NEW YORK — I've been making some pretty big Hollywood deals lately.

NEW YORK — I've been making some pretty big Hollywood deals lately.

I'm swimming with the sharks on Wall Street too, and raking it in thanks to my uncanny knowledge of the Fed and currency exchange markets.

So how come I'm not rich?

Alas, it's all a game. Through the magic of the Internet, I've been participating in simulated financial markets that let people wager fake money on the outcome of real events.

Economists have long been known that financial markets can forecast events with remarkable accuracy. That's the whole idea of futures markets — people trade based on what they think the price of crude oil or winter wheat or pork bellies will be at some future date. Even if most individual traders guess wrong the market as a whole almost always comes very close to the actual price.

Recently, some economists have suggested that people ought to be able to trade futures in just about anything, from the weather to medical and technological advances to the outcome of political contests. These markets could help shape government policies, guide venture capital to the most promising new technologies or help society prepare for natural disasters and other catastrophes.

"This is an institution for getting people who know things to tell it to you," said Robin Hanson, a professor of economics at George Mason University.

Internet sites have cropped up allowing the general public to speculate in these "prediction markets." A site called Hedgestreet allows sub-millionaires to speculate in markets normally reserved for the Wall Street elite. The Iowa Electronic Markets, operated by the University of Iowa business school, have let players wager up to $500 on the outcome of every presidential election since 1988.

There are also play money markets such as the Washington Stock Exchange, which lets political junkies speculate not just on elections but the appointments, resignations, legislative showdowns and such that make life on the Potomac so uniquely thrilling. On the left coast, the Hollywood Stock Exchange lets the moviegoing public predict which new releases will turn out to be blockbusters.

My account happens to be up nearly 300,000 Hollywood dollars, mostly thanks to my recognition that "Knocked Up" would do well at the box office and "Hostel: Part II" would flop.

"There's no doubt that the idea of using virtual stock markets is going to grow, and it is growing," said Hollywood Stock Exchange managing director Alex Costakis.

But how can a self-selected group of non-experts make accurate forecasts about complex events that are completely beyond their control?

That's the magic of markets. Spurred by the profit motive, people bring to markets everything they know, and place their bets accordingly. That means markets, with all their chaotic buying and selling and hedging and speculation, are really very efficient mechanisms to bring together information.

And all that information is revealed in one simple number — the price.

"The idea here is that prices aggregate information, the oldest idea in economics," said Justin Wolfers, a professor at the University of Pennsylvania's Wharton School of Business. "What we're seeing now is excitement about moving this beyond the standard financial sector."

Because of state and federal gambling laws, most currently operating prediction markets have to use play money instead of real cash. But even that doesn't seem to blunt their effectiveness.

The Hollywood Stock Exchange, which is owned by the financial services firm Cantor Fitzgerald, uses play money that is almost completely worthless (though 10 million Hollywood dollars will get you a $5 discount on a T-shirt). Yet the site's users forecast box office grosses so accurately that movie studios, advertising agencies and other clients pay real money for market data from the fantasy exchange.

The Inkling Markets web site, where users don't just trade but can also create their own markets on virtually any topic, uses a currency that is completely worthless. Right now you can use your Inklings — as the site's faux currency is called — to bet on who will win the New Hampshire Democratic primary, what the Federal Funds rate will be three months from today or which city will host the 2016 Olympics.

But the fact that you can't buy anything with them doesn't seem to hurt the performance of Inkling Markets. Top traders in the Inkling market say they are motivated by competitiveness, the simple desire to see themselves at the top of the site's leader board.

"When I log in the first thing I do now is check the top 10 list and see how No. 2 has been doing," said Mike Giberson, who was in first place for most of this spring but slipped into second himself just before this story went to press.

I, on the other hand, find myself somewhere in the neighborhood of 400th in the Inkling rankings. Whether that's enough to maintain my interest over the long haul remains to be seen. Last I checked, on Hollywood Stock Exchange my performance put me in an exasperating 519,179th place.

That's one reason some economists dream of the day any American can put real money down on, say, the number of people who will be killed by terrorists in the United States between now and 2010, or whether a hurricane will make landfall in Florida this year.

"Using these markets as forecasting tools could substantially improve decision making in the private and public sectors," more than 20 economists said in a letter sent to Congress and federal regulators last month.

The letter urged the government to carve out a "safe harbor" in the nation's online gambling laws for nonprofit, small-stakes, real-money markets that produce useful information as a byproduct of traders' efforts to make a buck.

"It's time to let the genie out of the bottle," said Robert W. Hahn, executive director of the American Enterprise Institute-Brookings Institution Center for Regulatory Studies and an author of the letter.

The economists suggested that any "economically meaningful" subject ought to be fair game, but not the kinds of things that would traditionally be considered gambling, such as markets based on sporting events or the outcome of "American Idol."

Hedgestreet CEO Steve Race argues that there is a subtle but important difference between sports betting and buying a futures contract on, for example, the price of gold.

"Gambling really is the creation of risk. It's artificial risk that is created on a football game or a game of chance," Race said.

But when you bet that the price of gold will go up or down, Race explains, you are ultimately buying existing risk from somebody who already owns gold or needs it. It isn't gambling because you're not creating risk, just redistributing it.

Get it? Neither do I. But the issue of what constitutes gambling looms large over the future of prediction markets. The Pentagon's Defense Advanced Research Projects Agency learned that the hard way a few years back when it funded a few economists to set up experimental markets that would track America's national security interests in the Middle East. The markets would have allowed traders to buy futures contracts on measures of military activity, economic growth and Western terrorist casualties in eight Mideast countries.

The hope was that prices in the Mideast security markets might give U.S. intelligence officials a leading indicator of developments in the region. But the project was aborted in 2003 when members of Congress protested that it would allow people to gamble on terrorist attacks the way they would on a football game. Opponents of the enterprise objected to the possibility that people might profit from a terror attack, or that terrorists themselves might manipulate the market to their advantage.

The episode convinced George Mason University's Hanson, who was one of the economists involved in the project, that for all their promise prediction markets might not be ready for prime time.

"This is all very new," he said.

How would the average CIA analyst react, for example, to the suggestion that an anonymous group of outsiders can produce intelligence information as well or better than traditional sources? It would require a faith in markets that is seldom found outside university economics departments and conservative Washington think tanks.

And that's not the only public relations problem prediction markets face. Researchers at Hewlett-Packard have been setting up internal prediction markets at the company for years, and they've found that they are enormously accurate predictors of everything from next quarter's printer sales to the price of memory chips six months down the road.

The problem, said researcher Leslie Fine of HP's Information Dynamics Lab, is that many people get tired of them. They don't want the additional chore of logging in every day to check on their shares. Some of them conclude very quickly that if the market is so good at getting the right answer, they can't expect to beat it. So why try?

Others may simply make mistakes, sort of like those befuddled Democrats in Miami Beach who accidentally voted for Pat Buchanan instead of Al Gore in the 2000 presidential election, thus helping send George W. Bush to the White House. The little technical errors of inexperienced traders can impair what economists refer to as a market's "efficiency."

For example, I recently made quite a (virtual) bundle in a practice market operated by Hedgestreet. Someone else on the site apparently thought they were betting that the interest rate after a recent Fed meeting would NOT be 4.75 percent, which would have been a very good bet. But they were really betting that interest rates would be 4.75 percent OR LESS after the meeting, which was very unlikely. I cleaned up, and then proceeded to blow my winnings on a series of risky currency trades.

Employers also face the opposite problem — some players understand what they're doing a little too well. At Google, another Silicon Valley company that has experimented with prediction markets, some savvy employees have developed complex arbitrage schemes and market-maker strategies that might help them win the game, but don't really improve the overall performance of the market.

That's why Hewlett-Packard has created BRAIN, for Behaviorally Robust Aggregation of Information in Networks.

"It's not even a market any more, really," Fine said.

Instead of having participants buy and sell shares or complicated futures contracts, BRAIN gives them a certain amount of real or play money and simply asks them to put it on one or more potential future outcomes. All the oddsmaking and other calculations, including an adjustment for each player's risk tolerance, happen in the background.

"It feels like a survey," Fine said.

Which brings us to the naysayers. Many skeptics argue that surveys work just as well or better than prediction markets in accumulating the knowledge of the masses. A recent study by Columbia University economist Robert S. Erikson compared the accuracy of the Iowa Electronic Market to that of public opinion polls during the 200 days leading up to the past five presidential campaigns.

Though he found the raw market predictions were generally closer to the eventual election vote percentages, Erikson argued that the polls were better if you applied some simple adjustment factors for things that they chronically misjudge, such as the incumbent's early advantage.

In other words, polls — combined with the knowledge of an experienced pollster — can outperform the market.

But even though he is skeptical about markets in the context of head-to-head presidential contests, Erikson still believes they have value in other situations, such as the rough-and-tumble free-for-all that is the early primary campaign season — i.e., now.

Most markets currently give Hillary Clinton an edge over Barack Obama in the contest for the Democratic nomination. The Republican side is more of a tossup, with John McCain, Rudy Giuliani, Mitt Romney and most recently Fred Thompson in the mix.

Who do I have? Nobody. I sold my Barack Obama shares at a profit, and rolled my Inklings into markets I know more about, like whether Barry Bonds will break Hank Aaron's home run record (of course) and whether Alberto Gonzales will leave office during the month of July (unlikely, I figure — if he's not gone by then I'm guessing he'll make it until August).

Play what you know — that's one of the secrets to success, the top traders say. Though you don't necessarily have to know much. Giberson has never watched a single episode of "The O.C.," but he knew there was an Inkling market to predict which character would be killed off at the end of the show's third season. So when he saw that actress Mischa Barton had just revealed her character was a goner the day before the big season finale, he used Google to find out which one she played.

Then he bought all the Marissa he could.