Having the Edge on the Market
by Ken Kurson
Back to The Reading Room
Ed Thorp looks you in the eye and tells you he intends to live forever. And because he invented the first system ever proved to beat blackjack, because he helped build the first wearable computer, because he's a former math and statistics professor who happens to run one of the most successful hedge funds in the country--a 47.14 percent return in 1998 and better than 59 percent in the 12 months ended this April, with virtually no risk!--you figure if he lives that long he's going to make a lot more money than he already has.
In 1962, Ed Thorp published Beat the Dealer, which proved mathematically what had only been suspected: that by keeping track of the cards that have been dealt, a blackjack player can produce a positive expected return--anywhere from 0.25 percent to 5 percent, depending on the number of decks and how deeply into them the dealer penetrates before shuffling. It's tempting to think of a successful card counter as a very talented gambler, but a real card counter isn't gambling at all; eventually a positive result will--not may, but will--be produced. A card counter removes the risk from the game and grinds out a profit. And a grind it is: Table-stakes limits and intimidating pit bosses render card counting better suited to those who actually enjoy the danger and secondhand smoke associated with casinos. Thorp made a fortune testing and refining his strategy, but his pens-in-the-pocket style never quite fit in with the glitz of the Vegas strip. More important, Thorp had begun to apply his thinking to a much bigger game of chance: the stock market.
"The overlap of interest between gambling and the stock market is very high. It's an amazing phenomenon," says Thorp. "But there are so many similarities and so much one can teach you about the other. Actually, gambling can teach you more about the stock market than the other way around. Gambling provides an analytically simpler world, and you can see principles and test theories."
To test those theories, Thorp founded Convertible Hedge Associates in 1969 (renamed Princeton-Newport Partners in 1974). He and his band of sequestered statisticians pored over charts and patterns alongside room-sized computers in Newport Beach, California, while an office in Princeton, New Jersey, mostly handled the firm's administration. The hedge fund's bread and butter were arbitrage deals that resembled Thorp's blackjack strategies--no risk, high return.
In 1974, American Motors convertible bonds were selling for the same price as the stock underlying them and paying 5 percent in the meantime (actually, more like 8.3 percent, since the bond was selling below face value). Thorp, playing both ends against the middle, shorted the stock while buying the bonds. "Situations that simple and clear are few and far between, but we made a large part of our living off scenarios just like that," says Thorp now, wistful for the days before information was sliced and diced by thousands and distributed to millions with the click of a mouse. In 1983, Thorp executed what was then the New York Stock Exchange's all-time largest dollar-value block trade. With Ma Bell about to dismantle itself, investors were panting for shares of the new Baby Bells. Thorp bought $330 million of the "old" AT&T and sold short $332.5 million of the new "when-issued" stock, which was bundled as ten shares of the new AT&T plus one each of the seven Baby Bells. That sliver of arbitrage--a gap of just 0.76 percent--enriched Thorp's investors by $2.5 million. With no risk.
Unfortunately, Thorp's talent at analysis and trading is matched by his lack of interest in personnel and marketing matters. The Princeton office "interfaced with the Street and did taxes and accounting," in Thorp's words--and some of that "interfacing" involved Michael Milken's circle at Drexel Burnham Lambert. Thorp, never accused of misdeed, realized he was better suited to running a small shop, one where the half-dozen employees could fit into a small but elegant suite overlooking the yachts of Newport Beach. So in 1989 he created Edward O. Thorp & Associates.
These days, Thorp trades about 4.5 million shares in as many as 3,000 transactions a day. Looking for "short-term mispricings," he shorts the overpriced and buys the underpriced. If he's right just better than 50 percent of the time, he'll make money. "Each trade has about a one-half-a-percent edge when it's put on," Thorp claims, "but we end up losing about a quarter of a percent to transaction costs--spread, commissions, SEC taxes--so we end up with a net edge of about a quarter of a percent on each trade. If we can do a tenth of a percent a day, we're going to have a real good year."
The goal is simple. "What we try to do is make 15 percent or more, annualized, for investors year after year, with low risk," says Thorp, who, because he holds hundreds of positions, will not be undone by a single mistake. Since its start in August 1994, Thorp's Ridgeline Partners fund has returned about an annualized 30 percent before fees (about 23 percent after fees). He says the fund has had positive returns in 82 percent of its months in operation. What makes this extraordinary performance all the more impressive during the market's long rise is that a typical day's return has only a slight positive association with the direction of the market: The fund's beta is just 0.1. (A beta of 1.0 means the performance of a stock or a fund essentially matches that of the Standard & Poor's 500 Stock Index; index funds that track the S&P usually have betas between 0.98 and 1.02. Betas above or below 1 indicate, respectively, more or less volatility.) In short, Thorp is beating the market while taking on less risk than the market itself offers investors.
That someone making 3,000 trades a day might be considered risk-averse isn't so strange given the care with which Thorp trades. Unlike most so-called hedge funds, which really don't hedge at all, Thorp's fund is truly market neutral: With equal dollar amounts invested long and short, his fund is far less vulnerable than most to the gyrations of the market. He keeps his fund's assets at about $300 million by closing it often and distributing profits regularly. And while he enjoys the usual hedge-fund-manager compensation (a 1 percent management fee and 20 percent of profits), Thorp humanely charges no performance fee for years in which the fund gains less than 6 percent. With a waiting list of over $100 million, Thorp operates knowing that the risks of being too big are as great as those of being too small.
Thorp doesn't analyze specific stocks--"We don't read brokerage research"--and laughs at the idea that anyone could know enough about all the publicly traded stocks to suss out the ones with long-term appeal. "We look for an unusual change in the price of a stock," he says. "There are certain company events that we track that are significant. We've got a collection of patterns of behavior, both technical and fundamental--I don't want to say what the things are, but we have this huge pile of data. The machine looks at the data, finds patterns in it, and selects whether to buy or sell short. It basically ranks the stocks from best to worst."
Thorp pauses, wondering if even this general description has revealed too much. The mathematician in him, however, can't help but refine the thought. "Ranking is too simple. We forecast a return for the near future on each stock, and the ones that have the highest return we buy. The ones that have the lowest return we sell short. And we make sure that we sell short about the same dollar amount as we buy. Now, that doesn't necessarily mean it's market neutral, because we could be either long much riskier stocks or short much riskier stocks. But it turns out, because the portfolio is so diverse--250 stocks on either side--it turns out it's very close to being market neutral. The actual unlevered data on the computer is about 0.06--so close to zero that it doesn't make a whole lot of difference. It's slightly positive, which tends to be good, because the market tends to go up."
Thorp's interests vary widely, and he's a voracious consumer of all sorts of information. He can speak as easily about Girolamo Cardano's 16th-century treatise on games of chance as he can about the time his drink was drugged at a casino. He's an accomplished amateur photographer and astronomer. And he completes one or two marathons a year. But Thorp's main passion continues to be the intellectual challenge of systems and a consuming need to find patterns and predictability where others see randomness.
Drawing another analogy between gambling and investing, Thorp makes clear that a situation that others see as a wash can at times yield a return of the kind that gets you written up in magazines. "Say a blackjack player is dealt a ten and a six, while the dealer's showing a ten. You can calculate that situation, and anyone who's played any cards knows you're 'supposed' to hit. But what if your 16 is comprised of two fours and four twos? In a deck that's ten rich, it's a definite stand." Just as the shares of AMC were valued differently from the right to buy those same shares, not every 16 can be treated equally at the blackjack table.
To explain why these anomalies continue to go unnoticed in an information-efficient economy, Thorp excoriates slot machines and the people who mechanically pour change into them. "They're the most moronic devices ever, one of the stupidest activities of humankind. People play negative-expectation games," says Thorp. "That's something I'm not willing to do."
Thorp pauses, to make sure he hasn't overlooked a behavior that would make his final thought untrue, then smiles at the reassurance that he has remained true to his logician's heart, his statistician's brain, his probability-calculating instinct: "I've never even bought a lottery ticket."
Ken Kurson, a former staff writer for Worth, is the financial columnist for Esquire.
Back to The Reading Room