Can Technology Build a Better Buffett? In theory, artificial intelligence should outwit the best minds on Wall Street. The fact that it hasn't teaches some valuable lessons about how investing really works.
By Carla Fried

(Business 2.0) – If ever there were a field in which machine intelligence seemed destined to replace human brainpower, the stock market would have to be it. Investing is the ultimate numbers game, after all, and when it comes to crunching numbers, silicon beats gray matter every time.

Nevertheless, the world has yet to see anything like a Wall Street version of Deep Blue, the artificially intelligent machine that defeated chess grand master Gary Kasparov in 1997. Far from it, in fact: When artificial-intelligence-enhanced investment funds made their debut a decade or so ago, they generated plenty of media fanfare but only uneven results. Today those early adopters of AI, like Fidelity Investments and Batterymarch Financial, refuse to even talk about the technology.

Still, artificial intelligence has steadily improved in sophistication and quietly made itself indispensable on Wall Street. According to Andrew Lo, director of the Laboratory for Financial Engineering at MIT, every investment firm embracing a math-driven strategy uses some form of AI in its research, and Lo expects the terminology to appear again soon in promotions for retail investments like mutual funds and privately managed accounts. Before the hype machine cranks up this time, however, it would be smart to figure out just what AI can and can't do for investors.

AI can't think like Warren.

True artificial-intelligence software is designed to model human decision-making and, like humans, to "learn" from experience. That's far more ambitious--and potentially more useful--than standard data-crunching and stock-screening programs.

But how do you "model" the mind of a Warren Buffett or a Peter Lynch? Data flows in not just from standard databases but from everywhere: CNN, hallway conversations, trips to the drugstore. (Remember Lynch's inspiration to invest in Hanes after his wife gushed over the company's hosiery?) Lo's own research found that feelings are another extremely important input for professional investors. Lo concluded this after attaching electrodes to traders working in Boston; the devices measured increased skin conductivity--and, hence, heightened emotions. "Unless you can put an emotional value on certain events and actions," he explains, "you can't get the job done."

Naturally, investors don't process this hodgepodge of inputs according to some set of explicit, easily transcribed rules. Instead, the mind matches the jumble against other jumbles stored in memory and looks for patterns, usually quite unconsciously. "Often, great investors can't articulate the nature of their talent," Lo admits. "They're like pool players who make incredible trick shots on intuition." Fine for them, but how do you code that?

That's why the notion of replacing money managers with AI will remain science fiction for a long time to come. In the real world, AI works best in the background, helping investors make better use of well-defined sets of data. Experience shows that such intelligent data munching can boost performance and consistency. But as Lo concedes, "Buffett's in no danger of losing his day job."

AI programs are only as smart as the researcher running them.

It's a big data universe out there, and even the most powerful computer can't examine all the possibilities. As a result, the most crucial decision in any AI-driven investment strategy--which data to analyze--has to be made by the human in charge of it.

Andre Archambault, for example, manages Standard & Poor's Neural Fair Value 20, an AI-enhanced model portfolio open to subscribers of S&P's Outlook newsletter. His AI software analyzes the 18 financial variables that he uses to calculate fair values for his universe of 3,000 stocks, looking for attributes most closely associated with the top performers of the last six months. The program then matches those attributes with equities in the database. Since adding AI in 2000, Archambault's portfolio has increased in value by 55 percent, while the S&P 500 has declined 26 percent.

Even so, Archambault is measured in his enthusiasm for artificial intelligence, attributing 80 percent of the portfolio's success to his core "fair-value" strategy and just 20 percent to AI. "You have to have something solid to start with," he says. "If you assume [artificial intelligence] is going to be the answer on its own, you are going to have a garbage-in, garbage-out problem."

AI won't prevent a system from blowing up.

The most visible poster child for artificial intelligence in the early 1990s was Fidelity's Disciplined Equity mutual fund. Managed by Brad Lewis, a handsome former Navy pilot whom the personal finance press found irresistible, the fund ran up a spectacular record--for a while. Disciplined Equity beat the S&P every year from 1989 through 1994, and its margin over the index during that stretch averaged 3.8 percentage points. Then, in 1995, the system abruptly stopped working, and the fund dropped a cumulative 7.8 points behind the index by the end of 1996. (Overall, Disciplined Equity has slightly underperformed the S&P during the last decade. But Fidelity won't say if the fund still uses AI.)

While Lewis and his bosses have never offered any explanation, the fund's breakdown is unsurprising. History shows that such lacunae eventually strike every quantitative investment approach, from blunt rules of thumb like the "dogs of the Dow" (buy the 10 highest-yielding Dow stocks each year) to the arcane strategies cooked up by Ph.D.s at hedge fund D.E. Shaw Group. The reasons are endless: Financial conditions can change; other investors can catch on, eliminating a winner's edge; or tastes can shift, and what excited the market once can subsequently leave it cold. Not even a wily computer can foresee everything.

The good news is that these breakdowns are usually temporary. After some tweaking of the formula or a turn in the market cycle, performance often returns. Just know that virtually no system is infallible.

AI's future is in personal finance.

Lo believes AI's next breakthrough will be to personalize investment services. His particular brainchild is something he calls a "smart index." Instead of measuring your portfolio's performance against, say, the S&P 500, an AI system would construct a personal index based on your financial goals, risk tolerance, tax bracket, and so on. If your performance lagged your smart index, the system would suggest appropriate changes to get you back on track. That won't make you into a Warren Buffett. But if it helps you keep your long-range plan in the groove, that's intelligent enough.