Man vs. Machine on Wall Street: How Computers Beat the Market

Wall Street, meet your post-human future. Uber-"quant" Cliff Asness bets that his high-speed computers and trading models can churn billions of dollars in profits in booms and busts alike. But can artificial intelligence really out-smart the market? 


With the winter's second blizzard raging outside, Cliff Asness sat in his relatively modest office in Greenwich, Connecticut, surrounded by three of his partners, his PR guru, an impressive collection of unread books, and a sea of foot-tall hard-plastic replicas of Spiderman, the Incredible Hulk, and friends. "Let me be technical," he said. "It all sucked."

Asness--intense, bald, and bearded, with a $500 million fortune and a doctorate in finance--was reflecting on the dark days of 2008, when capitalism seemed to be imploding, when Bear Stearns and Lehman Brothers had collapsed and the government had hastily arranged bailouts of Merrill Lynch, Morgan Stanley, Goldman Sachs, and AIG, among others.

His own business, Applied Quantitative Research--one of the world's leading quantitative-investment, or "quant," funds--had also suffered painfully. The money his team managed fell to $17.2 billion in March 2009, from a peak of $39.1 billion in September 2007, as clients headed for the exits with what was left of their cash.

Such losses can be fatal for fund managers like AQR, since sophisticated investors pay them big fees for exceptional performance and, understandably, have little patience for anything less. As AQR's founders felt the tremors from Wall Street rippling through their offices, Asness said, "we worried about the stability of the financial sector, the stability of the economy, and the stability of society." To Bloomberg Markets magazine, last fall, he was even more explicit: "I heard the Valkyries circling. I saw the Grim Reaper at my door."

Yet they survived. And AQR--which makes its fortune, like other quants, by using high-speed computers and financial models of extraordinary complexity--has made a stupendous recovery in the past two years. At the end of 2010, AQR had $33 billion in assets under management. Its funds' performance was up nearly 20 percent last year, after being up 38 percent in 2009.

This is all the more striking because many analysts believe the quants helped cause, or at least exacerbated, the meltdown by giving traders a false sense of security. The risk-control models these firms pioneered encouraged Wall Street to take on excessive leverage. Their trading strategies, which deliver excellent returns in normal times, functioned poorly in the irrationality of a financial panic, and reinforced a frenzy of selling. Although predictions of the death of AQR and its ilk, by the writer and investor Nassim Taleb, among others, turned out to have been greatly exaggerated, worries linger, even as some high-profile quants have surged back. Taleb and the other critics think their overreliance on computers gives quants excessive confidence and blinds them to the possibility of seemingly rare economic catastrophes--which seem to be not so rare these days. (This was the theme of Taleb's best-selling book, The Black Swan, which examined the effect of the "highly 
improbable" on markets, and on life.) As Exhibit A, they point to the extraordinary events of May 6, 2010, when the Dow dropped by nearly 1,000 points in a few minutes after an algorithmic program executed by the investment firm Waddell & Reed, in Kansas, triggered a terrifying blitz of automated buying and selling by other financial computers. The market quickly recovered, but many worry that the episode was a preview of greater turbulence ahead as machines gain control of more and more trading.

Scott Patterson, a former Wall Street Journal reporter and the author of the 2010 book The Quants, told me he can envision a world, not too far away, in which artificial intelligence could vanquish human trading altogether, just as it has Garry Kasparov on the chessboard. "I'm not totally against quants at all, because I think they are a very powerful way of investing," Patterson said. But, like a number of other critics, he thinks they might encourage a cycle of booms and busts, and possibly intensify the next crisis. "Go to a trading room, it's just guys on computers," he said. "And a lot of times it's not even guys, it's just the computer running the machine. I don't want to demonize it. I think there has to be a happy medium. But I'm personally worried that it can run off the rails."

As much as anyone else, Cliff Asness has shaped and 
embodied this world of automated high finance. And though his experience--from academia to Wall Street to Greenwich--has been marked by recurring crises, and though he admits that no one can predict when the next big one will hit, he's more confident than ever in the power of data and mathematical models, in his hands, to beat the market consistently over the long term. And, once again, the data are telling him he's right.

Next page: Learn what Asness calls "the only free lunch in finance"

Traditional professional investors--Asness calls them "quals"--search zealously for a handful of undervalued companies in which to invest. They interview management teams, evaluate corporate strategies, and analyze demand for products and services. AQR, like other quant firms, relies mostly on a roomful of powerful computers running proprietary models to evaluate reams of publicly available data. AQR never talks to management before investing. Asness and his partners were among the first to build a stock portfolio--and now a very successful business--by using computer models to combine two simple concepts: buying undervalued stocks (a strategy known as value investing) and betting against overvalued ones (which are called "momentum" stocks, referring to the tendency of securities that are rising in price to keep going up for a time, even when they're overvalued). Using a variety of metrics, the AQR models spit out the names of hundreds and hundreds of stocks that are undervalued (which the firm buys and holds) and hundreds more stocks that are over-valued (which they short, or bet will fall).

Asness explained the differences between quants and quals this way: "A qual digs very deeply into potential investments, but he can only do that with so many stocks, so he needs to have a relatively high level of conviction that he is right, since he's going to hold a pretty concentrated portfolio, say 10 or 20 stocks ... A qual needs to be careful about not making mistakes--one bad mistake in a 10-stock portfolio can get ugly!" He continued: "A quant, on the other hand, has the ability to study thousands of stocks at once, and thus can hold much more broadly diversified portfolios. Because quants hold so many stocks, ones that are even slightly misvalued may still make sense ... If you can find 500 stocks to bet on where each has a 51 percent chance of beating the market, then through diversification, the odds of your overall portfolio start to look pretty good."

AQR's September filing with the Securities and Exchange Commission shows that the firm owns stock in roughly 2,000 companies, from ADC Tele-communications to Zymo-genetics. "What a quant tries to do is spread their bets," he said. "You can still be dead wrong as a quant, but you're dead wrong if the average result doesn't hold. You should never be dead wrong because three stocks did the wrong thing." In a television interview recently, Asness said, "Diversification is famously called the only free lunch in finance. Some people ... don't fully believe in that. I do. And when you see a free lunch, the rational thing to do is eat it."

In essence, AQR is betting that markets will revert to the mean, and that investors will act "rationally" once information about individual companies is fully digested. (As an 
example, he said he recently declined Goldman Sachs's invitation to invest in Facebook at a $50 billion valuation, which he considered excessive.)

Quants sometimes speak about having found the Holy Grail of investing, the ability to construct vast portfolios of stocks that are nearly perfectly hedged--bets on value on one side, bets against momentum on the other--and that they believe should be, over the long haul, relatively immune to the market's increasingly violent vicissitudes. Scott Patterson refers to this phenomenon as the belief that they have discovered "The Truth." "Quants will say, 'No, we don't believe anything like that.' But actually when you sit down with them over a couple drinks, they'll start talking like that," he explained. "It's this belief that it's all captured in mathematics and models."

While quant investing remains a relatively small part of the financial world--perhaps $500 billion now, according 
to eVestment Alliance, out of the tens of trillions of dollars invested worldwide--in their effect on professional investors, the quants punch well above their collective weight, because what they're doing is considered so cutting-edge. Firms like D. E. Shaw, Renaissance Technologies, and Barra attract waves of Ph.D.s eager to create models that might lead to the next great investing technology, and to personal fortunes. Other investors can't help but take notice and try to emulate them.

Next page: "Potentially devastating consequences"

Sometimes, though, the quants get too clever for their own good, with potentially devastating effects. Such a moment occurred in the second week of August 2007, when a wave of selling by a group of quant funds using the same trading strategies led to terrible losses, as the firms all tried to sell the same stocks at the same time. As Andrew Lo, a professor at MIT's Sloan School of Management, observed in a September 2007 paper on the event, an "apparent demand for liquidity" that week "caused a fire sale liquidation." Patterson estimated that AQR lost $500 million in a single day, and close to $1 billion in the four-day rout before the markets steadied and started to recover on August 10.

(AQR doesn't dispute this estimate.) "No one really knows how much money was lost over this one-week period," he said. "But I've heard estimates that $100 billion evaporated out of these quant funds. It could be more, it could be less. No one ever really knows."

Things were so bad, in fact, that AQR, which was pursuing an initial public offering, soon dropped the idea. "They knew these strategies worked and they deluded themselves into thinking that they would always work," Patterson said. "And that's a very dangerous way of looking at the world, because it allows you to take more risks. That's why I think the quants sort of represent the entire financial system in a way. They got blinded by science."

Asness thinks this assessment is far too simplistic. He told the New York Post that he blamed the sudden losses not on AQR's computer models but on "a strategy getting too crowded ... and then suffering when too many try to get out the same door" at the same time. He told me he finds the argument that quants are "black boxes" of dangerously opaque trading strategies annoying and wrong. "We don't think of ourselves as 'black box,' " he said. "It is a great irony to us that even though a quant can, if willing, fully describe his investment process, it's often called 'black box,' even as the fundamental investor, who can never accurately describe his process, is not tagged with that label. A friend of ours, who is both a quant and fundamental investor, thinks quant is more accurately called 'glass box.' We think that's pretty accurate."

Asness abhors the idea of increased quant regulation. With some reluctance--given the vitriol with which he typically condemns Washington on his blog--Asness conceded that the government bailouts in September and October 2008 saved AQR by rescuing the firms with which AQR trades, an outcome at odds with his Chicago-school economic training, which champions Milton Friedman, free markets, and the survival of the fittest. The bailouts "saved any levered fund's bacon," he said. (Of the $33 billion that AQR currently manages, $13 billion is in levered funds, which use borrowed money to increase returns on the equity invested.) Nevertheless, he remains unapologetically critical of the bailouts. He thinks the government should have let the banks fail and the chips fall where they would. "Look, if Hank Paulson proposes, 'I'm gonna hand Cliff $50 million,' " he said, "Greedy Cliff could go, 'I kinda like that plan.' But if you ask me, 'Is this good for the country?,' I will publicly tell you, 'No, it's not good for the country.' "

Asness was born in Forest Hills, a ritzy Jewish section of Queens. His father, Barry, was a lawyer and a former Golden Gloves boxer. Early on, the family moved to a tract home in Roslyn Heights, on Long Island. His mother worked as a public-school teacher and later became president of World Health Communications, a pharmaceuticals marketing and education company. Both parents commuted to Manhattan, leaving Cliff and his brother, Brad, to fend for themselves. "Largely, I was latchkey," Cliff explained. (Brad is now AQR's general counsel.) At Herricks High School, Asness was a classic underachiever. His grades were middling. He quit or was fired from so many jobs that in his high-school yearbook, under the category of "Always seen at," Asness was described as "at a new job." He skipped a lot of classes. The turning point came when he scored the highest in the school on the SATs. "That actually had a weird kind of effect on me and I was like, 'Wait, I can do something.' " He was a mediocre calculus student but got a 5, the highest grade, on the AP Calculus exam. "That was kind of classic me," he told me.

Next page: "Nerds were suddenly cool, and everywhere"

He went off to the University of Pennsylvania and entered a dual-degree program in computer engineering and finance. He signed up for the LSATs but switched to the GMATs after his father convinced him that "people who can actually do math and engineering probably shouldn't be lawyers." He built a computer and wrote operating systems in FORTRAN. For extra cash, he worked for a few Wharton professors who wanted him to "code up" and test their theories. To his surprise, the task turned into more than just a job. He became engrossed by the idea of using computers to analyze stocks. He decided to pursue economic and finance theory in a doctoral program at the University of Chicago, under the professors Eugene Fama and Ken French. (Fama is considered a father of the "efficient-market hypothesis"--the increasingly controversial idea that because at any given time the prices of stocks trading in the market reflect all the available public information about them, the market is "efficient" and therefore tough to beat.)

While there, Asness wrote academic papers on subjects like "Changing Equity Risk Premia and Changing Betas Over the Business Cycle and January," and slowly absorbed the principles that would become the bedrock of both his financial strategies and his political discourse. He wrote his thesis on the "puzzling trait" of winning and losing stocks to keep winning or keep losing, even when data suggested they had overshot their true value in one direction or the other. He thus became one of the earliest theorists on momentum investing, which AQR still uses to its advantage by shorting stocks that seem to be on a momentum-fueled bender.

Meanwhile, Asness's best friend from college, Jonathan Beinner, went to Goldman Sachs to work in its asset--management business, known as GSAM, then a backwater at the firm. In 1992, Beinner asked Asness to come work for the summer at GSAM in the fixed-income division, building quant models. "It was very low-cost, low-risk to them," Asness said. While there, he developed inchoate versions of what he still uses at AQR to choose stocks, bonds, and currencies. At the end of the summer, Goldman asked Asness to stick around in the asset-management program for another year. "I was really on the fence between going back to academia versus taking it," he continued. "I was writing my dissertation at night." But the money Goldman offered him--about $100,000 a year--quickly changed the calculus, and he accepted. At GSAM, he specialized in trading mortgage--backed securities. "It was a great baptism," he wrote in a 2007 essay, "as every investing disaster that can befall you happens to some variety of a mortgage--back in reasonably short order." (He apparently learned his lesson: while mortgage-backed securities played a huge role in the market crash in 2008, they played no direct role in AQR's problems at the time.)

After reading a paper he wrote about models and steep bond-yield curves, which Asness wryly called "a page-turner," an executive at PIMCO, the huge asset manager, offered him a job starting a new quant-investing group. When he told his boss at Goldman about the offer, his boss said that GSAM was starting its own quant group and asked Asness to stay and run it. This would be a genuine "present-at-the-creation 
moment," when Asness and his team would combine for the first time the exponentially growing power of computers with the academic theories he had learned from Fama and French, along with his own pioneering thinking on momentum investing. A new long-term-investing strategy had been born, and it quickly proved very successful.

By then Asness had recruited John Liew and Bob Krail--two former classmates at Chicago--to come to Goldman and join him. In 1994, the first assignment for the new quant group was to investigate whether models could help stock pickers evaluate risk in foreign countries, where Goldman's asset-management team had a poor record. Using their proprietary models, they bought stocks in countries whose markets were undervalued and sold stocks in countries whose markets were overvalued. "It turned out that the value-and-momentum strategy for picking countries performed quite well," Asness wrote in the essay. In their youthful enthusiasm, they published their findings in an academic journal, effectively giving the milk away. (They quickly learned not to do that. Publishing current findings would be "a capital offense" at AQR, he said.)

Asness and his team soon replicated that success by creating models for buying bonds and currencies. He persuaded Goldman to start a hedge fund, using partners' money to employ his quant strategies. In short order, the team turned the partners' $10 million grubstake into $100 million. Goldman then opened the fund, named Global Alpha, to outside investors, with the pitch "You get the Goldman secret sauce with this smart team," Asness recalled. The fund was a rocket ship, up 111 percent in 1996 and 42 percent in 1997. By that time, they were managing $7 billion in total.

Despite a promise from Hank Paulson, then Goldman's No. 2, that he would soon become a partner at the firm, Asness started to feel the lure of starting his own hedge fund with Liew and Krail, free of Goldman's bureaucracy and compensation restraints (which likely would have put a lid on his annual income in the $10 million-to-$20 million range, far below what top hedge-fund managers make). The catalyst for their departure from Goldman was probably the single--mindedness of David Kabiller, whose job at the time was to meet with chief investment officers of pension funds and endowments and persuade them to let Goldman invest their money. Kabiller was convinced that he and the Chicago boys could create something special. "We can have more fun," Kabiller told them. "We can be more unfettered."

One of the problems at Goldman was that the firm would not leave the men alone to just manage the $7 billion--their first love--and kept after them to mentor others and to provide quant tools and analyses for the marketing department. "We were a support group as well as an asset-manager group," Asness explained. The final straw came one night when Kabiller was on a blind date with a woman whose father, a big hedge-fund manager, was going to invest in a fund being started by one of their recently decamped Goldman colleagues, who the team thought was a mediocre investor. In the middle of the date, Kabiller called Asness and told him how outrageous this seemed. "We had a feeling like, 'Gee, are we gonna be the guys sitting in IBM and watch this entrepreneurial guy go and leave and become really successful?' " Kabiller said. The departure process lasted a year. "Of agonizing," Asness said.

Finally, in early 1998, the four men left Goldman to start Applied Quantitative Research. By August, AQR had raised a $1 billion hedge fund, thought to be one of the largest initial hedge funds of all time, only slightly smaller than the one raised by the infamous Long-Term Capital Management, which opened in February 1994 with a group of star bond traders from Salomon Brothers and two Nobel laureates. This was a time when the allure of hedge funds of many different stripes--not just quantitative hedge funds--was growing across Wall Street. Physicists and mathematicians were fleeing academic departments to enter finance. Goldman was losing alpha males convinced of their own investing prowess. Hedge funds were cultivating an air of invincibility. Nerds were suddenly cool, and everywhere. (One hedge-fund manager I spoke with recalled working next to a group of quants in his office and thinking to himself, "This is the geek tank. They were behind glass. They had their headsets on all day, just computer programming.") According to Lo, the MIT professor, the number of quant-style hedge funds grew from something like 50, in January 1994, to nearly 700 by the peak in 2007.

As it happened, AQR had started just months before Long-Term Capital blew up (and needed to be rescued), and in the midst of the Internet bubble, when anything related to the Web seemed to double or triple in price overnight. It was a world of irrational momentum, an environment that could not have been worse for Asness's investing style. In 1999, AQR owned a bunch of seemingly undervalued stocks, in businesses like banking and manufacturing, while holding short positions on seemingly overpriced tech stocks. The firm was getting killed, bringing to mind Keynes's famous observation that the market "can stay irrational longer than you can stay solvent."

Within AQR's first 20 months, its $1 billion fund was reduced to $400 million. The firm was near complete collapse, but Asness fought hard to keep it alive. He added to his value investments as the bubble inflated and kept his short positions in place, with the hope that he could capitalize when it popped. He met repeatedly with investors, and argued that he would be proved correct once the hysteria subsided. And indeed, things soon turned around. From 2000 to 2002, in a bear market, AQR "made a ton of money," Asness said, and then for the next few years "made decent money" in a generally bull market. The firm nearly lost it all again in the August 2007 fiasco, and suffered along with everyone else during the financial crisis the following year. But because AQR was now more diversified--with products ranging from mutual funds for small investors to a variety of funds available only to sophisticated institutional investors--the threat to its existence was not nearly what it had been in 1999.

Then, in the past two years, AQR roared back, thanks to both a near-doubling of the Dow from its nadir in March 2009, and a return to the efficacy of the long-term-investing strategies Asness had pioneered since his days in academia. With that success, of course, came a return to confidence for him and his firm.

Next page: The future belongs to hedge funds

Asness thinks Nassim Taleb is right about the increasing frequency of Black Swan events. And quants still face severe threats from market volatility: according to a report from Morgan Stanley, quant funds betting against momentum stocks in Europe took a beating in early January that reminded many of the 2007 crisis, causing as much as a 10 percent overall loss. "In only a few days," the report said, a number of quants "experienced unprecedented losses in seemingly 'normal' market conditions." But Asness thinks the episode was overblown, and says AQR was unaffected by it. And he does not believe that quants, or other more traditional hedge funds, had any role in the catastrophic events of 2008. "The crash was about credit and real estate," he said. (That seems to be the conventional wisdom, at least as reflected by the report of the Financial Crisis Inquiry Commission.) "If you look at our history--even AQR's aggressive hedge funds--we've made money, and we have smoothed the path," he said.

They've also made themselves fabulously rich--beyond anything they could have imagined--and proved to themselves the wisdom of Kabiller's original vision. Asness thinks AQR has as much chance as anyone of predicting and protecting against the next Black Swan. Even so, he and his partners are occasionally given to intense periods of introspection, wondering whether the quantitative approach really still works. "I do have a recurring nightmare about being hacked to death by a pack of rabid black swans," he said. "What do you think that means? Seriously, anyone, quant or not, with a shred of intellectual honesty recognizes that there is some chance their historical success is just luck."

Still, he thinks AQR and the other quants have bounced back, from crisis after crisis in the markets, for one very important reason. "We had to first convince ourselves we were right," he said. And so they have.