(FORTUNE Magazine) – It's a morning in March, and a trader at Robertson Stephens, a top technology investment bank, sits in the sprawling trading room, the marshaling ground for the firm's regiment of Bloomberg terminals. The trader (who spoke on condition that his name not be revealed), a soft-spoken man who belies the stereotype of the arrogant trade jockey, is ready to open trading in a hot new issue that Robertson Stephens is underwriting. On his computer he has orders from other market makers to buy the stock for as much as $37 a share; he also has sell orders from his institutional clients who have already decided to flip their shares and take their profits. But at $37 a share, he has buy orders for 100,000 more shares than he has to sell. He looks over his order book. The next batch of "buys" is clustered at around just $30 a share. He doesn't think the stock will hold up at $37. The decision takes just a minute; the trader makes the "print," executing the buy order at $37. That means he has now sold 100,000 shares that he doesn't have.

The moments following are tense; if he has bet wrong on his "short" and the price rises, he could very quickly be in the hole for a lot of money--in the current market, once a new stock issue starts going up, it might not stop. In this case, however, the trader bet right. The stock settles back to $30 a share. He covers his short, making $700,000 for his firm in mere minutes.

Welcome to the hottest game in the financial world, the IPO free-for-all--a high-stakes sport in which investment banks, institutional funds, huge market makers, and, yes, a growing mass of day traders and Net true believers compete to bring home instant prizes sometimes worth tens of millions of dollars.

Tales of hit-and-run scores on such initial public offerings are now all too familiar, of course. From 1980 through 1998, according to data compiled by University of Florida finance professor Jay Ritter, a total of 39 IPOs ended their first day of trading at double their offering price or better. In 1999 alone, the number of one-day doublers soared to a startling 119. (The only frenzy in the U.S. financial markets that is even comparable is probably the tsunami of quick-rising penny-stock offerings in the late 1960s.)

Sure, as many small-time investors have jealously come to accept, the IPO market has yielded a bonanza for well-connected types who get share allocations at low offering prices and then immediately flip them for several times what they paid. But in fact others--especially the underwriting banks--are playing the system more cleverly than that. And in a hot new issue, much of the game happens in the wild first minutes of trading.

Before it all starts there's a brief period (no more than 30 minutes) in which the lead underwriter of the offering and other market makers send one another electronic indications of the prices at which their clients are willing to buy and sell shares. Both the underwriting firm and the market makers--chiefly Knight Securities, Herzog Heine & Geduld, and Schwab Capital Markets--aggregate orders from a large number of individual investors and online brokerages. Ideally the indication window will let these trading maestros find an equilibrium price at which the stock will start off; in the hottest offerings, though, buyers and sellers often fail to get even close to equilibrium, leading early trades to be made at wildly divergent prices.

And that is when things get really interesting. The lead underwriter--the first place that institutional investors turn to when they want to flip their shares quickly or, in some cases, buy more shares to add to their IPO allocation--and the big Nasdaq market makers each begin with orders to buy or sell hundreds of thousands of shares in their queues. The moment the stock is free to trade, the desks roar with activity.

Investors often look at the opening and closing prices on the first day. But to the banks and market makers, what happens in between is what's important--especially what happens in the first official minutes of trading. Some stocks start high and spike up. Others can walk off a cliff--Palm dropped from $140 to $100 in its first 20 minutes of trading, as some ten million shares changed hands.

The lightning-quick pace is set by the lead underwriter's trading desk, which is responsible for maintaining liquidity in the stock. Naturally, investment banking firms present themselves as impartial brokers, trying to balance supply and demand in a dangerous market. But make no mistake about it: Their clear sense of where the market is headed benefits their own trading desks too.

Look back to that trader at Robertson Stephens. Underwriters, as it happens, operate with a number of advantages over every other investor, and certainly the typical retail investor. The lead underwriter--and only the lead underwriter (not even the co-managers listed on the cover of the prospectus)--gets to see the syndicate book, the list of institutions that received shares in the IPO. That gives the underwriter an edge in predicting what will happen to the new stock issue, since all major underwriters keep track of what institutions do with their allocations--for example, whether they're likely to flip their shares. Moreover, by looking at the huge number of buy and sell orders from its own clients, the underwriter's trading desk can, as in the case of the Robertson Stephens trader, decide to buy shares for its own account or sell short. Bankers can still lose money on a wrong bet. But on the whole they're likely to pay off. Cornell University professor Maureen O'Hara showed in a 1998 study that even before the current IPO frenzy, investment banks made money in after-market trading of their issues. Now, she says, they're undoubtedly making much more.

But there's another twist to the story. Privately, investment banks say that now they are being left on the sidelines as thousands of retail investors bid up new issues. How is that possible? A great proportion of the orders from online brokerages are routed to market makers like Knight and Herzog. And that "limit order book," in turn, gives those firms an insight into the stock's movement that even the underwriter doesn't have. Increasingly, such dealers--armed with retail orders to "buy at the market" or with sky-high limits--are opening the trading. They, like our Robertson trader, often bet that demand will flag after those heady early minutes, and then sell the shares short at lofty prices.

So where, dear reader, does that leave you? Put bluntly, treading water. In this fast-moving game, whether it's the underwriters or other market makers who profit, it's investors who absorb the cost. Even some of the fund managers who score big on skyrocketing IPOs now complain about the speculative fervor of the Big Bang. Alberto Vilar, head of the powerhouse Amerindo technology funds, says that he routinely tries to buy shares in promising new issues with the intention of holding them for years, but he gets priced out of the market when the stock opens too high.

And hey, when even brand-name money managers complain about the process, you can be sure that some change is afoot. Seeking to counter the power of the Knights, Herzogs, and Schwabs, some of the biggest banks have recently pushed for Nasdaq to open the limit order book, thereby giving them access to the data too. The thinking is, the more players that can accurately gauge supply and demand, the more price stability there should be. Of course, the glasnost could go even further: The exchange could open not only the limit order book but the syndicate book as well, letting everybody see which institutions receive allocations at the IPO price (and by extension, who's likely to flip the new shares). Unfortunately, none of this makes it any easier for you, the individual investor, to get into an IPO at the offering price. But that's another story.

THE TRUTH ABOUT THOSE CHAT ROOMS | STOCK OPTION CATCHES | SERWER'S 'COCKROACH' WARNING