(FORTUNE Magazine) – It's the kind of record that would do almost any team proud: For 14 of the past 17 years, the Alpha Tilts large-cap stock fund managed by Barclays Global Investors in San Francisco has beaten the S&P 500. Not impressed? Okay, check this out: The active management crew at Barclays Global also runs a small-stock fund that has outperformed the Russell 2000 index ten years out of 12. Plus an international-stock fund that has whipped the MSCI Europe, Australia, and Far East index eight years out of ten. It keeps going like that on down the list: four more small-stock funds, a Britain fund, a Europe fund, an Australia fund, a Canada fund--all of which thumped their benchmarks on a regular basis.

In a world where most money managers fail to beat the market over time, such consistent thrashing of the market averages is unusual, to say the least. But what's truly weird is that the people doing it work at Barclays Global, a place that owes its very existence to the notion that you can't beat the market and are better off just trying to match it in an index fund. In its previous incarnation as a division of Wells Fargo, Barclays Global pioneered index investing back in 1971. It is now the world's biggest indexer and the fourth-biggest asset manager overall. If you've never heard of it, that's because its main customers are pension funds, foundations, and endowments--not retail investors.

But while Barclays Global will always be linked to the academic belief in unbeatable markets (the term of art is the "efficient-markets hypothesis"), that belief has fallen out of fashion. It is no longer considered heresy--at the University of Chicago, at MIT, at Barclays Global--to argue that very skilled, very careful investors can consistently beat the market. Which is exactly what Richard Grinold, a former finance professor at the University of California at Berkeley, Harvard physics Ph.D. Ronald Kahn, and the overeducated research staff they command at Barclays Global set out every morning to accomplish.

How do they pull it off? By doing the things careful investors have been doing since time immemorial--or at least since 1934, when Benjamin Graham and David Dodd published Security Analysis. But they don't look at just a few dozen companies; they look closely at thousands. They don't follow hunches; they follow computer models. They don't try to hit home runs; they try to hit "lots of singles," as Kahn puts it. "Central to our approach is a certain amount of humility," he adds. "We don't believe that markets are efficient. But we don't think they're very inefficient either."

This is modern quantitative investing. It falls somewhere between stock picking as it has been practiced for generations and the computer-driven "black box" techniques made famous--and infamous--by highfalutin hedge funds like Long-Term Capital Management in the 1990s. It's not magic--though they beat the market, the Barclays Global funds still lost investors a lot of money last year. But it is increasingly popular. Barclays Global now actively manages $136 billion, up from $108 billion at the end of 1998, despite the intervening bear market. Other money managers who engage in the sort of fundamentals-based quant investing practiced by Barclays Global report similar gains.

So far it is a game just for the big boys. None of Barclays Global's active funds are available to retail investors, and the same is true for all but a handful of similar offerings from the competition. That may change--investing ideas have a tendency to drift down from the institutions to the little guys over time. But retail investors also prefer investing in ideas they can understand. And it's an open question whether the particulars of this investing idea can ever be made comprehensible to the layman.

To answer that, I've flown out to San Francisco, to see whether the Barclays whizzes can make their work comprehensible to this layman. Over lunch in a 32nd-floor conference room in Barclays Global's nondescript Financial District headquarters, Grinold and Kahn do their best. Grinold, 64, studied physics at Tufts, served as a navigator on a Navy destroyer in the early 1960s, then got a Ph.D. in operations research from Berkeley. While teaching at Berkeley's business school in the early 1970s, he slid over to the then wildly fashionable field of finance. Kahn, 46, studied physics as an undergrad at Princeton, as a grad student at Harvard, and as a post-doc at Berkeley before deciding to go into the money business. Both men are pleasant enough, and keep the finance mumbo-jumbo to a minimum--but they are prone to long, awkward pauses and maddeningly abstract statements. So after a while they take pity and escort me to another conference room, where several bright young faces from their team have gathered to tell me about what they do.

"There's been a revolution in access to financial data," says Jonathan Howe, a UCLA finance Ph.D. "It's really been global, and that's been great for us." The challenge, says Ernest Chow, who has a Ph.D. in applied math from Australia's Monash University, is figuring out which data to heed. The core measures all have to do with value: Price-to-book, price-to-earnings, price-to-whatever. But the Barclays Global gang also pays attention to signs that a stock is on a tear: Analysts are ratcheting up their earnings estimates; a company is beating the estimates every quarter. These "momentum" measures don't work as well as they used to, because investors have caught on and corporations have learned how to manipulate earnings targets, but they still matter.

Then there's earnings quality, which not surprisingly seems to be Barclays Global's richest research vein at the moment. Amid the corporate scandals of a year ago, for example, research officer James Claus, a Columbia accounting Ph.D., started looking into companies with large leasing businesses, because lease accounting offers much leeway for fudging. "We saw that the numbers at Ford reflected very optimistic assumptions," Claus says. So they underweighted Ford. Good move: It's down almost 50% since mid-May 2002, compared with a 12% loss for the S&P 500.

Finally, when their exhaustive research and fancy computer models don't seem to be doing the trick, the team punts. In the first half of 1998, Grinold says, all the valuation models were telling them to underweight Yahoo, Excite, and the rest of the dot-coms. But underweighting those stocks was causing Barclays Global's funds to fall behind the indexes they're so focused on beating. So they threw up their hands, threw out their valuation models, and simply added the dot-coms to the funds at their index weights. The result: They rode the rampaging Net boom up and back down, although not as far down as the overall market. "If there's a truck coming down a one-way street the wrong way, do you jump out in front and say stop or do you get out of the way?" says Grinold. "We decided getting out of the way was a good idea."

In between dodging trucks, Grinold and Kahn have found time to write a book. It's called Active Portfolio Management: A Quantitative Approach for Producing Superior Returns and Controlling Risk. Despite the title, a $75 price tag, and lots of pages lousy with Greek letters, it has sold more than 10,000 copies. "The art of investing is evolving into the science of investing," the two men write portentously in their introduction. "As new generations of increasingly scientific investment managers come to the task, they will rely more on analysis, process, and structure than on intuition, advice, and whim. This does not mean heroic investment insights are a thing of the past. It means that managers will increasingly capture and apply those insights in a systematic fashion."

The roots of this "scientific" approach are to be found in the revolution that transformed academic finance in the 1950s and 1960s. Its best-known and most controversial component is the aforementioned efficient-markets hypothesis--the argument that markets work so well that everything worth knowing about the price of a stock is already reflected in that price. But possibly the greatest insight of the new finance is this: Investment returns are in part a reward for taking risks.

William Sharpe, a gregarious Southern Californian who got his economics Ph.D. at UCLA in 1961, put those risks into terms that investors could understand: Risk meant the possibility that a stock--or a portfolio--would go down more than the overall market. There was no point in talking about a money manager's performance without looking at how it compared with the market's performance, Sharpe argued. An investor who got a higher return than the market simply by taking bigger risks didn't really "beat" the market. Start looking at the world this way, and the margin of error for active investors quickly shrinks. To crib from a lecture often delivered by Sharpe: The overall return of the market is, by definition, the overall return of the active investors in the market. So the average active investor makes the market return, minus costs. Since the costs of active trading are more than the cost of buying and passively holding stocks, someone who simply buys the market--or a facsimile thereof--will invariably do better than the average active investor.

Thus was the index fund born. Sharpe--by then at Stanford--was involved as a consultant with Wells Fargo's pioneering index efforts in the early 1970s. (The better-known Vanguard 500 Index didn't come along until 1976.) But Sharpe's teachings about risk and return also began to find their way into the world of active money management. The man most responsible for that was Barr Rosenberg, a Berkeley economist who was also doing consulting work for Wells in the early 1970s. Sharpe had devised a simple gauge of risk that he called "beta"--which measured a stock's historical volatility in relation to the overall market. Rosenberg improved on that by adding data about companies' liabilities, earnings, industry sector, and the like--and soon went into business peddling his "Barr's better betas" and "Barr's bionic betas" to professional investors. BARRA, as he called his consulting firm, was a massive success. In 1979 he brought in his former Berkeley colleague Grinold, who went on to become BARRA's director of research and later president. In 1987, Grinold hired physicist Kahn, who'd come west to do postdoctoral research on the extinction of the dinosaurs.

BARRA's work using computer models to comb through companies' books in search of risk inevitably led its practitioners to try out similar techniques to pick stocks. Rosenberg left BARRA with much fanfare in 1985 to start his own asset-management firm. That same year, to much less fanfare, the indexers at Wells Fargo Investment Advisors launched their own foray into stock picking, with BARRA's Grinold devising the research strategy. That was Alpha Tilts, the perennially market-beating large-cap fund mentioned at the beginning of this article. Grinold came over to work at Wells full-time in 1993, and Kahn joined him in 1998. (British banking giant Barclays bought and renamed the outfit in 1995.)

Rosenberg sold most of his money-management firm to French financial giant AXA in 1998. He now spends a lot of time teaching classes in Buddhism at Berkeley's Nyingma Institute, and he did not respond to FORTUNE's requests for an interview. But AXA Rosenberg is still going strong, managing $25 billion according to strategies that--as current CEO Bill Ricks describes them--sound a lot like those of Barclays Global. Grinold says Goldman Sachs Asset Management is another kindred spirit, and many other money managers employ similar tactics. But the Grinold-Kahn team stands out. "I think Barclays Global is the best on the equity side," says Eugene Flood, a former Stanford finance professor who now runs fixed-income quant manager Smith Breeden.

As mentioned above, most of this quant expertise is available only to institutional investors. But that's changing. AXA Rosenberg has a few retail mutual funds. Barclays Global, whose main retail offering now is a series of exchange-traded index funds called iShares, would love to do the same with its quant strategies if it could figure out a marketing angle.

But can it ever work as an investing answer for the masses? Probably not. Active investing is simply too much of a zero-sum game. "When we're trying to beat the market, we each try to beat each other," says Sharpe. "If I have a better computer algorithm than you have, you'll eventually either quit trading or get an even better computer algorithm." Needless to say, Grinold and Kahn have a pretty danged good computer algorithm. But what do these perennial market beaters do with their own money? Stick it in index funds! (Although Kahn does cop to a couple of hedge fund holdings.) They advise you to do the same.