(FORTUNE Magazine) – I WAS SITTING in a bikers' strip bar in Salem, Oregon, next to a naked young woman named Shelby when it struck me that this wasn't turning out to be a typical FORTUNE research project.

I was there--honest--to talk to the males at the table about quantitative analysis, computer technology, and the new Wall Street. They are a band of entrepreneurs (you'll meet them in a few pages) launching an offshore money fund based on cutting-edge artificial intelligence and infotech, and they're dead serious about their business. Shelby, it turned out, was an...er...acquaintance for whom they had agreed to design an erotic World Wide Web page on the Internet (http://www.DynamixTrading.com/shelby).

These guys sure aren't the buttoned-down Wall Streeters of yore, and they may not be totally typical examples of quants who are becoming increasingly important in finance and money management. But they're not all that atypical either. "Quant" is Wall Street jargon for "quantitative analyst," or someone who uses mathematics, statistics, and souped-up computer technology to solve financial problems--like how to beat the markets. The quants' credo is that there is method to the madness of finance, and patterns in the turbulence of world markets that can be exploited for profit. The trick is to recognize the patterns, and quants--many of whom have doctoral degrees in fields like physics and computer science--employ the latest techniques from academia to do just that.

DOYNE FARMER and NORMAN PACKARD PREDICTION CO.

Might it seem hubristic to name a stock-picking firm Prediction Co.? Doyne Farmer, one of its founders (and a chaos theory pioneer), points out, "we don't call ourselves the Perfect Prediction Co., or even the Good Prediction Co. Just Prediction Co."

Farmer and his partner, Norman Packard, are physicists by training--and onetime professional gamblers. They've been predicting the difficult since the late 1970s, when they devised a way to beat roulette by using computers built into their shoes. Now they're predicting the ups and downs of financial markets out of their office in a former artist's studio in Santa Fe. Farmer, 43, and Packard, 41, are native New Mexicans and have been friends since childhood. Farmer cut his academic teeth at Los Alamos (as in A-bombs), where he ran the complex systems group. Packard was a fellow at the Institute for Advanced Study (where Einstein once worked) in Princeton, New Jersey, before becoming a physics professor at the University of Illinois. Both quit their jobs in 1991 to form Prediction Co.

Here's their approach to solving problems, both scientific and financial: Take a seemingly random phenomenon, like sunspot cycles, and build predictive models using mathematical formulas. "Just because something is chaotic doesn't make it unpredictable in the short term," says Farmer. If the model works, it will predict the future of that system with reasonable accuracy. As Farmer and Packard learned from their gambling days, reasonable accuracy doesn't mean perfect accuracy. To beat the house odds--whether at a casino or in the stock market--you don't need to win all of the time, just most of the time.

Farmer and Packard see their work in financial markets as a logical extension of their earlier academic research. The main difference is that success and failure in financial markets are more concrete. Outside of quantum mechanics, says Farmer, nothing's harder than forecasting financial markets. "Frankly," he says, "the fact that this was supposed to be impossible appealed to us."

There's also the money, of course. Prediction Co. doesn't trade on its own but rather has an exclusive agreement with Swiss Bank Corp. The software in Santa Fe sends trading signals to Swiss Bank traders in Chicago, and Prediction Co. gets a cut of the profits. Neither party will say how well it's doing, but David Weinberger, a Swiss Bank executive, says he's delighted. He has a growing confidence that Prediction Co.'s success so far is sustainable and isn't just dumb luck.

MARTIN FREEDMAN PARALLAX

JIM STEINER PARALLAX

A transcript of a weekend spent with Rafael deNoyo (shown at right with client Shelby Rose), Martin Freedman, and the other members of their band of merry quants reads like a mix of David Mamet, Isaac Newton, and Robin Williams. Conversation careens from pure mathematics to (real) war stories to fetish parties in London. Whatever room they occupy is filled with peals of laughter and cheap cigar smoke, at any venue from a nice Italian restaurant to a casino or strip joint.

Not really a single company in the traditional sense of the term, they are members of a loose alliance of quant-entrepreneurs trying to put their knowledge of sophisticated statistical analysis and computer science to work. Last spring deNoyo started his own research company, called Dynamix Trading, out of his garage. Freedman and his partner, Jim Steiner, formed an offshore statistical arbitrage fund called Parallax, run out of the British Virgin Islands. Freedman also runs a separate company in London called MetaModel Research. All the principals do contract work for one another; keeping straight who does what for whom is like following the strands in a sieveful of spaghetti.

Despite their hijinks, these edgy quants say they're completely serious about making money. Judging from the initial returns of Parallax, they're doing something right. The fund's software sits on a computer file server in the British Virgin Islands and is monitored via the Internet from Oregon and London. Parallax just opened to offshore investors. Freedman and Steiner won't speak for attribution about Parallax's performance, but FORTUNE was told that the fund has earned 45% since it went online last January.

The resumes of this cast of characters are as offbeat as their business approach. DeNoyo says he left home at age 15, never went to high school but attended Berkeley for three years, and has made a career working at various investment companies, including Fox River Financial Resources, now defunct. Jim Steiner, Parallax's dealmaker, claims to have fought in Nicaragua with the Contras--and to prove it sent FORTUNE some gruesome snapshots of himself crouched over dead soldiers. Freedman, the Londoner who did much of the math for Parallax, says his career has spanned everything from engineering to finance to...dancing.

As the boys shoot the breeze around deNoyo's kitchen table in Salem, Freedman mentions that he and Steiner came up with the idea for Parallax around the time they met at a kinky London nightclub. As soon as this fact is revealed, the whole crew bursts into laughter and is off again. Here's a snatch of the conversation:

DeNoyo: "It's a little like Mardi Gras..."

Freedman: "There's a little whipping on the dance floor..."

Steiner: "Isn't this how all businesses get started?"

BRAD LEWIS FIDELITY

You don't know algorithms from Alka-Seltzer? Well, you can still take advantage of high-tech investing. All you need is a quant fund, like the ones Brad Lewis runs for Fidelity. Although they underperformed the market a bit this year, Lewis's Disciplined Equity Fund has had average annual returns of 18% since its startup, and his Stock Selector Fund has returned 23%--both somewhat better than the overall market.

Lewis developed an interest in quantitative problem solving as an operations research major at the Naval Academy. After putting in five years in the Navy flying helicopters (he still flies his own plane, pictured above), Lewis got an MBA at Wharton and found a job as an analyst at Fidelity. At first he did research the traditional way--traveling to companies and picking over balance sheets. But he soon started fiddling with computerized analyses of data series (such as trying to find relationships between a company's earnings or price-to-book and the fluctuations of its stock price). He found that he was able to forecast prices pretty well using his software, and off he went into the wild blue yonder.

Today, Lewis has seven researchers and money managers working for him, looking at about 65 variables ranging from macrostatistics like GDP to analysts' earnings estimate revisions. They plug the numbers into their computer software model to get their stock picks.

Like most quants with a good thing going, he won't divulge details of the technology he uses. Until last year he used a type of artificial intelligence called a neural network, a self-instructing program that uses lots of miniprograms to solve complex problems--and a common method used by quants. Now, Lewis says he's switched to "hard-core artificial intelligence" technology based on an academic paper he came across last year.

ANDREW LO MIT'S SLOAN SCHOOL

His model rocketry hobby notwithstanding, Andrew Lo isn't a rocket scientist per se. He just trains them. Lo is the architect and chief engineer of a quant factory at MIT's Sloan School of Management. At age 35, with an endowed chair in the finance group, Lo has created a special track in "financial engineering" for MBA students yearning for quanthood.

The goal of his program, says Lo, is to meet the rising demand for financial sophistication as markets become more complicated. In addition to the usual B-school classes like accounting and marketing, students take courses in financial technology, which Lo describes as "a combination of mathematics, statistics, and financial analysis rolled into one." Lo's students go on to careers as fund managers, risk management specialists, and other quantish jobs.

To augment the financial engineering curriculum with a real-world look and feel, Lo recently set up a mock trading floor that has all the trappings of the real thing: authentic trading desks; live stock quotes running across Trans-Lux boards; even clocks on the wall that tell the time in New York, London, and Tokyo.

Lo talks as easily about his favorite science fiction novels as he does about his research. Actually, his career in quantitative finance was inspired by science fiction, notably Isaac Asimov's Foundation, the sci-fi classic about mathematicians who buck conventional wisdom and predict the future. The good guys can't foretell the actions of individuals, but they figure out how to predict, among other things, the mass movements of galactic populations.

The idea of the predictability of groups of people--like financial markets--grew from a pulp fiction plot line into a recurrent theme of Lo's research. Some of his early research skewered the random walk hypothesis, the staple of finance theory holding that market forecasting is impossible. His most ambitious goal sounds suspiciously familiar: He hopes to develop a theory to predict, among other things, the mass movements of financial-market participants. "Asimov was a terrible writer," says Lo, giving credit where credit is due, "but he had some incredible ideas."