Using Cool Math to Find Hot Stocks
For an investing guru, Yale University finance professor Zhiwu Chen had a pretty poor start at playing the stock market. In 1996, Chen poured $20,000 in savings into equities, hoping -- along with half of the U.S. population -- for a major windfall.
He would drag himself out of bed at 6:30 a.m. to watch CNN Moneyline and CNBC, read analysts' reports, and buy or sell shares before heading off to teach finance at Ohio State University, where he worked at the time. Chen would stop by his department's Bloomberg terminals during the day to check on his portfolio and read business news. He followed analysts' advice to the dot.
In two years, his savings melted like snow in his native China's summer sun. "I always got in at the high, got out at the low," he says. "By [the markets' close at] 4 p.m., I was mentally exhausted." So, Chen decided he needed to come up with a better way to play the market, one that would require no emotional involvement.
BEATS A COIN FLIP.
To improve his accuracy as an investor Chen, now 39, turned to math. After four years of work developing a new quantitative investing model, Chen in 1999 launched ValuEngine.com with a partner from a hedge fund. One of the few independent quantitative-analysis firms on the Street, VE uses mathematical calculations to forecast fluctuations in share prices as far as three years into the future.
ValuEngine.com claims a 70% accuracy rate in predicting price trends for individual stocks, according to President Paul Henneman. That's far better than average investors who diligently follow the news and analysts' reports: They make good bets only 50% of the time, according to VE's research -- no better than if they simply flipped a coin.
The accuracy of its forecasts is the main reason VE's analyses are used by more than 25,000 financial advisers, hedge-fund managers, and individual investors, Henneman says. VE's user base has been doubling every quarter, and the company is profitable on a monthly basis, he says.
The site now charges a $14.95 monthly subscription fee to individuals. Institutional investors who want to use an expanded version of its service, in the form of software, pay $6,000 a year. It allows investors not only to see what a stock's future price might be but also to track their portfolios and make comparisons between different investing methods.
A few clients have enjoyed even higher levels of success than the 70% accuracy rate would lead them to expect. Ross Antony, a 43-year-old unemployed network engineer in Pineville, La., reports a 92% return on his $10,000 investment since he began using the service in October. "I don't know what's behind it, but it's working," he says.
Maria Peters, of Santori & Peters investment advisers in Monroeville, Pa., says she can't even remember when a VE recommendation backfired. She has been using it for years.
ValuEngine.com went through some rough times early on. As it struggled along with the rest of the dot-coms, a year ago it had to lay off 15 of its 25 employees, cut its marketing expenses, and move into a smaller office in Stamford, Conn. But persistence is one of Chen's trademarks. Before coming to the U.S. to get his PhD, he listened to Voice of America every day for four years to improve his English and understanding of American culture. He enrolled at Yale in the mid-1980s.
Today, the 10 ValuEngine workers who help Chen -- who teaches in New Haven, a 40-minute drive away -- crunch the numbers for reports on 6,000 stocks and can barely keep up with soaring client demand. Individual investors snap up VE's reports at Web sites such as Yahoo Finance and Hoover's Online. VE is now in talks with 50 other possible distribution partners, and it plans to start doing reports on the Asian markets in local languages.
Even rivals seem respectful of Chen's approach. "I hate to say this about a competitor, but it's in the hands of competent people," says Samuel Eisenstadt, research chairman at asset manager and researcher Value Line (VALU ). He says Chen has "an ambitious project here."
Indeed, unlike better-established competitors such as Value Line, in business since 1931, VE goes well beyond simply ranking stocks by parameters such as safety. It develops charts showing exactly what it predicts will happen to a stock's price for up to the next three years. VE can do so because it not only incorporates such common parameters into its forecasts as past and future earnings and interest rates but also attempts to find patterns in past stock-price fluctuations to help forecast future share prices.
ADDING IN IRRATIONALITY.
Chen also stands out from many other quantitative analysts because he makes assumptions about how earnings and interest rates will evolve over time -- something Value Line doesn't do. He believes the market often behaves irrationally and doesn't necessarily move each stock toward its fair value right away, creating opportunities for investors.
Chen's methods go against the grain of most academic thinking. Today, an MBA fresh out of school would simply calculate a stock's fair-market value by the discounted cash-flow method: Estimate how much cash a company will generate in the next several years, then translate that into current dollars. Since a company's performance is in large part tied to its ability to generate cash, that helps determine how much its stock is now worth. If the resulting number shows the stock is undervalued, investors might want to snap it up.
Although Chen teaches this method in his classes, he doesn't believe in it because it doesn't take into account investors' reactions, he says. An example: Shares of computing giant IBM don't go down as much as those of some smaller companies because some investors are always waiting on the sidelines to get in when IBM dips, Chen explains. Thus, IBM shares always enjoy more demand and command a higher price. (Joseph Granville, a top 1970s investment guru and an early quantitative analyst, pioneered this idea.)
NOT "ONE ANSWER."
Chen has his critics. While admitting that his model "might work short-term, it's unlikely to work reliably," says Lewis Johnson, a finance professor at Queen's University in Kingston, Ont. That's because stock-price behavior is mostly rational, he explains. The moves up and down can be the result of a pickup or a decline in orders. Furthermore, "predicting irrationality is more difficult," says Johnson. "One cannot predict [irrational price moves] from past data."
Certainly, no quantitative method can work at all times. "There isn't one answer," says Charlie Pluckhahn, an analyst at investment firm Stephens Inc. "If there was one answer, there wouldn't be a stock market." After all, whenever someone makes money on the stock market, someone else loses it.
However, Chen contends that his model "can make the best-effort assessment based on what we know today and reduce risk." And in the wake of the Internet boom and allegations that analysts promoted questionable stocks, many investors are leery of brokerage-house research. Unbiased, unemotional figures like VE's seem more trustworthy to many investors these days.
In any case, Chen is putting his money where his mouth is. Several months ago, he co-founded a private hedge fund with several partners as investors. Chen, who has put all his personal savings into the fund, is in charge of creating stock-picking models for the fund: modified versions of VE models, with parameters such as loss control added in.
Chen says he doesn't participate in picking the stocks for the fund, which so far has made an 18% return. But he's hoping it will continue to do better than his first fumbling efforts as an investor back in his Ohio State teaching days.
By Olga Kharif in Portland, Ore.
Edited by Thane Peterson