The Unrules E-Book

Igor Tulchinsky

• Herausgeber: John Wiley & Sons
• Kategorie: Fachliteratur
• Sprache: Englisch

Seitenzahl: 248

Veröffentlichungsjahr: 2018

Beschreibung

Learn from a master of quantitative finance the rules that made him a success. The UnRules presents the dynamic rules for success in the age of exponential information. Written by Igor Tulchinsky, the trader behind global quantitative investment management firm WorldQuant, this book is more than just another Big Data guide for financial wonks -- it's a prescriptive, inspirational book for everyone navigating the tidal waves of the information age. Data is everywhere, coming at us in a never-ceasing, ever-rising river that threatens to overwhelm us. Tulchinsky shows us, however, how natural patterns underlie that data -- patterns that may dictate life or death, success or failure. The marriage of man and machines has allowed scientists to explore increasingly complex worlds, to predict outcomes and eventualities. This book demonstrates how to exercise real intelligence by discerning the patterns that surround us every day and how to leverage this information into success in the workplace and beyond. Igor Tulchinsky has spent his career discerning meaningful patterns in information. For decades, Tulchinsky has been at the forefront of developing predictive trading algorithms known as alphas -- a quest that has led Tulchinsky to explore the nature of markets, the fundamentals of risk and reward, and the science behind complex nonlinear systems. Tulchinsky explains what we know of these systems, both natural and man-made, in accessible and personal terms, and he shares how alphas have driven his success as an investor and shaped his central "UnRule," which is that no rule applies in every case. As markets evolve, even the most effective trading algorithms weaken over time. Decades of creating successful alphas -- and learning how to effectively transform them into strategies -- have taught Tulchinsky about the need to combine flexibility and focus, discipline and creativity when building complex models. At a time when data and computing power are exploding exponentially, The UnRules provides an expert introduction to our increasingly quantitative world.

Leseprobe

Table of Contents

Cover

Foreword

Preface

CHAPTER 1: Quake

CHAPTER 2: The UnRule that Rules the Rest

CHAPTER 3: Parallel Universes

CHAPTER 4: Signal and Noise

CHAPTER 5: Waves

CHAPTER 6: Correlation

CHAPTER 7: Scaling Up

CHAPTER 8: An Exponential World

CHAPTER 9: Quant Biology

CHAPTER 10: The Age of Prediction

Index

End User License Agreement

Guide

Cover

Table of Contents

Begin Reading

Pages

C1

iii

iv

vii

viii

ix

xi

xii

xiii

xiv

1

3

4

5

6

7

8

9

10

11

13

14

15

16

17

18

19

21

23

24

25

26

27

28

29

30

31

32

33

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

51

52

53

54

55

56

57

58

59

60

61

63

65

66

67

68

69

70

71

72

73

74

75

76

77

79

81

82

83

84

85

86

87

88

89

90

91

92

93

95

97

98

99

100

101

102

103

104

105

106

107

108

109

111

112

113

114

115

116

117

118

119

120

121

122

123

125

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

E1

The UnRules

Man, Machines and the Quest to Master Markets

IGOR TULCHINSKY

This edition first published 2018

© 2018 Igor Tulchinsky

Registered office

John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, United Kingdom

For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.

Wiley publishes in a variety of print and electronic formats and by print‐on‐demand. Some material included with standard print versions of this book may not be included in e‐books or in print‐on‐demand. If this book refers to media such as a CD or DVD that is not included in the version you purchased, you may download this material at http://booksupport.wiley.com. For more information about Wiley products, visit www.wiley.com.

Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book.

Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. It is sold on the understanding that the publisher is not engaged in rendering professional services and neither the publisher nor the author shall be liable for damages arising herefrom. If professional advice or other expert assistance is required, the services of a competent professional should be sought.

Library of Congress Cataloging‐in‐Publication Data

Names: Tulchinsky, Igor, 1966– author.

Title: The unrules : man, machines and the quest to master markets / by Igor Tulchinsky.

Description: Chichester, West Sussex, United Kingdom : John Wiley & Sons, 2018. | Includes index. |

Identifiers: LCCN 2018003403 (print) | LCCN 2018005290 (ebook) | ISBN 9781119372110 (pdf) | ISBN 9781119372127 (epub) | ISBN 9781119372103 (cloth)

Subjects: LCSH: Success in business. | Strategic planning. | Information technology. | Information society. | Tulchinsky, Igor, 1966–

Classification: LCC HF5386 (ebook) | LCC HF5386 .T82865 2018 (print) | DDC 650.1—dc23

LC record available at https://lccn.loc.gov/2018003403

Cover Design: Ed Johnson

Foreword

Igor Tulchinsky and I had very different formative experiences. His childhood was constrained by the spiritual oppression of life in the Soviet Union, while mine was enriched by the opportunities available to middle‐class kids in 1950s' America. Yet we had much in common: caring parents, a love of reading, and a fascination with math.

As one of today's leading quantitative investors, Igor understands better than most the numbers that underlie dynamic markets. “Markets can be seen as waves,” he writes. “They resemble the regular oscillations of a musical instrument.” That's a valid observation, although different from the way I came to learn about business and finance. As a college student, I was influenced by the writings of the late Nobel Laureate Gary Becker, and by personal experiences that made me realize how many aspiring entrepreneurs – especially minorities and women – were being denied access to capital.

Igor's approach has relied on rigorous and sophisticated mathematical analysis to identify trading opportunities. This might seem very different from a reliance on theories of human capital – the talent, training, and experiences of people – and the effects of societal trends on business success that I use. But in reality, we both seek to predict the most likely future based on what we observe. Understanding numbers and understanding people can both yield important insights that contribute to financial success. And we concur on several important points that are discussed in this book:

All markets contain risk, and without risk there are no gains. Careful research can discover the price of risk more accurately.

Markets also contain psychological traps, such as confusing correlation with causation. If most people are ensnared by these traps, an objective investor who follows the research – like the proverbial one‐eyed man in the land of the blind – has an advantage.

The best investors seek and distill advice from widely diverse sources.

The study of markets and the study of biology have much in common. Each is a data‐driven information science; each uses predictive algorithms in seeking a needle in a haystack of data. As Igor points out, the next great disease breakthrough might be discovered using the same mathematical techniques he uses to analyze financial data.

Talent is distributed around the world. Genius lives everywhere.

Igor and I both also believe in history's important lessons. A 2010 book about financial markets said that “real estate prices collapsed, credit dried up and house building stopped.” That sounds like a description of 2008. But it actually refers to 1792, during the administration of George Washington. More recently, stock markets dropped sharply, banks curtailed lending, and unemployment rose to double digits. Again, that wasn't 2008, it was 1974. Live long enough and you begin to appreciate what remains constant through cycles of history. Yet also note that history isn't a sine wave that repeats patterns exactly; it's more like a helix – similar events return in a different orbit. This is why research is crucial.

Investors who conduct careful research are usually better insulated against inevitable market downturns. They understand that the value of debt securities underpins all capital markets, that leverage is a dangerous tool in volatile markets, that ratings are not always a reliable measure of credit quality, that interest rates are not predictable, and that government actions often distort markets.

Although these basic investing principles change little over time, the tools of finance have changed dramatically. When I studied quantitative economics at Berkeley in the 1960s, computers were expensive, relatively inaccessible, room‐sized machines with little power to model investment scenarios. By 1976 processing was speedier, but the storage cost for the IBM System/370 that my business installed was still $1 million per megabyte. Today data processing is millions of times faster, available to nearly anyone on earth, with virtually infinite storage in the cloud at a cost that approaches zero.

This technology revolution has changed the world in many fields. Its impact on biomedical research and precision medicine, for example, has accelerated clinical science and saved untold numbers of lives. There is great opportunity for it to advance beyond its current state through partnerships such as the WorldQuant Initiative for Quantitative Prediction at Weill Cornell, which Igor founded. In the area of finance and investing, Igor and his colleagues now can do what 1960s' finance students could only dream of – simulate reality by creating millions of algorithms (called alphas) that identify trading opportunities with remarkable speed and accuracy.

Although we see markets through different lenses, Igor and I are in complete agreement on one of the most important social issues of our time: providing a path to a meaningful life for every worker, no matter how much traditional work is disrupted by advancing technology. In 2017 we co‐authored a Wall Street Journal opinion article about the challenges of automation and artificial intelligence. We concluded that digital innovation and robots are opening new possibilities for workers and that the future workplace can provide the opportunity for lives of purpose. We believe, in short, that technology leverages human capital and that wisely deployed technology creates more jobs than it destroys. The key, of course, is to provide abundant opportunities for training and retraining.

The workplace of the future can already be seen in the international operations of Igor's company, WorldQuant. Separately, the WorldQuant Foundation's WorldQuant University offers students a tuition‐free online master's degree program in financial engineering. By providing opportunities for a diverse group of bright people who are willing to work hard toward a clear goal, Igor is expanding human capital and helping assure a more prosperous tomorrow. The UnRules is a valuable guide for getting there.

Michael Milken

Chairman of the Milken Institute

Preface

People who know me well are aware that I'm a man of few words. In fact, I joke that you only have so many words in life, and when you use them up, you die. Of course, now that I've written this book, I'm living dangerously.

When we are born, our languages are bestowed upon us. I was born in the Soviet Union, in Minsk, now the capital of Belarus, and I grew up speaking Russian. When my parents and I left the Soviet Union and came to the United States, in the late 1970s, we had to master English. As a child, I grasped the new language more easily than my parents did, but – as with the challenging task of adjusting to a strange new culture – we coped. Mathematics was a language I felt comfortable with. I had played chess as a child, and my parents were professional musicians; both pastimes are rooted in a mathematical, rules‐based order. Soviet schools excelled at teaching math, and when I was in middle school in Wichita, Kansas, I discovered computer programming. From the start I was drawn to the precision of early computer languages: BASIC and, later, C. When I stumbled into video game development at age 17, I was assigned to co‐write a book about video game programming. My experience in early video gaming – coming up with characters (and jokes), writing the programs, working on the book – convinced me that just about anything is possible.

This book, The UnRules, is about languages of many kinds: scientific, mathematical, computer, financial, biological. It's about codes, patterns, and signals, and the attempt to extract order from a noisy world. The notion of the UnRule, which lies at the heart of this book, is a kind of philosophy, based on empirical experiences both in financial markets and in life, where no rule, dogma, ideology, paradigm, or model lasts forever and no trading or market relationship performs as you expect all the time. Like a tether on a balloon, the UnRule limits the reach of all the other rules I've gathered over the years. For me, an intense involvement in competitive markets, and in building my career and my quantitative investment firm, WorldQuant, led me to develop rules that apply not just to trading but to life. Many of those rules are rooted in an always uncertain future. This is reflected in my firm's deep involvement in developing alphas – that is, algorithms that seek to predict certain market relationships. The alphas we develop, now numbering in the millions, consist of mathematical expressions and computer code. We rigorously back‐test them with historical market data to “simulate” their performance, just as video games simulate different realities. Much of this investment process is extensively automated.

And yet we do not just hand over trading to machines. People matter. Over the years we have learned a lot about alpha design and development. We've learned that no matter how well an alpha is back‐tested, it will probably not perform as well when we put it into real markets, and like the rules, no alpha lasts forever. We've learned about the use and dangers of correlation, the management of risk, and the deployment of extraordinary numbers of alphas. We've developed a sense of when to assume risk and, very importantly, when to take losses.

Along these lines, I have found that some life decisions have no clear solutions. For many years I made disciplined but incremental empirical decisions – hiring, for instance, only when I could find genuine talent. There was no master plan. Eventually, we discovered we could find the brightest people in quantitative fields and teach them finance. Smart, motivated people learn quickly. That search for talent transformed WorldQuant into a global firm, exploiting the fact that talent is universal but opportunities are not.

My parents and I had to risk a long journey to America to find the freedom to take advantage of opportunities. Today WorldQuant offers citizens of many nations those same chances, while allowing them to remain at home – in Bulgaria, China, India, Israel, Russia, and Vietnam, among other countries. That recognition that talent requires opportunity also lies behind my recent philanthropic efforts to provide free online education in quantitative disciplines through WorldQuant University, a not‐for‐profit entity legally separate from the firm.

Today we find ourselves in exciting scientific and technological waters. The drive of any investment firm is to try to predict the path of a market's complex turbulence, which we have labored to decipher and define through alphas. But prediction is never easy. There is an unresolved tension captured by the UnRule. We have been riding great leaps in computer power and an explosion of data of all kinds. We have only just begun to explore this new world, which has amazing possibilities and profound challenges.

The UnRules ends with that curve of exponential growth in alphas bending toward the sky. In WorldQuant we have built a company uniquely suited to this dawning age of broad exponential growth. The UnRules is not a long book, but I hope it conveys a sense of the ceaseless searching and testing and experimentation that occur at a firm like WorldQuant. In fact, this book is about beginnings rather than endings. I'm still not a believer in using too many words, but there will be more to say as we explore this new world in more profound ways.

Many books have deep roots. The UnRules goes back to my childhood, listening to my parents practice their music every day in our apartment in Minsk. Authors often thank their parents; none of us would be here without them. But mine embodied many of the virtues that found their way into my rules: hard work, persistence, discipline, goal‐setting, the willingness to take a risk to reach a valuable end, all bound together by love. And without Millennium Management's Izzy Englander, WorldQuant would not exist. He has been my boss, my mentor, and my friend for many years.

Parts of this book were first composed in an internal publication for the WorldQuant community in 2013. Wendy Goldman Rohm, my literary agent, was instrumental in conceptualizing aspects of the book and finding a publisher. Weill Cornell Medicine's Dr. Christopher Mason, the subject of Chapter 9, has entertained and enlightened me in conversation for a number of years, and kindly made sure I got my biology right. Several WorldQuant colleagues read parts or all of this book in draft, offering comments and suggestions, pointing out errors, refreshing memories. They include Scott Bender, Jeffrey Blomberg, Anuraag Gutgutia, Richard Hu, Geoffrey Lauprete, Nitish Maini, and Paradorn Pasuthip. And ably overseeing and managing the editorial process was WorldQuant's global head of content, Michael Peltz. Finally, I'd like to acknowledge all my many colleagues at WorldQuant over the years. This book, and our success, would not be possible without your faith and support.

Igor Tulchinsky

December 2017

CHAPTER 1Quake

“Take aggressive risks, but manage losses.”

On the morning of August 6, 2007, a Monday, I arrived early at WorldQuant's office in Old Greenwich, Connecticut. I had a lot on my mind: I was in the middle of moving, my head filled with the logistical details of movers, schedules, and the kids. By 10 a.m., however, I knew something was wrong. We had been hit, seemingly out of nowhere, by a wave of losses on our statistical‐arbitrage trades – a strategy, common to a hedge fund firm like WorldQuant, that takes advantage of pricing differentials between related financial securities.

As the hours ticked by, anxiety quietly gripped the office. Because our trading is automated, the atmosphere at a quantitative investment management firm like WorldQuant resembles a library far more than it does a frantic trading floor. Nobody's screaming or rushing around. But that Monday you could feel the tension. There was little laughter, and the portfolio managers, clearly nervous, drifted in to discuss their exposures. The next day it got worse.

WorldQuant had been in existence for only six months, although I had been engaged in quantitative trading, which involves using sophisticated math and large amounts of data to identify trading opportunities, since 1995. At WorldQuant we had poured resources into developing about a hundred predictive algorithms we call alphas: mathematical expressions and computer source code that we rigorously back‐test before putting them into production in live investment strategies. All that effort went into ensuring that we wouldn't take a hit like the one we were suffering. We knew that individual alphas regularly weaken or fail, and we were no strangers to drawdowns – we experienced significant declines roughly once a year back then. But our alphas were not supposed to fail collectively. This was bad.

You know what they say: When the CEO moves into a new house, it's a signal to sell. What we didn't know immediately was that similar losses were hitting our competitors at other quant firms. Renaissance Technologies, D.E. Shaw, AQR, and Highbridge Capital Management all saw their finely honed strategies take a sudden nosedive. Goldman Sachs, which at the time had one of the largest quant books – $165 billion – eventually lost more than 30%. Just like us, our rivals must have been struggling to figure out what had happened and why it seemed to be happening just to quant firms.

There had been some ominous signs in the surrounding financial world. For much of the summer, fallout from the unfolding subprime mortgage crisis had been sending shock waves through the markets. Bear Stearns was forced to close two mortgage‐backed credit funds, and there were signs that European banks were growing wary of lending to one another. But our investment strategies were designed to be market neutral – that is, uncorrelated with the broader market. Those subprime issues, in theory, should not have affected the quantitative strategies we employed at WorldQuant. But then, nearly every quant shop probably thought the same way.

Quant firms are only a slice of the hedge fund world, which in turn is only part of the investing universe. Though firms like WorldQuant were hit hard on August 6, 2007, there were no signs of a broader collapse. The next day the Federal Reserve decided to leave interest rates unchanged. Stocks fell after the announcement, then recovered; that week the S&P 500 edged down only very slightly.

As we tried to figure out what had happened, all we really knew was that our relative‐value and statistical‐arbitrage alphas were not working, as if their plugs had been pulled. We suspected that someone out there had taken a hit and was liquidating, setting off a chain reaction of selling, but we lacked the time, the distance, and the data to comprehend fully what was going on. We watched nervously as the problem spread from the U.S. to Japan.

Over my trading career I'd learned a number of lessons that had served me well: Don't get emotional about your trades. React instantly to bad news. If it's scary, run. Take aggressive risks, but manage losses. Back in August 1998, when I was just building my trading portfolio, the Russian government suddenly devalued the ruble and defaulted on its debt. In the resulting violent drawdown, I saw my entire year's gains evaporate in a few days. A month after that, hedge fund firm Long‐Term Capital Management needed a bailout by major banks to avoid causing damage to the American financial system. Now, almost nine years to the day later, that chaotic time was on my mind.

The problem of looking ahead, of course, is that you can't know how big or how long the declines will be. After the first losses on Monday, I made the decision to start liquidating the entire portfolio on Tuesday, giving up all the year‐to‐date profits. Some of this was my memory of the Russian default, when I held on too long, and some was intuition – observing the fear in people's eyes. Liquidating was difficult to swallow, but on Wednesday the carnage deepened, and we felt lucky to be out of it. On Thursday I came into the office early and made a decision to jump back in with 50% of our capital. I was aware that the market could sweep lower, but once again I was relying on intuition – not just on instinct, but on instinct shaped by experience.

In fact, the markets righted themselves as suddenly as they had declined. Just like that, most of the participants were making money again, though we took a few months to get back to 100% invested. We ended up having a pretty solid year. But those who hesitated to sell, had trouble liquidating, or sold into the recovery doubled their pain.

That August 2007 episode became known as “the quant quake,” and it contained a number of lessons: There are risks that you've never thought about, and there are uncertainties. Sometimes you have to act quickly with too few data points. At WorldQuant we may practice quantitative trading, but we also know when to rely on intuition born of experience.

The firm went on to generate stable returns again, and as we accumulated the alphas that we use to build strategies, we experienced fewer significant drawdowns. In the industry the quant quake triggered a rethinking of investment models and a considerable amount of debate. Were too many quantitative hedge funds chasing the same strategies and eliminating the profits? What did happen in early August 2007?

To this day the evidence remains circumstantial and no one really knows for sure what set off the quake. But in the subsequent years, we've developed a better idea thanks to academic research. A month or so after the quake, two finance academics, MIT's Andrew Lo and Amir Khandani, tried to unravel what had happened by building quant portfolios and simulating the episode – in a sense, running the history backward. They concluded that somewhere in the markets a large player – Lo and Khandani thought it was a bank, but Bob Litterman, who ran Goldman's quant fund at the time, later argued it was a multistrategy hedge fund – may have taken a hit and quickly sold a large relative‐value position to respond to credit‐related margin calls or to take risk‐reduction measures. Given what was going on at the time, there may have been a link to the growing subprime mortgage problem. Liquidating positions in turn put pressure on quant firms with similar positions heavily invested in equities, made worse by leverage, which magnifies gains in rising markets and losses in falling ones.

Then a contagion effect developed, with the stress in one part of the market spreading to others. Prices fell, and the more they fell, the worse it got. The fact that the quant quake seemed to target relative‐value trades may have been a coincidence, but it did suggest that unrelated markets had inadvertently grown more correlated, creating a so‐called crowded trade without realizing it, and raising the risk for everyone.

We would see far broader and more dangerous correlations emerge when the global financial crisis broke upon us all. When Lehman Brothers collapsed in September 2008, WorldQuant had another scare: Lehman was our prime broker in Asia and Europe, and its failure meant we couldn't trade our overseas portfolios for several days. But in this case, at least, we knew what the problem was. We quickly negotiated a new prime brokerage relationship and got back into the market in about a week.

As the world struggled to recover from the financial crisis, WorldQuant continued to perform and grow. Today we believe our greatest growth is still ahead of us. We have seen remarkable increases in people and data, computing power and market experience. In fact, it has become clear to me that we are part of an exponential revolution in quantitative finance.

What does that mean? I believe that nearly all aspects of WorldQuant's business, and perhaps our broader business lives, are undergoing not linear but exponential growth. As a result, goals that seem shocking today will look normal tomorrow and useless the day after. Exponential thinking requires audacity, not complacency. It means not believing in limits, which are temporary and meant to be broken. It calls for risk‐taking as a way of life. In exponential thinking the terrain ahead is always unknown. In unknown terrain there are always bumps; it's a world of turbulence and risk. And the rewards are growing exponentially for those who can digest all this information.

When WorldQuant launched, in 2007, we had 37 employees. Today we employ more than 600 researchers, portfolio managers, technologists, and support staff in 25‐plus offices around the world, including over 125 Ph.D.s. Though the number of alphas at our command seemed large in 2007 – and it was, relatively speaking – it has since exploded. We now have more than 10 million alphas archived in the WorldQuant databases, and over the short term our goal is 100 million in the next few years and 1 billion in five to 10 years. That's big, exponential growth, which we expect to happen.

We have built WorldQuant around a handful of core ideas.

Alphas, like ideas, are infinite. Trading can be taught. We believe we hold the future of trading in our hands. We believe that talent is statistically distributed globally but opportunity is not, so we must go out and try to match talent to opportunity. The competitive demands of the market drive us to reach out and continually seek a diversity of opinion – and of ideas, which produce alphas. That's one of the lessons of the quant quake: Don't get sucked into a crowded trade. Think differently.

This means three things. First, WorldQuant is, in part, a technology company that must operate globally to tap talent. Second, WorldQuant is a global alpha factory, whose output is an ever‐growing stream of diverse investment ideas. Last, WorldQuant must shape itself by exponential thinking – by thinking big. Our view is that with great success comes great responsibility. And some of that sense of responsibility extends to educational efforts, particularly in quantitative fields.

Among the most important responsibilities is translating these core beliefs into concrete actions, finding ways to use WorldQuant's insights and resources to provide people around the world with opportunities to develop and demonstrate their talents.

In 2014 we launched the WorldQuant Challenge, inviting participants to build high‐quality alphas. It's part competition, part learning opportunity – contestants use and experiment with our proprietary simulation and back‐testing software, WebSim. Just as impressive as the alphas we've seen generated have been the locations from which they were generated. We've had participants hailing from the eastern coast of India to rural China, reinforcing the fact that a few major cities, or even a few countries, don't have a monopoly on talent or great investment ideas.

In 2009 we started the WorldQuant Foundation, which furthers charitable initiatives, including making high‐quality education more accessible worldwide, through targeted donations to organizations and helping students continue their journey in education. To date, we've offered scholarships to talented individuals who have graduated from esteemed universities in China, the Middle East, and the U.S.