By Irwin Speizer
Photographs by David Deal
Jaffray Woodriff dislikes interruption. He spends most of his time cloistered in a windowless, soundproof office, surrounded by more than 500 books on various business and technology topics, hunched over a computer terminal that has no e-mail access. If he takes two outside phone calls, it’s an unusually hectic day.
As chairman and chief executive of Quantitative Investment Management, Woodriff runs what has become one of the largest commodity trading advisers in the world, with investors scattered across the country and around the globe. Yet he almost never travels to meet his clients or prospective investors, sending out his managers instead or having clients and others make the trek to his headquarters in Charlottesville, Va.
Woodriff prefers monastic devotion to the computer models that are at the heart of his investment strategy—models that he created and that he continues to develop and tweak day after day. “I would be more monastic if I could,” he says.
Those who know him invariably describe the 40-year-old as “quirky” but are quick to add that his skill at running a quantitative managed futures fund is impressive. QIM’s flagship fund, the Global Program, has about $5 billion in assets and an annualized net return of 18.53% from inception on December 1, 2001, through November, 2009, with a return of 16.5% in 2009 through November. The Global Program has a Sharpe Ratio of 1.34 and annualized volatility of 11.6%, and it boasts low correlation with other managed futures funds.
QIM’s growth has been rapid, aided by the firm’s welcoming attitude toward managed accounts, which helps offset the severe, black-box nature of the strategy. About 70% of the firm’s investor money is in managed accounts. Assets in the Global Program increased 10-fold in four years, soaring from a bit more than $500 million in 2006 to $5 billion in 2009. Growth at that clip can trip up managers, and investors have expressed concerns. But QIM has so far avoided the pitfalls through a combination of changes it made in its execution and the overall growth in futures trading, which has provided more room for firms like QIM to maneuver. While 2009 proved to be a tough year for many CTAs, QIM continued to produce steady profits.
“Just look at their returns,” says Rick Boerke, head of Prudential Bache Commodities Midwestern futures and options business in Chicago, who has known and followed Woodriff for years and has invested in QIM. “Correlate that to their asset growth, and you see virtually no degradation in their return stream.”
QIM has been adding new funds to its mix over the past few years, as money has poured in, branching out from its core strategy of short-term futures trading to longer-term futures trading and an equity strategy. Those new strategies now include more than $500 million in additional assets. The newer programs also target 12% volatility and, in addition, offer more volatile, three times levered versions.
At the heart of Woodriff’s method is a complex series of predictive models of his own design that he continues to adapt and modify. The programs analyze historical performance data of futures markets to produce a steady stream of predictions about market direction that he uses to buy and sell a variety of futures around the globe, 24 hours a day. Run out of his Virginia headquarters, the entire procedure is computerized, so the job of his traders is mostly to monitor the system. The Global Program uses a short-term trading style that typically holds positions for an average of seven and a half days—a much shorter period than many CTAs that look for longer trends and often hold positions for several months. While his firm includes 29 employees, only one produces the essential computer models that run the operation: Woodriff himself.
Exactly what takes place inside his models is somewhat mysterious, despite Woodriff’s eagerness to offer explanations. Boerke at Prudential Bache is one of several futures commission brokers that do business with QIM. Boerke says a number of his clients have made the trek to Charlottesville to meet with Woodriff.
“I have had clients visit QIM to try to understand how its models work, but few have ever come back and said they had a thorough understanding of their methodologies—and these are very smart people,” Boerke says. Asked to provide a concise description of what he does, Woodriff responds: “What I have done is to systemize a very tight feedback loop within the realm of prediction so that I can create real understanding of what actually works to identify and utilize the predictive edge that emerges from the collective behavior of participants with very partial information.”
In more direct language, Woodriff uses a statistical technique called the ensemble method, which is a way of mining data to produce something akin to the wisdom of crowds. A bundle of computer models, each searching for patterns in different ways, are linked together to produce a consensus statistical prediction—a sort of prediction by algorithmic committee. Scientists use the method to help predict ozone levels, for example. Woodriff uses it to help predict where futures markets are headed over a 24-hour period. His predictions are derived from four basic bits of historical pricing information: the open, close, high and low of specific markets.
Rishi Narang, whose Telesis Capital is a longtime investor in QIM, says other fund managers use similar methods and techniques. “The core idea is not so magical,” Narang says. “It is how he puts it together. Getting the program correct is very challenging.”
QIM uses about 5,000 model variations, each one looking at a slightly different short-term cycle. Despite all the complex mathematical computations, the system’s predictions are often wrong. Michael Geismar, QIM’s president, says the firm’s models produce profitable trades at a rate of more than 55%, which he notes is far better than those of many competing CTAs, particularly those that look for longer-term trends and may hold positions for months. Some of those funds may consider 40% a good win rate; such firms make up for the lower accuracy by aggressively boosting bets on predictions that pan out and backing out of ones that lose.
“The question we get a lot is, how is one person able to do this?” Geismar says. “Competitors have teams of researchers and PhDs, yet we outperform them.”
Woodriff chalks up his success to a combination of skills, hard work and a contrarian streak that allows him to view problems differently than others. He points out that he has no formal training in the key computer, mathematical and statistical skills needed to do what he does. He is a self-taught quant—and proud of it.
Despite his love of solitude, Woodriff is an engaging and affable conversationalist who delights in ticking off his own shortcomings as someone destined to create and run predictive computer models that guide investment in futures markets around the world. A middling student (he graduated in 1991 from the University of Virginia with a bachelor’s degree in business and a 3.1 grade point average), he regrets that he didn’t apply himself even less to his classroom studies.
“Frankly, I wish my GPA had been lower, because that would have meant that my creatively skeptical contrarian streak would have been more advanced at that point,” Woodriff says. When he started spending time after graduation hunched over a computer terminal in his parents’ home trying to write programs that would predict movements in futures markets, not even his own mother believed he could eke out a living from his labors. She encouraged him to get a real job.
Woodriff grew up on a Virginia farm run by his father. The family—Woodriff has one sister—lived on 500 acres not far from Charlottesville, and his father farmed another 1,000 acres nearby, raising corn, soybeans and cattle. In high school, Woodriff excelled in math and did well in math competitions. But he says he was never a good student in the traditional way, rarely grasping ideas and concepts on a regular classroom schedule. “I wasn’t really a student that you taught how to do something,” Woodriff says. “I was more, gosh, I don’t understand this, and then a light bulb would go off, and I would know how to do it perfectly. In calculus, I didn’t get it, and then one day I got it, and I went from being the worst student to the best student.”
Woodriff began dabbling in the markets while waiting to start classes at the University of Virginia in Charlottesville in 1987. He used a few stocks that his grandfather had given him as a stake and discovered the beauty of options. “I began trading options like a gambler,” Woodriff says. In the fall of 1989, Paul Tudor Jones, the billionaire hedge fund manager who heads Tudor Investment Corp. and also happened to be a University of Virginia alumnus, delivered a talk at the business school. Woodriff was aware enough of who he was to attend and recalls shaking Jones’ hand. He has not run into Jones since, but the experience piqued Woodriff’s interest.
“I wanted to work with raw numbers,” Woodriff says. “The fact that a lot of money was being made on the futures side with just price data was very compelling.”
Woodriff began spending his spare hours reading up on technical analysis and after graduating in 1991, he moved back home and continued working on his ideas, creating spread sheets crammed with daily futures price data. He decided he needed more computer power, so he went back to campus looking for a computer lab where he could camp out. “My mom and sister were saying, ‘what is wrong with you? You are crazy,’” Woodriff says. “But my dad was like, ‘you seem to be passionate about this; go for it, kid.’”
Woodriff’s modeling advanced, and he hit upon the idea of having several models work together to learn and reach a predictive conclusion. His breakthrough came during a double all-nighter at the computer lab. “I was trying it out on 20 or 30 computers at once,” he recalls. The approach he developed is still the core of QIM’s Statistical Prediction Integrating Robust Ensembles method that Woodriff continues to develop.
Through an aunt and uncle, Woodriff met an older UVA alum, Bob Jordan, who was an asset manager in Roanoke, Va. Together they formed Blue Ridge Trading in 1991, with Jordan raising money and managing the business and Woodriff developing his models and trading futures. Looking for a cheap place to live and work, Woodriff spotted an ad for a room in a student apartment near campus.
He made his way to the apartment in an ice storm and knocked on the door. His future president answered. Michael Geismar was an undergraduate at the time, and he eyed Woodriff suspiciously. “We had a lot in common,” says Geismar. “Everything he told me made a lot of sense.”
Woodriff moved in, set up his computer in the living room and spent his days fiddling with his models and trading client money for Blue Ridge. He taught Geismar how to trade, then hired him after Geismar graduated in January 1994. Woodriff says Blue Ridge reached $2.5 million in assets and had a return rate of more than 20% trading futures in stock indices, currencies, and several other markets.
The arrangement was short-lived. In July 1994, Woodriff split off from Blue Ridge and formed his own CTA, Woodriff Trading. There was no money to pay Geismar, who left for a job as an actuarial accountant at Towers Perrin in Alexandria, Va. Over the next few years, Geismar worked at Coopers & Lybrand, William H. Mercer and Capital One Financial.
Woodriff continued to work solo at his new company, moving into a house in Charlottesville that doubled as his office. He reached a peak of nearly $3 million in assets under management by 1997. But a hefty drawdown in 1997 was enough to make him realize how tenuous his position was. In addition, he found himself spending so much time on client relations and business management that he was left with little time to work on his real love, which was researching and programming his models.
“I looked at my year, in terms of what I spent time on, and realized I did virtually no research. I hadn’t changed the models. I was doing things I wasn’t nearly as good at. I thought that was stupid. I thought if I were a proprietary trader at a bank in New York, all of the other work I was doing would be delegated. Plus, I was going to get to go to New York and get that perspective. The plan was to spend two or three years there.”
New York was initially unwelcoming, as Woodriff amassed a pile of rejections from banks, asset managers and hedge funds, including D.E. Shaw, which told him that despite his wonkiness he wasn’t their type. But Woodriff persisted—and finally landed a spot in 1998 with Société Générale, which was expanding a proprietary desk and was looking for traders with diverse skills and backgrounds. Woodriff says he got hired based on his ability to generate returns with his CTA strategy.
“I couldn’t believe my luck,” Woodriff says. “I was hired on as a vice president and then promoted to director.” Woodriff left after two years, intending to start a new hedge fund with one of his bosses at Société Générale that would be based around his futures system. Woodriff called Geismar and asked if he would like to join up, too. Geismar agreed, quit his job at Capital One and moved into an apartment in New York with Woodriff. While waiting for the new hedge fund to gather enough clients to start operating, Woodriff began trading equities in a personal account that held about $300,000, with Geismar helping out. But the clients never came, and the new hedge fund never launched.
Both shared a love of Charlottesville, so they left New York in 2001, moved back south, and set up a new trading operation called DHR. Although they started out by continuing to trade individual U.S. equities, the main thrust quickly turned to trading futures based on Woodriff’s system. They launched the Quantitative Global Program in December 2001, trading their own money. In 2002 the program returned 13.6% and held $1.68 million. The fund returned 20.13% from January to September 2003, when Jaffray and Geismar decided to open it to outside investors. In October 2003, they set up a new company, Quantitative Investment Management, bringing in as a third co-founder Greyson Williams, another UVA alum who had done work for them as a private consultant in financial analytics. Williams is QIM’s chief operating officer today. DHR was closed in 2005.
QIM chose to charge no management fee but a 30% performance fee, and it encouraged investment in separately managed accounts, which turned out to be a big plus, particularly with investors from Europe. From October to December 2003, the fund returned another 10.46%, and its assets under management grew to $8.75 million.
The Global Program has continued posting positive returns ever since, and the steady returns helped attract a flood of new investment, rising from $13.75 million in assets in 2004 to more than $5 billion as of November 2009.
The rapid growth encountered some bumps along the way. The Global Program jumped from $143.6 million in 2005 to $527.7 million in 2006, and Woodriff discovered he had become such a big player that his trades began influencing the markets. Once his models pointed to an opportunity, the Global Program would quickly place a big bet over a period of a few minutes, and the size of the trades became big enough to cause prices to react and throw off the expected result. The Global Program continued to make money but had its worst result since its inception in 2006, up 5.23%.
“We had become the elephant in the trading room,” Geismar says. “We were pushing markets too quickly. If we were buying gold for 10 minutes, we could see the price of gold move, and it was all us. We always knew there was slippage. The problem was we didn’t think we would affect the market as much as we were.”
The trading style was retooled. Trades that had once been executed over 10 minutes were now spread over several hours. A new rating system was devised for the signals generated by Woodriff’s models that sized trades based on the strength of the signal. If a signal continued to build in strength over the next few days, the trade size could be increased. The Global Program rebounded with its best year ever in 2007, up 28.4%. Assets soared to $2.74 billion.
“Prior to making the change, we were hurt by being too aggressive,” Woodriff says. “We are so much bigger now, but we have so much less of a problem because we made so many changes to become much more passive in how we put our positions on and take them off. It is amazing that there was so much room to become so much less aggressive.”
The market meltdown of 2008 hurt many hedge fund strategies, but the Global Program continued to do well, posting an 11.94% return that year. But investors took advantage of QIM’s liberal withdrawal policy, which has no lockups or gates, to cash out during the tight times. QIM’s assets went from $3.7 billion to a low of $2.6 billion in the fourth quarter of 2008. But the drain almost immediately reversed into a flood of new money in 2009, pushing the Global Program to $5 billion.
QIM has also been adding new funds. The equity trading strategy that Woodriff and Geismar had practiced back in New York became the Tactical Program, a short-to-medium-term market neutral equity fund that was launched in November 2007. It has so far had positive returns, and it counted $436 million in assets as of October. But a managed futures fund that maintains a much longer, one-year holding period has not fared as well. The Ultra Fund, launched in September 2007, was down 6.28% in 2008 and lost another 4.49% through November. Woodriff has been working to improve the performance of that strategy. QIM also now offers more highly leveraged versions of its funds.
The heart of the QIM system remains a work in progress that is not immune to significant short-term swings. On Friday, December 4, the Global Program got caught on the wrong side of a dip in long bonds and foreign currencies. Long bets in those issues cost the Global Program 3.7% in one day. But the next week started off much better, and by Wednesday the Global Program had made up about half of the loss.
QIM figures it still has plenty of room to grow and is accepting new investment in all of its funds. To help accommodate growth, the Global Program has been adding new markets around the world, with nine launched in 2009 and plans to enter another nine. Among the newest trading venues are four interest-rate futures markets, some agricultural commodities, and one new currency, the Swiss franc. The Global Program currently allocates about a three-quarters of its futures trading to stock indices, interest rates and currencies, with the rest spread among oil and gas futures, metals and agricultural commodities.
Geismar says QIM intends to continue adding assets without adding extra manpower—while relying on Woodriff as the brains of its systematic trading methods.
Woodriff says he is happy to be doing what he loves and delegating all the rest to Geismar, Williams and his other managers and staff. While he plans to continue building QIM and fiddling with its models, he says he has already exceeded his wildest dreams as an asset manager. Indeed, he would love to have a second act as a software developer, transforming the models he has built for investment management into general-use predictive models that could be deployed for other businesses or for science. He would enjoy nothing more than to be allowed more time for solitary communion with his computers.
“I would like to be completely alone for half of my week,” Woodriff says. “Programming with full focus, with no chance of the phones ringing.”
