ofIBM inc. Source: CQC, Inc. c,"2010. All rights reserved worldwide.

Trade: \aiural Gas Futures This trade illustrates aJ'i osswitial aspect ()fsuccessful speculative tradin'g. ndn'wly. rim iwing n'gid with entry—(L's Oppom‘d to risk mmiagvmoni— order plar'i-irwni levels. On Apn'l 20, 2006. the natural gas market made itsr'yr-Ir- high at $12.48!) basis January futures: By.Si‘ipivmhvr 29, 2007, it put initsr'yr'lt‘ low at $7.444. A! this point. traders started examining vimious Fibonacci rotrar’miwni low-ls for establishing short positions into the

rm. 8.7 Regret Minimization with IBM Inc, Share Price

Purchased

Avg Share Price

$102.85 $97.09 $91.34 Actual Totals:

500 500 None 1,000

$102.85 $99.97 $97.09 $99.97

Shares

Market Price

Mark-loMarket

Profits at $121.46

$102.85 $97.09 N/A $121.46

$0.00 ($2,880.00) N/A $21,490.00

$9,305.00 $12,185.00 N/A $21,490.00

NOIe' Performance results shown exclude commissions. SOWCe: CQC. Inc. 9*2010. All rights reserved worldwide.

TRADING TOOLS AND TECHNIQUES

l22

. _.p a“... .I “It 5

SW.

8 trading 60;: Illor O 1 444

. .--~

l 117‘

i-

is

2

p

‘2

n-

I ..

_

le~mwtn ‘ .x» -« .n “maxmher'L-v-x-m-HV-Lm -n“ .»"mummy-m o\-J> an (.m— min

,5

countertrend pullback. Accordingly, assuming a trading account with $3 million in assets under management and a risk tolerance of2percent oftotal assets under management, we could place. sellorders on three contracts atthe 38.2 percent retracement level of$9.371 per MMBTU, another three. contracts at the 50 percent retracement level of$9966, and a final three contracts atthe 61.8 percent retracement level of$10562. We would also place aprotective buystop order above the 61.8 percent retracement level at$10.73 on allnine contracts. The problem was that bymid-November, the market had tested the $8.75 to $8.90 resistance area for an entire month without making any significant upward progress toward our initial sellprice of$9.371. Those who rigidly adhered to Fibonacci sell levels irrespective ofthe market holding resistance around $8.90 missed the opportunity tosell January natural gas futures. However, those willing to modify their existing orders in light of the market’s resistance could have reduced the number of contracts offered at each ofthe Fibonacci price levels to one lot while adding a sell order atthe recent resistance level of$8.88. This flexibility enabled partial participation in' the bear move byselling one contract at $8.88 on November29,2006. This contract would have been covered atcyclelows of$7.444 on December 11,2006. for aprofit of$14,360.00 (see Figure 6.11).

174

“A v1”

'|

.101 ‘1 105

-. ~

10

n t

13"

FIGl‘Rl‘.‘ 6.“ January 2007 Daily CME Group Natural Gas Futures Source: CQG, Inc. ©2010. All rights reserved worldwide.

HI.

.

l2?)

Minimizing Trader Regret TABLE 6.8 Regret Minimization with January 2007 Natural Gas Futures

‘_______—__—————__—_____————.____

Entry Price

Contracts Sold

Sale Price

Market Price

Mark-toMarket

Profits at 37.444

M—

510562 $9.966 $9.371 $8.880 Actual Totals:

0 O 0 l l

N/A N/A N/A $8.88 $8.88

N/A N/A N/A $8.880 $7.444

N/A N/A N/A

$0.00 $14,360.00

N/A N/A N/A $14,360.00 $14,360.00

Note: Performance results excluding commissions. Source: CQG, Inc. ©2010. All rights reserved worldwide.

FINAL THOUGHTS Regret minimization techniques like taking partial profits and then movm'g stops tobreakeven on the remam'der are invaluable in'reprogrammm'g the trader away from irrational cycles ofeuphoria and fear in favor ofevenmindedness. (The final chapter of this book examines even—min'dedness intrading, which is our ability to embody an objective, emotionally tempered attitude toward trading opportunities.) Inaddition, regret tiontechniques are an antidote totheperfect trader syndrome. The perfect trader syndrome occurs when traders seek perfect entry and exit prices by buying the low tick and selling the high. Irrational attachment to buying the low and selling the lu'gh becomes an obstacle impedm‘g our ability to successfully enter and exit trades. Because regret mini'mization techru'ques train' us to exit at avariety ofprofitable price levels, they aid in‘ deactivat1n'gtheperfect trader syndrome. In summary, through regret nurunu"'zation, many ofthe Wall Street adages that formerly seemed like apipedream become emotional realities, including being right and sitting tight as well as the classic: cutting losses and letting wmn‘ers run.

I}II \l"l'l‘.'R 7

'l‘imeframe

Analysis

Acloud ismade ofbillows. upon billows upon bellows that look like" clouds. As you come closer to a (loud you don’t go! .s‘onwllu'ng smooth, but irregulam'tt'es at a smaller scale. ———Dr. Benoit, Man'delhrot

no ofthe most robust tools forgent-\rating positive expooumcy models is timefrzutw analysis. This chapter explores both traditional tinwt'nune zmal'ysis and timofrzune divergent-o. ’artit'ular emphasis isplilt"(‘d on usingvzm'ons.‘ mathematit-al l(‘(7hlli(‘¢.‘ll indit-ators to undmstan‘d likoly market bohaw'or invan"ous timel‘ramos.

TRADITIONAL TIMEFRAMI‘.‘ ANALYSIS lhogan .s‘tutjlying markvt lwhavior in “)8? in the hope ofdeveloping positive o.\'poct;mt'y trading modvls. and gravitated to the simplicity~—when compared to fundzunvntal' 1~utal'\.v's1'.s‘—-—of lt.’(‘l|lll(.‘2l] indit'ators. Ihogan to understdn'd the importance ofmarket trends and that var'ious tt't'hnical in— dicators could help in trend identification. When stntggling to determine thetrend, Iquickly realized that not all indicators werv ('roatod equal. and that technical indicators derived from mathematkral formulas offort-‘d an Objective mtswer to the question “What isthe trend?" As Icontinued study~ mgonce history, applying ('lifferent mathmnatir'al technical indicators to determine the trend, Ieventually lean'ted that there was" no single answer

l25

Mb"

TRADQNG TOOLS AND TECHNtng

to the quemion. The only satisfactory answer to the questi.'on is‘W questsion: “What isyour timetrame‘I" in other words; according to ohiect'ive mathematical techm'cal m'dica. tors like moving averages or Wilder‘s relative strength index ('Rb‘ll." an asset could be in a hull trend (nude-period RSI above M) on its monthly and Weekly charts, abear mar'ket (nm'e-pcn'od RSI below ot'l)according" in its dmly chart. and a bull market (mn'e-pen‘od RSI above 507) based on n; hourly chart (see Figure 7.1). Traditional tm'ieframe analysts consequently always begins with determining the aswt's trend. support and res:s'tan(.'e according to its longest .timeframe. Once we have a clear picture of this macro perspectiw 0! market behan'or. we can .saf‘ely zoom in and examu'ie the trend. support. and remkstance based on .sh‘orter-tenn ttn'ieframes. This is illustrated by Figure. 7...". which shows achart of30—minute bars't. tn‘ICE mm 20‘“ sugar futures. Based on this chart. we would conclude that' sugar was In'abear mar'ket an-d look to saw" rallies near' the (werhead resistance level ofmm} per pound (see figure 7.2). Now compare Figure 7.3 to Figure 7.2. how that by lengthenmg’ our titncframe from 30 to bt‘Lnun'ute bats. we define overheat twist/since not at anew. but at $0...."Q3O per pound ms'tead (see. figure 7.3). \.'ext. t‘ontmst both the 30-, and 60-minute charts tothe dailychart and our definition ofthemset‘s trend changes from beansh“ to bullish while‘out View of technical support levels shifts from $0-2703.’lb. on the mmm'ute timet‘rame (or$tl..."o‘i':)"’lb. on the60-minute tin'ieframe) to$0..')o."3(.ly’lb. on the daily timeframe (sew Figure 7A). Fumll'y. byexamining Figure 75. we gam‘ a macro tuiderstandm"g of sugar's trend——wlu‘ch is decidedly bullish-as~ well as aclear picture ofmultiple long- and intermediate—term support and resistai'ice levels (see Pigm'e 7.5). it is‘ also important to realize that by switchm‘g from the daily to weekly nm'eframe m’exchange~traded futures we shift from a specttfi'c contract month——in our example. the March Null [(‘F. sugar futures. contract———to a rollm‘g front-month combination chart. and therefore older data pom‘ts on the weekly chart‘ (such as the tent term support level of 30.13% established dunn'g May 3111th reflect 509 port and res-L."tance levels offront-month sugarcontracts at that time inthe asset's history: The most robust and commonly employed. solution to timetrame midi" ysisissues is‘mam‘taim‘ng asingle screen that instan‘tan'eouxlsy shtmws multi~ pletimefrutnes forthetraded ax‘se‘t (see‘ Figure 7.6). Thetu’nefmmes should provide a clear picture of the trend. support. and msistancc accord“ to the timefmme that you me trading. as well as longer tm'iefmmes in other words. Figure 7.6 would be pnfifemble to those‘ malw'u’tg trade decr sions based on :flminute bars Bycontrast. shorter—term traders nughl'W" W'ute bars for their pn‘man; market an'al\_ssi.“. while looking at‘ 30» 31"

.4; u‘htfu’ln"‘. . .- Mum. ~14~~Vux a-nw.»mm.w.'~wwuw&

y.~‘

FIGURE 7.I Mommy. Weekly‘ Daify. and Hourly Charts of Realty Income Corporation on December 3. 2010 Source' CQG, Inc. @-2010. All rights reserved woridwide.

g, ,2:

{I

a}

R:

MW w.._.....«-...,.._._w.._...w‘fi7. -m g...

'

~

m

l27

~

«In

M'-

«a

vu-

-: v-

u w...

«m

km

~-

«A

.w

a”

4'"

u”.

.n

FIGURE 7.2 ThirtyMinute Chart ofICE March 201 1Sugar Futures on December 3.2010 Source: CQC, Inc. mt market View (see Figure 7.7).

flflEFRAH-E CONFIRMATION TR..\~I)ING WWW

Traditional- timefrmne trading suggests buying when all tiinefrzunes anamed ar'e bullish or selling when all indicators are beari'sh aeeording to mathematieal' teehnieal' indicators such as moving averages or percentage willatoxs like Wilder's relative strength index (RSI). There are two eoue manly employed versions of generating timefrmne continuation trading sign‘aI's: the first, suboptiinal' version blindly buys or sells anytime all titne~ frames are' bullish or bemlsh rt-‘gardless ofhow long ago the tiniefraxnt-‘s confirmed. buyor sell signals (see Figure 7.8). Bycontrast, the second, more robust version oftimc-‘t'ranie confirmation allows trade execution only when longer timefnu-nes show trend eonfinnation an'd the shortest timefrmne shifts from divergence to continuation. 'I‘hsu’ is illustrated by exan'iining Figure 7.10, which shows the purchase“ pn’ce for GB of $16.65 per shar'e when RSI (Tossed 50 at 2:40 l’.M. with buying atthe, close for$16.75 per share (see Figures 7.9 and 7.10.).

Trade: International Paper (,‘ompany FlIgure. 7.11 shows dm'ly, hourly, mid .‘l(')~tuinute Chart-s emitinning a bullish trend in Intema‘tional ’aper. In addition, the 5—minute ehart ol‘ International I’di’x'r shifts from diverge-nee to confirmation of this bull trend at 13:00 noon EST on December .‘3,2010. tn'ggvering our buy signal at $26.03 pershar'e. Although there are minty ways to manage risk on this trade, one "fIn? most popular isbyplzu'ing n proteetive sell stop at the previous ey-« (‘l'elow of the longer, .‘ltl-minute titnefrmne chart at $.5’5.8ll per share (see figure7.11). There are aim at wide Variety of tools for exiting with profits, Including theregret iut'nimil.’ation let‘hniqm's covered in(‘hapter ti.Here we Offer adifferent nteel‘umt‘sm from those previtnisly shown, namely, exiting when the relative strength index of lntemational' Paper's ‘FHIIII’IIIN‘ ('hart CNS/581‘s." above 7.)". This oet‘urred at ...’v’:3-"tl l-I‘o'T and we sold the stock at $313136 Der share (see: Figure 7.12).

ammonites sumo LN”EFMMEW‘

flu...“ w.._....t~..—~....l

. . Al . . . “(Tu-lull“ Ibelieve that trading in the direction of the long—term trend dj"'05! always a good idea" one, of the, most. valuable tools Ihave.

k

.. a

n.» nu

"\

mcnuru Jvhns

if).

~= M; -

ism?»

in"? '1,

M:

*4»

1;;

” ,n

_/

Ana—"m... ..»4.. A m..- , >._--,W..«wfl , 7,. .V

\J‘n

a‘a.nu1_

3|“ val]

nu "u, -,

«13%;!

‘Mnnun 'Auvn

'unIMpuuM pqvuofiwu equhu “v

-

1 u); anw w suvuj axnuiw 5 puv '«(n

- w» 1w...ua~mvu~wmmadip§1mmww\mmw/‘Jh/«nhug, «“(HAI A.‘ . mm...“ w?”

\‘x‘ m..-“

x- 4/‘\

. - “ Ilm . a» Maw. . ww: .fi.— mpm. may“: 1 M

,A

‘

.W.ngmwg,-.mw.mw.m.~;,;gxwmg .. .4 «

,4'“

Close > ?0rit. we look at summn lm‘vls ofgold's longer tinwfnmws. Act‘ording to our llnn'nuu- har than. this .:s1n);x)n was the swing low price of $1,402.60. Sm’rrp tradvrs.‘ using twn-minntv bars tend to look for quick prulits dn-d high"! “inning rwn‘rz'ntng'os. lplm'ml nur limit order to exit with pmfits at the highs of tlw bar' lie-{um the rag-latiw strmgth index gave its oversold n-ad'ing. This limit order to mt at $l.~107...‘$(') de" triggered dr'ound four m‘nutm later (5w l‘t'gnnw.‘ 7,13 and 7.14 l.

l0filler-Trrm Trade: (ME (.‘mup (.‘rude 0H l’e‘mire 7.13: .. Imoku2(@,26l [-1]

Long Exit: Imokul(@,9) [—1] < Imoku2(8,26) [-1) Short Entry: Imoku1(@,9) [-1]

< Imoku2(@,26) [—1)

Short Exit: Imokul(@,9) [—1]

TABLE 8.5

> Imoku2(@,26) {-1}

Ichimoku Crossover System

Asset

Profit

#Trades

#Days

Max Draw

MCL

P:MD

96w

P:L

ES CL NC CC 25 ZN EUG 1Y6 53

(10.200) 129,160 163,380 (6.230) 23,655 37.159 60,475 33.095 1.592

105 89 97 105 117 95 115 108 123

25 29 27 25 22 27 23 24 21

(46.287) (31.3901 (67.220) (45,300) (29.8071 (12.226) (37.740) (23.1 57) (12.101)

15 3 5 6 8 6 10 10 13

N,"A 4.11 2.4 N/A .79 3.04 1.60 1.43 .13

33.33 46.07 43.30 42.86 38.46 47.37 41.74 42.59 35.77

1.82 228 2.00 1.25 1.94 1.82 1.96 1.75 1.86

Total: 432,086 16 4.66 40.99 1.98 954 25 (92,771) W Notes. Portfolio summary includes data from Januaw 1.2000, to December 31. 2009. and assumes $10 round~turn deductions for slippage and commissions.

Source: CQC. Inc.

2010. All rights reserved worldwide.

I54

TRADING TOOLS AND TECHNIQUES

Next, we will make the simple Tenkan/Ku'un trend-following crossover model more robust by incorporating the clouds. Specifically. long entry now requires bullish Tenkan/lx"iiun crossovers to occur above the clouds. and shortening bean"sh crossovers must occur below the Clouds. Also, open positions are exited whenever a reversal crossover occms or the close isno longer above or below the cloud (see Figure 8.7). Using (‘QG, the programming code for the Ichimoku Clouds crossover system iswritten this way: Long Entry: .Imokul(@,9)[«ll

> Imoku2(@,26)[—l]

Imoku3(@,52)[—27]

>

MD Close(@)[—l]

m Close(@)[-1]

> Imoku4(@,9,26,81m,1)

[‘27] Long ml't.Imokul(@,9)[—ll

< Imoku2(@,26)[~ll

Imokut3(@,52)(-27l

OR Close(@)[-1]

OR Close(@)[-1}


MA(@.Sim,200)[—1l

Long Exit.-

.....

RSI(@.9) [-1] > 65 OR OpenPositionAverageEntryPrice (@,ThisTradeOnly) - DollarZPricetG,7500) / OpenPositionSize (@,ThisTradeOnly) Short Entry.RSI(@,9)[—1]

> 65 AND Close(@)[-l]

< MA(@,Sim,200)[—1]

Short Exit: RSI(@,9) [—1]

< 35 OR OpenPositionAverageEntryPrice

(@,ThisTradeOn1y)

- Dollar2Price(@,7SOO)

/ OpenPositionSize

(@, ThisTradeOnly)

Asexpected, the mean reversion system enjoyed asuperior percentage. ofwinning trades as well as a deterioration ofaverage profit to average loss ratio when compm‘ed with various trend-folkiwu'ig systems such as Bolln'iger Ban'd breakout, lchimoku cloud crossover, and so on. RSI Euremes System with Volume Filler Since electronic trading haselinm‘iated the problem oflagging volume data on exchange-traded futures contracts. Iinclude. amodified version ofthe original- RSI Extremes system with avolume filter that allows entry only ifvolume decreases as themarket achieves its extreme RSI reading (see Fi'gire 8.10).2 Usm‘gCQG, theprograi'iunu'ig code for RSIExtremes with volume filter iswritten this way: Long Entry.-

RSI(@,9)[—l] < 35 AND Close(@)[-1] V01(@)[~2] > Vol(@)[-1]

> MA(@,Sim,200){-1] AND

Long Exit: RSI(@,9) [-1] > 65 OR OpenPositionAverageEntryPrice

(@,ThisTradeOnly)

k

lhl)‘

TRADING TOOLS AND TECHNIQUES

Short: Entry.-

RSI(@,9) [-1] > 65 AND Close(@) [—1] < MA(@,Sim,200)l-11 AND Vol(@) [-2] > Vol(@) [-1] Short Ex1’t.RSI(@,9)[—1]

< 35 OR OpenPosx't1'onerageEntryPr1'ce

(@,ThisTradeOnly)

— DollarZPrice(@,7500)

/ OpenPositionSize

(@,Th1'sTradeOnly)

Although the volume. filter did not. dramatically improve our mean re version system, we did enjoy fairly consistent, moderate improvements when compared with the original version in terms of profit to maximum drawdown ratio as well as average profit to average loss ratio (see Tables 8.8an‘d 8.9).

123730 - ImYO , - IZMM' y 45°

mm: > .’-.‘ '.' ow mun-oi m u .' must-twainOll'T'ml* L. A v.‘lm I «mum ‘ vormluiw l Mind“)! I I -

W m .

‘xCoMechuls. vComocLoun' ~- Iamfduxor-v u- lolilLouflalio: Inollqnmn ' i VITMIIUminimum tor-Mn

‘

vgfit

| l u 70 v20 no

FIGI’RE 8J0 Equalized Active Daily Continuation Chart for CME Group E-Mim S&P 500 Index Futures Using RSI Extremes with Volume Filter Note: Trade summary includes data from January 1, 2000, to December 3i, 2009! and assumes $10 round-turn deductions for slippage and commissions. Source: CQC. Inc. ©2010. All rights reserved worldwide.

lb‘l

How to Use Trading Models TABLE 8.9 RSI Extremes with Volume Filter

ES CL NC 6C ZS ZN EU6 1Y6 SB Total:

Profit

10,100 23,480 36,830 13,390 (13,750) (8,078) 21,160 1,535 11,630 96,297

#Trades

#Days

Max Draw

MCL

PzMD

°/.w

Pu.

20 26 43 25 15 25 24 24 22

32 24 17 29 26 34 22 26 29

(9,145) (35,180) (45,130) (17,050) (27,665) (16,128) (14,390) (20,117) (5,331)

1 4 4 1 3 2 2 3 1

1.10 .67 .82 .79 N/A N/A 1.47 .08 2.18

70 73.08 51.16 64 60 56 66.67 58.33 86.36

.74 .63 1.19 .98 .45 .52 .75 .75 .78

224

27

(75,901)

00

Asset

1.27

64.29

.72

Notes: Portfolio summary includes data from January 1,2000, to December 31, 2009. and assumes $10 round-turn deductions for slippage and commissions.

Source: CQG, Inc. {C2010. All rights reserved worldwide.

Combining .Voncorrelated Sistems Aquick glance at the Totals row ofTables 8.1 to 8.9 suggests that trading adiversrfi‘ed portfolio ofassets makes many positive expectancy models more robust. Although asset class diversrfi'cation isthemost common form seen In trading, tlu‘s section introduces readers to trade system le'efSIfi— cation. Ingeneral, trade system diverSIfi'cation capitalizes on the prm‘ciple that markets can do only two things, trend or trade in arange; therefore by Simultaneously executm‘g both trend—following and mean reversion mod— els, we can improve. our risk-adjusted rate of return. Since both models enjoy positive expectancy overall and capitallz'e on dif'ferent typesofmarketbehavior. itisreasonable toassume that when atrend-following model experiences itsequity drawdown because ofchoppy markets. the. mean reversion model's abih'ty to capitahz'e on this same choppin'ess will temper theseverity ofthe drawdown (andvice versa for trending markets). Although anyofthetrend-following andmean reversion models shown in‘Table 8.10 could beused to illustrate trade system diversification. Table 811 combines two ofthe most robust throwaway models shovm m‘ tlu's Chapter, RSI Trend with a 10—day stop and RSI Extremes with the volume filter. As expected, theprofitsofthetwo models are additive. but the worst

' 4::m.is. ‘

peak-tovalley drawdown 1n' account equity isnot. Therefore, the profit to minimum drawdown for the combination ofthese two models issupen‘or 10 either model as a standalone system. In addition, al'though the combmled perfonnance of the models still experienced less than- 50 percent “111mg trades. the addition ofour mean reversion model did m‘crease the

TRADlNG TOOLS ANDrem

I62

m1 8J0 Comparisons of Mechanical Trading Systems P1

System

Profit

#Trades

#Days

Max Draw

MCI.

PMD 36w

RSI Trendl O

341.854 115.644 319.080

884 1,595 1.020

18 6 14

(48.281) (55.207) (48,249)

13 15 12

7,08 2.09 6.61

41 29 1.97 36.74 1.92 41.18 1.85

243.409 Ichi Xover 432.086 Ichi Clouds 368.665 RSIx Volume 96.297 RSlx 89.l89

822 954 951 224 314

13 25 18 27 28

(77.073) (92.771) (73.546) (75.901) (79.012)

16 16 14 8 10

3.16 4.66 5.01 1.27 1.13

37.78 2.21 40.99 1.98 32.70 2.89 64.29 .72 64.65 .64

RSl Trend3 RSI Trend exits BB Break

Notes: Portfolio summary includes data from January 1,2000, to December 31, 2009, and assumes $10 round-turn deductions for slippage and commissmns. Source: CQC. Inc. :6,2010. All rights reserved worldwide.

1531118.! I Combining RSI Trend and RSI Extremes with Volume Filter System

Profit

#Trades

#Days

Max Draw

MCL

P:MD %W

RSI Trenle

341 ,854

RSIx Volume

96,297 438.151

884 224 1,108

18 27 20

(48.281) (75,901)

13 8 9

7.08 1.27 9.34

Combo

(46,892)

M.

41.29 19.7 64.29 .72 45.85 1.60

Notes: Portfolio summary includes data from January 1,2000. to December 31, 2009, and assumes $10 round-turn deductions fer slippage and commissions. Source: CQC, inc. C’;2010. All rights reserved worldwide.

percentage ofwinning trades from 4129‘» percent for RSI Trend as astandalone to 45.85 percent forthe combined portfolio (see Table 8.11).

NM:I‘lls‘CJIM'IC.-\I. “(IIIELS Although may ofthr- intlicatom used in classical technical analysis, such as trcndlines, rctraccnwnts. divergences, and 5(7) forth, are difficult to in— corporate into mechanical trading models, they me still quite robust and can' be used to develop rule-basal insitivc expectan'cy models.” Akey m' developing robust trad'ing models using these: indicators isrecognizm"g thal the indicator isnot a complete stand-alone trading system an'd must there fore be augmented with rules ofns'k management aS' well as exiting with profits. Although the spectrum ofpossibilities regarding exit for CIaSSI'Cfll indicators isvirtually limitless... Ifocus on regret mim’miza‘tion techniques introduced in Chapter 6.

How to Use Trading Models

I63

Fibonacci Retrarcmcnt Models

5 8..

Despite lacking a universally objective answer regarding which C\_’(‘ll(‘dl‘ highs mid lows to measure from. speculators are attracted to models incomor‘i'ting retracement theory because they participate in the dominant, longer-term trend dun’ng pullbacks, thereby ide.nti_fvm'g kiw-n'sk/high— rewar'd trading ommnunities. Although, ("har'les Dow published his concepts of buym’g or se‘lling m‘to retracemcnts within the major trend around 1%). most speculators nowadays use retrzui‘cment levels introduced by Ralph Nelson Elliott in the. 1930s. which dr'e based on Fibonacci number sequences. The most popular—an'd therefore commonly used—~01” these. Fibonacci retracement levels we the 38.2 percent. :30percent, an'd (31.8 percent rt.*tracemen$.‘ Figure 8.11 shows how Fibonacci retracement levels allow disme tionary traders to objectively quantify entry levels. risk. and reward. As stated earlier. retracement trad'ets should ask, “Do 1want more trading signals mid more false, posithfes or fewer, higher quality signals with fewer falst" positives?" For example, aggressive traders (who fear missing the trade) could sell cash U.S. dollarqlapanese yen (l."SD—JPY) at the 38.2 percent retracement level of91.17, placing a protective stop loss order above

Cycle m“

8‘ .“ Inhomnl. LII-o. pvotocliu y flops chow

i m. m

an»

5m “unusual

M, m “I” m

‘_

1

WWW?

CycI0 low I

"(SIRE 8.! I Weekly Chart ofSpot USD-JPY with Fibonaccv Retracement Levels Source' CQG, lnc. {CUZOl 0.All rights reserved worldwide.

it"

Wauw.._—u~ .. .o--. - ._...\ i.”’J

‘i’ll

l h vbnml

_ 5.{3.5. L5,

1 ‘8

;’

:mviyy7mm

3‘7Ten-mm

arr~..\:v>'r.13r.’-‘mr—

.uvs

226"

TRADER PSYCHOLOGY

Monthly Performance Record My daily trade—b_v~trade actixities are initially recorded and tracked m'a monthly perfonmuwce record. This spreadsheet is the burl'dmg' block for the longer-temi perfonnance anal'ysis spreadsheet that Icall Monthly Sum. mary Totals. Each indin‘dual monthly performance spreadsheet mll' be labe'led as the unique ('al'endar month mid year combm'ation to track pet. fonnar-ice. The monthly performance record 15' composed ofeither 21 or .3)" columns oftrack record data (ifyou are recording the time ofentry and tm'te ofexit. your spreadsheet will have, 23 columns ofdata; if'not, itWill' have 21). The columns are as follows: I. Asset—This isthe asset traded and should m'clude month and year of expiration for futures (al'ong with stn’ke and Wm or call for op.tions).

2. Position—Long or short. Although in generally we should not see a bias toward either the bull or bear side. the caveat here is' that trend followers should be. long in bull markets and short in' bear markets.

3. Entry Date—I suggest usrn'gthetm’te and date stamp usedbyyourbroker (as‘ opposed towhatever your particular tun'e zone 15'). 4. Entry Time—The shorter your hold tm'te. themore un‘portant trackrng' trade entry and exit times becomes. Scalpers and daytraders learn alot about performance and probability ofsuccess based on un'ie oftrade execution. For swu‘tg traders, trackm‘g ofcolumns 4and 8might sull‘ prove instructive, though itisusually not helpful forlong-term position traders.

5. Volume—Number ofcontracts or shares traded. watch out for trading biases such as tradln'g heavier volumes incertain assets. long

or short biases. and so forth. 6. Entry Price—Remember to average entry prices on split price fill orders. 7. Exit Date—In general, hold Me on mnm‘hg trades should be longer than on losm‘g trades. 8. Exit Tln'te—See column 4.

9. Exit Price—If you use the. regret mmm"uza'tion techniques. described in Chapter 6,parse out the volumes accordingly and show unique lme‘ items foreach exit price. 10. Gross P/L—This shows profit and loss before deductions forcommits.~ sions.

Analyzing Performance

227

11. Cash Actions—Use this column to record deposits and withdrawals to the. trading account. 12. Conunissions—Unless you are executing scalping or day trading methodologies, commissions should represent a small percentage when compared with average gross profits shown in column 10. l3. Net P/L—This shows profits and losses after deducting for commis-

sions. I4. Days—This shows the number ofcalendar dayswith an open position in'the market. In general, we should see higher numbers for winning trades and smaller numbers on losing trades. . MTM Drawdown—The column shows the worst intramonth peak-tovalley drawdown in equity. In general, we want to see this number below 5 percent of total- assets under management. Drawdowns in excess of 10 percent: are problematic an'd above 20 percent would trigger the account’s closure because ofthe triggerln’g of its fail-safe stop loss. 16. Winning Trades—If we are using the regret mininu'zation techniques described in Chapter 6,you will record percentages of the winning trade inthis column.

“mi

l7. Losing Trades—The total number oflosing trades can be. larger than the number ofwinning trades dS' long as they are small (this isespecially common to long-term trend—following models). [8. Profits—Record all partial and fullprofits inthis column. I9. Losses—Work on keeping the numbers inthis column smaller than the numbers in’theprofits column. 20. Model—If you are simultaneously executing two or more methodologies, use tlu‘s column to record which methodology was employed on

each particular trade. 2|. W Trades—If you are, usm'g the regret mm‘imization techniques described in Chapter 6, you should add the partial profits of each complete wrnn'ing trade together in this column. (Note: We do not need L Trades column Since it would be identical with the Losses column.)

22. WTime—.Use the exit time (number ofcalendar days) ofthe longest portion oftheprofitably exited trade as thewinning trade tune forWinning trades. 23. LTime—,.This isthenumber ofcalendar daysofyour losing trades. Table 11.2 is an explanation of asset symbols used in Tables 11.3 to 11.7.

. wow

228

TRADER PSYCHJLFOCY

all“, I1.2 Explanation ofAsset Swims-ms Lisa“: m Pe..r‘ormante Tables

V

».

Asset Symbol

Asset Description

RBUl O RBVl 0 111l0 CH!0 EMDUl O KCUl cm 0

r21 CME Cramp Sept 2‘; GR808 cnieaoad Casoime Futures CHE Group Oct 2m 6'R808 Unleaded Gasoline Futures

EURUSD GBPUSD AUDUSD USDCAD USDJPY

Casn Euro-t5. Doiiar Cash mssb Po-.I‘r~..d—L,‘/.S. Dollar

Source: CQC. Inc. g” 20-?

A-fz‘ {grits reserved wondmde.

CUE Group Dec 2640 Soybean 0:? Futures lCE Dec 2010 Canon Futures CME Croup Sept 2010 E-Mz‘m SP 400 Mid Cap Futures s‘CE $er 201‘ OCoflee Futures SPDR Cold Trust ETF

Cash Australian DollarvUS. Dollar Cash us. Dollar—Canadian Dollar Cash L15. Do.’lar—japanese Yen

‘5Mum-Month Performance Record ‘ Close exarmnan""on of the SPDR gold trust ETF trade (symbol GLD) en» tered on August 2010. ill'ustrates the lmutau"‘on ofour monthly performance record and why it15‘augmented with the mum—month performance record. We entered our trade inGLD on August 3ft)._.)?f'10. and were marked to-market on the trade for the month ofAugust at the settlement pn'ceof $120.91 per share. However. tth' was merely our end-of-month marl-Ho market settlement price. To best. know how the trade was termmat'ed. we must look at’ the multj'month performance record. which shows the trade entry. mf'ormation on August 51"). 3010. as well as trade exit data on Sep‘tem see Table 11.8). Also, the mum-month performance record is‘ her 1,2010 (, also used to track m'termonth peruk-to-valley equity drawdowns as well as the txm'e requtr'ed to achieve new highs m'account eqm'ty. ’' When the problems of m'termonth performance issu'es have been explam‘ed, traders someurn'es ask why we bother traclan’g monthly performance at all'. The answer is'that' hedge funds, CTAs. CP05. and so forth use monthly performance records to track endof-month performance stausr' ucs'.Even speculators. who donot aspire tomanagm"g money are adnsed.‘ to Kim'tain both monthly and mum-month track ra-ords bet.:ause your end of-year markotrrmar'ket account value “111' be med. for tax purposes (35 opposmd tothe more mf'onnatfl-e mum-month performance record}. Iuse thesame columns 111'themuln-‘month performance table asshown m' the monthly tables with the exception of the cash anions columnAlthough cash actions are important to momtr‘n, this is" already bemg' i,rar."ked m'the monthly performance tables. Bycontrast. the mum-month table is"only concealed. wuh' traclor’ig our trading performance results

229

Analyzing Performance

TABLE ".3 Monthly Performance Table—August 2010. Part 1 W Asset

Position

Entry Date

Entry Time

Volume

Entry Price

NA-

Deposit L0ng Short Long Short Long Long Long Short Short Short Short

8/1/2010 8/2/2010 3/5/2010 8/2/2010 3/6/2010 8/9/2010 8/9/2010 8/1 1/2010 8/11/2010 8/11/2010 3/1 1/2010 8/11/2010

EURUSD CTZlO

Short Short

N/A 1 200.000 1 200.000 100.000 100,000 200.000 100,000 2 100,000 100.000

N/A 2.1519 1.3161 2.1519 1.5897 1.0274 1.0274 85.43 1.3177 42.23 0.9136 0.9136

8/11/2010 8/11/2010

100.000

1.3177 81.13

EMDUlO

Short

8/11/2010

C1210

Short

8/11/2010

EMDUl 0

Short

KCU10 KCU10

Long

8/11/2010 8/16/2010

Long

8/16/2010

RBV10

Long

8/18/2010

EURUSD

Long

8/18/2010

200.000

1.2884

USDCAD

Long

8/20/2010

100.000

1.0496

GBPUSD

Long

8/24/2010

200.000

USDCAD

Long

8/20/2010

100.000

1.5512 1.0496

GLD

Long

8/26/2010

1,000

EURUSD

Long

8/27/2010

200,000

1.2713

CBPUSD

Long

8/27/2010

GLD Aug 2010

Long

8/30/2010

200.000 1,000

1.5529 121.45

N—l—l-fl—I—I—d

N/A RBUlO EURUSD RBUl0 GBPUSD USDCAD USDCAD USDJPY EURUSD 21210 AUDUSD AUDUSD

744.3 81.13 744.3 1.7865 1.7865 1.8999

121.38

Totals Notes: All performance table results are excerpted from hypothetical trading results and reproduced solely foreducational purposes. Since hypothetical back-tested results were derived from daily charts, the column for “Entry Time" is blank. Source: CQG, Inc. Q2010. All rights reserved worldwide.

Performance byAsset Record Although breaking dovm performance on an asset—by-asset baSIS’ 15' not an absolute prerequisite for understanding and elmun"at1ng' trading biases, Ipersonally find generatm'g a performance-by-asset table helpful. The performance—by-asset table should contatn‘ the same 22 colunms shown m' the multi-month performance table. The keypom'ts foranalysns‘

.mw

Mm

230

TRADER PSYCHOLOGY

Tm IL4 Monthly Performance Table—~August 2010. Part II Asset

Exit Date

N/A RBUlO EURUSD RBUIO GBPUSD USDCAD USDCAD USDJPY EURUSD ZUlO AUDUSD AUDUSD EURUSD UZ.‘0 EMDUlO mio EMDUlO KCUlO KCUlO RBVIO EURUSD USDCAD GBPUSD USDCAD GLO EURUSD GBPUSD GLD Aug 2010 Totals

N/A 8/3/2010 8/5/2010 8/5/2010 8/6/2010 8/10/2010 8/11/2010 8/11/2010 8/11/2010 8/11/2010 8/11/2010 8/11/2010 8/12/2010 8/12/2010 8/12/2010 8/12/2010 8/13/2010 8/17/2010 8/17/2010 8/18/2010 8/18/2010 8/23/2010 8/24/2010 8/26/2010 8/26/2010 8/30/2010 8/30/2010 8/30/2010

Exit Time

Exit Price

N/A 2.1745 1.3159 2.175 1.5965 1.0301 1.0389 85.1800 1.3048 42.2100 0.8999 0.8931 1.2826 80.3300 728.0000 80.9600 741.5000 1.7895 1.7870 1.8709 1.2886 1.0516 1.5464 1.0572 120.6700 1.2715 1.5531 120.9100'

Notes: All performance table results are excerpted from hypothetical trading results and reproduced solely for educational purposes; *Mark-to-Market settlement price on open position at end of month. Since hypothetical back-tested results were derived from daily charts, the column for "Exit Time" isblank. Source: CQC, Inc. ©2010. All rights reserved worldwide.

here are idenufi'cation ofany asset class biases inregard to long or short dxr‘ectional biases as well as ur'ational risk management biases.

Performance byTrading Model Record Ifyou are sun'ultaneously tradm'g multiple methodologly'es. this table compares the track records ofboth models. The performance bytradu'tg model table should also contam' the same 22 columns shown m'the mum-month

,4 .m

2:"

Analyzing Performance TABLE ILS Monthly Performance Table—August 2010. Part III

Mm

Asset

Gross

Cash

7/1.

Actions

Commissions

Net

MTM

P/L

Days Drawdown

M

N/A

N/A

RBUI 0 EURUSD

949.2 40

RBUl 0 CBWSD

970.2 ~1360

USDCAD

200,000.00 N/A

N/A 944.2 30

N/A 2 1

955.2

4

270

-1370 265

1150

2

USDCAD

1145

—500

3

USDJPY EURUSD

_510

1290

I

1285

24

1 I

ZLZI 0 AUDUSD

1370

AUDUSD

2050

2045

EURUSD

3510 400

3505 395

1630

2 2

1625

2

CTZIO EMDUIO

85 280

80

2

275

KCUIO

112.5 I8.7S

3 2

KCUIO RSVIO

1075

I 1

13_75 2 -2445 1

~2436

EURUSD

40

3o

USDCAD

200

195

I 4

—960

-970

1

760

755

7 I 4 4

CBPUSD USDCAD CLD

—71 0

EURUSD

40

—720 30

GBPUSD

40

30

CLD

. «me

I

14 1365

CTZIO EMDUIO

N/A

—~S40

—545*

N/A

a «J—

—3641

Aug 2010 Totals

8723.65

—160

8538.65 2.1

—3641

Note: All performance table results are excerpted from hypothetical trading results and reproduced solely for educational purposes. 'Mark-to-Market settlement price on open position at end ofmonth. Source: CQG, Inc. ©2010. All rights reserved worldwide. ‘

perfonnmtce table. The key pom‘ts for analysis here. are ensurm'g the robustness“ ofeach model as astzmd-alone.

\lonihlp Summary Performance Totals \rlonthly summan; totals tables offer a comprehensive View of longerterm perfomtmtce stanStics. Thespreadsheet m‘cludes 22columns oftrack record data

mam

232

TRADER PSYCHOLOGY

TABLE ILB Monthly Performance Table—August 2010, Part lV Asset

W

L

N/A RBU10 EURUSD RBU1O CBPUSD USDCAD USDCAD USDJPY EURUSD ZQTO AUDUSD AUDUSD EURUSD CTZlO EMDUlO U210 EMDUlO KCUlO KCUlO RBV10 EURUSD USDCAD CBPUSD USDCAD GLD EURUSD CBPUSD CLD Aug 2010 Totals

N/A 0.5 l 0.5

N/A

Profits

N/A 944.2 30 965.2

Losses

Model

N/A

N/A Trend MR Trend

—1370

MR MR MR

265 1145

0.5 0.5

—510

1285 14 1365 2045 3505 395 1625 80 275 107.5 13.75

0.5 l 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

—2446

30 195

MR MR Trend MR MR MR MR Trend MR Trend Trend Trend MR MR Trend

—970

0.5

755 —720

l l l 14

30 30 5

15099.65

—S4S* ~6561‘

MR Trend Trend MR MR Trend

Notes: All performance table results are excerpted from hypothetical trading results and reproduced solely for educational purposes. *Mark-to-Market settlement price on open position at end ofmonth. “Trend” is notation for trend-following model; "MR" is notation for mean reversion model. Source: CQC, Inc.

2010. All rights reserved worldwide.

The columns are as follows:

I. Month—This column displays both month and year. 2. Gross P/Ir—This shows total monthly profit or loss before deductions forcommissions. I» V

p 3:. x7. 4/ E- :

3. Commissions—This shows total monthly commissions. Unless you are executm'g scalpm’g or day—trad'ing methodologies, commissions should

t—

Analyzing Performance

233

WSW—Mill] Monthly Performance Table~~August 2010. Part v Asset

N/A RBUlO EURUSD RBUlO CBPUSD USDCAD USDCAD USJDPY EURUSD ZLZ10 AUDUSD AUDUSD EURUSD CTZlO EMDU10 U210 EMDU1O KCU10 KCU10 RBV10 EURUSD USDCAD CBPUSD USDCAD CLD EURUSD CBPUSD GLD Aug 2010 Totals

wTrades

WTime

I.Time “If”

30 1909.4

4

1410

14 3410 4790

475 1900 121.25

2

30

9'50 30 30 N/A 15,099.65

2.7

Note: All performance table results are excerpted from hypothetical trading results and reproduced solely for educational purposes. Source: CQG, lnc.

2010. All rights reserved worldwide.

represent around 1to 2 percent of average gross profits shown in column 2.

4_ Net p/L__This Shows total monthly profit or loss after deducting for commissions. 0".Number ofTrades—This column alerts us to overtrading tendencies. Although scalpers and day traders will have higher numbers, in general, the smaller your number ofmonthly trades, the better.

4%,

-

lune:

2M

TRADER PSYCHOLOGY

“Bil: IL8 Tables

Comparison of Monthly versus Multi-Month Results for GLD Trade

Exit Date

Exit Price Cross P/L Commissions Net P/L —$540.00* —SS.OO‘

Aug 20l0

8,1:305‘2010 Sl20.9l'

Mum-Month

9/1/20l0

Sl22.51"

$530.00

—SS.OO

Mum-Month

9/l/20l0

$21.47"

$l0.00

-»$5.00

Days

$545.00‘ N/A $525.00 2 $5.00 2

Note‘ All performance table results are excerpted from hypothetical trading results and reproduced solely for educational purposes. ’Mark~to-Market settlement price on open position at end of month. "Exit price on 500 shares (50 percent oftotal volumetric position). Source: CQG, Inc. 32m 0.All rights reserved worldwide.

6. Worst Draw—Tins isthe largest intramonth dl'ld intermonth peak-to valley drawdowns in dollars. Use an' asterisk to denote intermonth peak-to-valley drawdowns u'iequity.

. Start Acct Bal—Tlu's shows theaccount balmtce atthebeginning ofthe month.

ROR *\>—-This' shows the rate ofreturn for the month and is Net P/L from column 4din‘ded bythe Start Acct Ba] incolumn 7(unless intramonth deposits and withdrawals occurred that month). CD

. Cash Actions—This column shows all deposits and withdrawal's tothe trading accmmt.

IO. EOM Acct Bal—T'hls' shows the account balance at the end ofmonth and will also be used as thefollowing month’s Start Acct Ba] amount. ll. W—This' column isthe total number ofwinning trades for the month. 12. L—This column is'the total number oflosing trades forthe month. l3. AvgProfit—This column shows average profit on ilzlwinning trades. l4. Avg Loss—This column shows average loss forall losing trades. lo. %W

Tlu's column shows the percentage of winning trades for the month and is calculated by taku'ig column 11 dl‘ld din'ding it bycolumn 5. Although most traders find iteasier to stick with methodolo gies eru'oyin'g higher winning percentages. some. very successful longterm trend traders conSIS'tently experience less” that 50 percent “1an nn'ig trades. The key to their success isvery robust P:L Ratios (See column 17) and W/L Times (seecoltuun 23.

16. Time—Tins shows the number ofcalendar' days with an open position

inthe market. I7. P/L Ratio—This shows the ratio ofaverage profit toaverage loss andis.

calculated bytaking column 13and dividing itbycolumn 14.Ingeneral

j“flavor 7'flWfijWfinfifl new". ' ‘

v. Fri-Zr» ’h 33???? ':«hr-v rmeI-«Ww- efwtiv'r.ewe—«em

If

Analyzing Performance

233

TABLE ILO Monthly Summary Totals, Part l

M

Month

Gross P/L

Commissions

Net P/L

itofTrades

Worst Draw

W

Au92010 SepZOIO

8,723.65 10.556.00

«160.0 —197.5

8,563.65 10,358.50

19 21

-r3,64l -~—1.810

Total:

19,279.65

7-3575

18,922.15

40

-3.641

Note: All performance table results are excerpted from hypothetical trading results and reproduced solely for educational purposes. Source: CQC. lnc. QC" 2010. All rights reserved worldwide.

traders want this ratio to be significantly above, 1.0, unless their 96W (see column 15) issignifican'tly greater than 50 percent. l8. % Draw—This is the lar'gest percentag'e intramonth an'd intermonth peak-tovadley drawdowns. It iscalculated bydividing the largest dollar an'iount. drawdown shown in column 6bythe. total account equity before the beginning of that rlrawdown. Use an asterisk to denote an intennonth peak-to—valley drawdown in equity. 19. Days Draw—This shows the longest intramonth or m‘termonth drawdown. Remember that the longest drawdown is not necessarily the largest. Use' an as'tert‘sk to denote an' inter-month peak-to—valley drawdown. 20. Mont.th PzMD—This measures how much net monthly profit was generated Visa-W's how much risk was endured to generate that profit. The higher this' number, the. more robust the methodologies traded. It is calculated bydividing Net P/L from column 4byWorst Draw of column 6. 2|. P:MD——This' measures how much total (mum-month or multiyear) net profit was generated Vis-a—vis how much risk was endured to generate that profit. Because. this number iscumulative net profits and theworst peak-to-valley drawdovm will not necessarily change over time. this"

TABLE IL“)

Monthly Summary Totals, Part ll

Month

Start Acct Bal

ROR %

Cash Actions

EOM Acct Bal

W

Aug 2010 Sep 2010

200,000.00 208,563.65

4.28 4.97

200,000 N/A

208,563.65 218,922.15

14 16

5 5

Total.

N/A

9.46

200.000

218,922.15

30

10

L

Note: All performance table results are excerpted from hypothetical trading results and reproduced solely for educational purposes. Source: CQC, Inc. .9”,2010. All rights reserved worldwide.

236

TRADE’ Rl’SYC H01OCY

‘IWBH‘I II.“

Monthly Summary ‘1ottils. Part Ill

Month

Avg Profit

Aug 2010 Sep 2010

1161.51 1059.51"

Total:

1106.90

3‘W

Tlme

P:L Ratio

36Draw

-1093.50 919.17

73.68 76.19

1.80 2.77

1.06 1.15

1.72 0.87

1116.60

75.00

2.33

0.99

Avg Loss

M- ..- “finuv... .M ~. ..._...._--._ .-

1.72 .4. ... ._...___

Note' All performance table results are excerpted from hypothetical trading results and reproduced solely for educational purposes.

Source (QC. Inc. kc‘ 2010. All rights reserved worldwide.

TABLE II.” Monthly Summary Totals. Part IV Month

Days Draw

Monthly P:MD

P:MD

W/LTime

Aug 2010 Sep 2010

12 6

2.35 5.72

2.35 5.20

1.71 2.97

Total;

12

N/A

5.20

2.44

Note. All performance table results are excerpted from hypothetical trading results and reproduced solely for educational purposes.

Source: CQC, inc. \c‘2010. All rights reserved worldwide.

number should increase Sigmitictmtly over time when compared to the \iiouthly MID. 22. W1. 'I‘imcgllere the total average holding time of 11.11 winning positions isdivided bythe total average holding time ofall losing positions. In general. tmders strive to make this number larger as their skills improve.

Monthly Summan Performance Totals by “l‘radint.0 \lodel Ifyou are simultaneoush-' executing multiple trading models. it ishelpful to break down performmtce by each trmling model. as this alerts us to strengths mid weaknesses ofeach model as a stand-alone as well as how simultzuieous implementation enhances oven-1,11 perforuumce. The spreadsheets are composed ofthe same 22columns used inthe monthly stiiiiiiian," totals tahle.

FI\:\I. THIN GII'I‘S

___________.__‘_ __________._,‘.....—-r¢

People tend to have unrealistic beliefs about the growth and developmem process intrader psychology. We imagine the removal oftrading biases 35

av my “\,v»t-,-.I~

Analyzing Performance

237

a yes—n0. bias or no bias proposition. Think instead of emotional growth and development like a spiral ofmusical notes ina scale. Our psychomgical development (_ in most instances) isnot linear; we ms'tead expen‘ence stronger an'd weaker octaves of the same emotions such as fear, greed, pride, J'eFthllSV.'. and so on. For example, as we continue working on ourselves as traders, itmight appear that we have completely eradicated asst» ciations ofemotional pain with losses. This belief could even bevalidated byenduring a loss without any association ofemotional pain whatsoever. However. inmost instances what has happened 5'resolution ofastronger octave ofassociations of emotional painwith loss, and when alarger-thanaverage-sized loss or astring ofconsecutive losses occurs, emotional pam' resm’faces atthese weaker octaves. Comparing the C sharp note. to the emotional pam‘ expen'enced by traders enduring losses, as beginners we feel a stronger C sharp note, whereas intermediate-level traders feel the same Csharp note but it15'a weaker. less destructive octave ofthis same pain oflosing. Finally. even master traders feel this sat-tie Csharp' note ofemotional pam' after losses, but they have trained themselves to embrace and release the emotion almost ms'tantaneously. Byrealizing that the resolution of trader biases” occurs in stages. we are emotionally prepared for their recurrence on these weaker octaves. Consequently, when recurrence does instead offrustration ordespondency. we recogruz‘e tth' weaker octave ofemotion forwhat itis,en'dence ofour maturation from bem'g emotionally crippled bylosses to a tempered acceptance of this weaker octave. which still requrr‘es resolution to shift‘ us from the octave ofcompetent trader to that ofmaster trader. Other. more ideological, tradln'g biases also tend to be resolved tn‘ stages. For example, a beginnr'ng speculator nu‘ght display a bias toward gom’glongor short, trading on quiet daysas opposed todays when govemment reports are released, or Fridays versus \rlondays. Intermediate traders could have resolved such strong biases but could still display biases toward trading ICE Brent Crude Oil versus CME Group WI‘I‘ Crude 011'. Even advanced traders who have resolved such m'termediate-level biases might still dls'play biases toward trend following as opposed tocountertrend trading.”Finally, tradm'g biases correspond to developmental stages ofspeculators. Begrnn‘ing traders display extremely destructive biases such as biases agains't admittm‘g that theyare wrong as well as biases toward sm‘all profits and large losses. These biases are so destructive to traders that they lead to emotional breakdown and failure, which forces us to acknowledge the realities ofthe market’s multifaceted. uncertain. andever-changing nature. Itisat this developmental stage~following an account blomrp—that’ we have the potential to learn the reality ofbeing stuck between the prov-er.~ bial rock (ofnot wanting to lose) and a hard place (ofregretting missed‘

u,“

. :on

>1-un

'----'.n :-

“Ix-A

‘ -' wan.

«on...

cam

238

TRADER PSYCHOLOGY

profit opportunities). According to SunTzu, “Do not press an enemy atbay (because) . . . if they know there is no altemative, they will fight to the death.”“ In other words, byrecognizing that fear of loss leads to regret over missed opportunities forprofit and participation so as tominimize re gret over missed opportunities leads to fear oflosses, we are sufficiently motivated to develop and adhere to acasino paradigm method irrespective of the outcome, adopting a “whatever happens, happens” attitude. Ironically, itisoftentimes our pairing ofthis attitude with the casino paradigm methodology that marks the transition from novice to intermediate-level trading skills. R(*’l]l€lllb(,‘l‘, adoption of this “whatever happens, happens” attitude is not areckless disregard for risk. We in'stead manage the risk, acknowledgingour fear' ofloss, but we feel that fear and execute the casino paradigm method anyway (see (L‘hapter 12). We. acknowledge the. possibility ofloss, but have sufficiently matured as traders so that we recognize acceptance ofthis possibility as the pn'ce paidto minimize theregret ofmissing opportunities forprofit. The trzuisition from beginner to intermediate trader is unparalleled throughout thecareer ofthespeculator because itrepresents our shiftfrom failure tosuccess. Acommon problem among int‘emiediate-level traders is allowing moderate degrees ofsuccess and achievement ofinitial financial goals to devolve into complacency and risk aversion. Complacency and risk aversion typically ar'ise from laziness or an irrational clinging to suboptimal methods due to an erroneous belief that the only altemative isa return to previous experiences of emotional chaos and financial ruin'. If the issue is complacency leading to irrational levels ofrisk aversion, the antidote isrefocusing on probabilities and committing to spec1fi'c, cuttm'g~ edge performance goals. Bycontrast, ifthe problem isan irrational fear of failure, the antidote is research, including development, back testing, and optimization ofmechanical trading systems. The more time dedicated to research and testing ofother positive expectancy models, the more easily irrational bias'es regar'ding our model can be eliminated. Researching other models fosters an attitude ofopen-mm’ded inquisitiveness that promotes modification ofour own methodology inaccordance with our unique trading personality. The transition from intermediate to advanced trader is often marked byincorporation ofintuitive skills—~which come from experience—to augment purely mechanical methods. Advanced traders also use intemal irrational emotional'ism as' abarom— eter to access intuition regar'ding consensus mental-ity (which is almost always wrong). For example, whenever a par'abolic move in my favor occurs and Iimagine n‘itaclysmic events that could push the mar'ket [0 new all-time. highs. Icalmly notice the euphoria and inuuediately exit.

Analyzing Performance

TABLE II.I3

39

Overview ofTrader Skills

Skill Level

Methodology

Risk Management

Discipline

Challenge

Beginner

None

None

None

Develop casino paradigm method

Intermediate

Rule based

Adheres to

Complacency,

position

rules with

risk aversion

sizing

rare, minor

Stop losses,

breakdowns Advanced

Rule based, augmented by intuition

Risk management pyramid

Near-flawless to flawless

Develop new models, refine existing models

Note: These categories are offered for illustrative purposes only and should not be thought ofas hard and fast delineators ofbeginner, intermediate or advanced trading skill levels.

50percent ofthe position, while raising my stop on the remainder. More often than( not, emotionally charged daydreams ofcataclysmic events, al'l— time new highs, and so forth, mark the. peak ortrough ofamarket move. Advanced traders hone. self-awareness tosuch an extent that theycan distinguish hopu'ig from intuition. The easiest way to differentiate. hoping from intuition is an internal monitoring ofemotions. Wishing isan' emotionally charged superimposition of our subjective beliefs onto the market’s behavior. Bycontrast, intuition isan emotionally neutral, objective perception ofthe market's truth. Fln'ally, although Ihave tried to delineate. basic characten'stics andtendencies oftraders atvarious developmental stages oftheir careers, part of What distinguishes the advanced, or master, trader from u'ttennediate or begmn'ers isan ongoing commitment to growth and refinement. ofskills. Theyrecogmz‘e that there isno static plateau ofmastery in'trading. that itis' afluidprocess ofrefinement, and thatsuccess intrading, asthecliche goes, “.. . isa journey, not a destination.” Whereas intermediate-skilled traders struggle to getback into their old groove ofsuccessful trading aft'er asetback,master traders realize there isno idealized pastto return to.that you can never step into the same. river twice (because. n'vers. like the markets. are multifaceted, uncertain, and ever changing). They therefore. View so-._ called setbacks as opportunities toperceive previously unconscious blind Spots intheir skills and use, them toachieve new heights ofsuccess intrad~

m'g(seeTable 11.13).

(.‘ll‘l’Tl‘IR l2

Becoming an Even-Tempered Trader He who binds to himsr’jl'u joy both [he winged (Uh(Ins/my. But he who kisses (hejoy as itflies, Lives in eternity'ssunrise. —William Blake.

T raders cannot afford n'gid beliefs. While beginners are susceptible to extremely destructive biases such as exiting profits quickly, lettm‘g losers run, and so on, even experienced traders can in'iprove perfor— mance by overcoming subtler trading biases. For example. although expen‘enced traders real'ize that they cannot be rigidly- bullL'sh or bearish. their flexibility often falls apart when itcomes to modification ofpositive expectancy models or ns'k management methodologies. This chapter examines a wide array of psychological and somatic tools and techm'ques. including even-mindedness, meditation, visualization. and research to aid in‘ tempering emotionalism, promoting creatifity. and overcoming van-'ous trader biases.

THE “I DON’T CARE” 0LT While (lining with some childhood friends (who are not inthe industry) one asked what Itaught. traders. After explw'ning it. another replied. “I get it, you're the ‘Idon't care' guy.”Although a humorous simplification oftrader psychology, in many ways his FPSDOHM" was right. Emotional'ism in‘ tradingdoes not work. As longas you are not reckless“ about risk management

24]

242

'l'RAUI llPf.fl'HUIMi“!

while exeeuting a positive expeetaney model, you h‘ltouhl not be emotionally attaehed to the results ofyour Inides, Ityou do «me, then you either haven't done enough researi'h to be eerlaltl that itis n posiliw expt-w-tsim-y methodology, you’re not managing the risk, you're letting prevents newt live tr.'iding experienees s.'.‘ibot.'ige your edge, (ll you’re :uldirted to the waitbler‘s mentality ot'needing to win as opposed to knowing you will sou eedi This is the only industry in whieh individuals destined to (-xpt‘l'lt‘llt'r 10,000 to 1,000,000 data points (or trades) We: the eourse of their mieem obsess over the results ofa single data point, The «little about only helm; as good as your last trade isboth untrue :ind psyehillogimlly dmtmvtlw The antidote is remembering that, throughout your (".il'eel' you will expe rienee everything from protits when the market. missed your stop loss by one tiek‘ to losses when your stop was the high or low pl‘lt‘e. ityou cling to each of these experienees, you will ride the emotional roller eosmler ofeuphoria and depression ad inlinitum. The roller eu.‘th‘tel‘ is exlumsting, demoralizing, and leads to eareer burnout (espeeially in eleeminleztlly traded, 24—hour markets). But there isan alternative. Instezul ol' obsessing oVer past. losses, pre~ mature profitable exits, and so on, fonts on market. opportunities offered in the present. moment with emotional eveiHnindedness while stlnulljmn» ously learning from past, errors so that you trade more el't'eetiVely now and in the future. What is even—mindedness‘.’ it, means muting without attw'lt ment to winning or aversion to losing on any single trade. 1often eontpare nmster trmlers to .‘i'etuaries who pore over statistieal tables so as to better determine probabilities and risk. When the unlikely loss does oeeur, they do not. imagine themselves as failures and abandon the profession; lhey in stead reeognize itas aeost ot'doing lmsiness. Eveninindedness tm-hniques stn’ve to temper emotional reaetivity to the results ot‘a single trade. Insteml ofobsessing over the outeome of this single trade, realize that as long as" you are managing the risk and adhering to a positiVe expeetaney model, a year" from now you will not even remember this trade. As opposed to defining sueeess or failure based on whether a single trade was aprofit or a loss, foeus on the (.‘db’llltt paradigm proeess and iiieasure .s'ueees‘s bythe degree to which you demonstrated diseiplined .‘nlherem-e to the positive expectaney model and rules ol'risk managtfluent. Even—mindedmiss means eonsishently operating in the middle ground of tempered enmtiionalism. 'I‘ypieally, beginning (and to a lesswr extent inte.miediate-level) traders cycle from extremes of greed-driven reckless ness inwhich prudent risk management ruhts are abandoned and panic in which fear ofloss sabotages their ability to stleeessl'ully exeeule the positive expeetwtey model byeither preventing them from plaeing entry order“; sett ingstop loss orders too tight ( too (:lo.s(.- toentry priee levels), or moving stops from loss to bredk'even levels prematurely (see: Table 12.ll.

Becoming an Even—Tempered Trader “nu: IZJ

w; .

Emotional Spectrum ofTrading

W

Even-

Reckless

Blind

Paralysis Fear Recklessness After Entry Mindedness M Unable to

Adjusting

initiate

Stops

positions

prematurely; Premature profit—taking

i

V.

2-13

Flawless execution of model

Deviating

Not quantifying

from

risk wtth stops;

model» adding risk

Overleveraging

after entry

After a trade, irrespectivr.‘ ofwhether itwas" a profit or loss. traders must fight tendencies toward fearfulness or recklessness. Following a winning trade. l‘earfulness can inai-u‘fest as rut unwillingness to give back profits. Some speculators stop trading altogether once tltey achieve their monthly profit tar'get so they do not lose itall back. The antidote to this tendency isremembering each trading opportunity isunique an-d unrelated. to pren’ous or subsequent opponunities. Just because you me up 10percent on the month has absolutely no hean'ng on whether the next trade, will beprofitable. Remember. the market isunaware ofwhether you are up 10 percent or down percent. Itisinstead continuing to offer opportunities to pzuticipate inyour positive expectancy model. itisup to you whether you superimpose artificial ceilings on perfomian'ce merely because you have. not yettumed the calendar ahead toanew month. On the, other hand, recklessness can arise from euphoric delusions of in'xincibility as well as traders imagining they can abandon disciplined ad~ herence to risk management or the positive expectancy model since they are now playing with the house’s money. The. antidote here isrememberingthat there isno such thing as the house's money and that as soon as the market generates aprofitable mark—to-market. those profits me yours. That the profits are yours is reflected byyour brokerage statement. Itis consequently just as irrational and irresponsible for you to be more reckless after profits as itisfor fear to derail your continued adherence to a positive expectancy model after losses. After losingtrades fearfulness can mzuu'fest asparalysis preventing the placement ofentry orders, setting stop loss orders too tightly (that too close to entry price levels), or moving stops from losss‘to hreakeven levels prematurely. The antidote isremembering that the model enjoys positive expectancy and our fear" istherefore irrationally tied tomemories ofprefious losses and iscounteiproductive. Reckless alumdonment ofthe model

or risk management can also arise utter losses from our desire to recuperate quickly. The antidote to recklessfness after losses ispatience and rememben‘ng that markets offer opponunities for profit more. quicldy and

TRADER 5ch,

m-amnwex’mw,‘“9

V

afi'er suffvn‘ng 3 km. T)“1 “51m

ads on. whmhvr

x-

j

u

‘

.

‘

c

“'9 9‘“ “"‘d ""r l"~--"‘"'i\‘9 (*Xl’i‘t‘tflqm‘:

‘ R‘~"“"mh“r. than» an:- al. managmg “‘9 “Sk' {Mam f a gnud hm. Winning ’ from 'hkllefiflsuccfl."fufl)’ (“or winning Mngodde 7 “ i if outmmes: " ’ w trading had bet. a (‘nw‘qummy mm possible beL m1dlm..mg gem a “r discipline from managpnwm’ 105mg risk badfmnm amistake in mmk‘hng. mm 090mm mom ofivmvflyfl h 1. errors 5‘) we such from to iearn Our flzlwlms Pmufim we “5%” “Turf?” (5951)"? sx‘wh»MW as an m the H nwd m-avat the {mum Howewr. merely we o;the am paradigm, and continue our dlwmlinm Momma pnfimfiion ' {our ‘ { mndfl. exxw‘tmwy "mans m xxm‘uw Pfl'fl-mlmkwhwm of1h? ('arfulnoss, . or mum’s m M‘klm

A; acm’ummn‘ymg (W,'