In November 2004, a man with half a million dollars was looking to double his money. He could have gone to a high-stakes table in Vegas. He could have put his money in a hedge fund and waited. Instead, he signed on to a Dublin-based website, Intrade, the night before the presidential election and put everything on George W. Bush, then running even with John Kerry. The next morning, he was a millionaire.

Carl Wolfenden, the acting exchange manager of Intrade, explains the logic of prediction markets: “Our members sign up and trade with the intention of making money. One of the byproducts of that, the pricing information that they generate, translates into probability. The market pricing is measuring the probability of uncertain future events.” The results are eerily accurate.

Tens of thousands of people bet money on who would win each state in the 2004 general election, and Intrade’s political futures market predicted the winner in all 50. Two years later, the Intrade favorite won every single Senate race. Investors, or, if you prefer, gamblers, were generating political predictions far more accurate than professional pollsters’. They were also winning and losing piles of money.

But the political prediction markets didn’t begin with high-rolling political junkies. They started with a few dozen college students. In 1988, the University of Iowa business school opened the Iowa Electronic Markets as an experiment. They allowed anyone to buy contracts based on how they thought a given candidate would do in an upcoming election. The market developed a price per share. If the market moved Candidate A’s price up to 50 cents a share, it was saying that Candidate A had a 50 percent chance of winning.

“We collected almost 1,000 polls that came out during the election cycles, and compared the poll prediction to the IEM prediction,” says Professor Joyce Berg, “and in 75 percent of cases the IEM was closer to the actual outcome than the polls were.” One study showed that IEM’s prices on the eve of an election were off by an average of just 1.37 percent.

Berg says that the people trading are nothing like a random sample of voters. “In 1988, everyone was from Iowa, and we only had 155 people in the voteshare market. Even now, when we have thousands of people in each market, we are not distributed among states by population. Our traders are overwhelmingly male. They have more education than the average voter. They have a higher income than the average voter. But the market mechanism is one where we don’t need a random sample of voters, we need people with information.” In other words, a large representative sample of the electorate cannot accurately predict its own behavior when asked a simple question. But a group of students betting their spring break money can.

The IEM limits its traders to accounts between $5 and $500 in order to avoid a crackdown by the government. Those who want to make larger wagers have to go overseas and deal with Intrade. The Internet Gambling Enforcement Act of 2006 prohibits American banks from making credit-card payments to offshore gambling sites. The only Americans betting on Intrade have offshore accounts and use foreign addresses. If Wolfenden set foot on U.S. soil, he would probably face arrest and prison. But the exchange he runs is the most hailed prediction market worldwide.

This year, over 50,000 contracts were bought on the New Hampshire primary, and so far 2 million contracts have been purchased on the Democratic nomination. Clinton’s futures were more highly valued than Obama’s even after a string of Super Tuesday defeats. But once news broke that Clinton was lending her own campaign money, Obama’s price surged ahead in both the IEM and Intrade.

There was almost as much interest in Republican outcomes. Those poor souls who were bullish on Fred Thompson either sold short or stayed in their bad position, losing everything. But one couple gained a small fortune. Last summer there seemed to be no chance that John McCain would be the nominee. His amnesty bid backfired, his poll numbers in Iowa plummeted, and he couldn’t raise any money. His price on Intrade dipped below 5 cents a share. But one trading duo, Bethen and Jonathan, saw an opportunity. If McCain won the nomination, each share they bought for a little under 5 cents would pay a dollar. Already they’ve seen their investment increase in value by a factor of 19. Barring any dramatic developments between now and the convention, the pair will be tens of thousands of dollars richer.

The small trend of prediction markets officially became a phenomenon when James Surowiecki included them in pop-economics bestseller. A financial journalist for The New Yorker, Surowiecki landed on the lists with The Wisdom of Crowds in 2004. He touted sociological studies showing that a group of people guessing the number of jellybeans in a jar was almost always more accurate than any one single person. In case after case, crowds of people in the free market could aggregate and organize information like nothing else. Surowiecki promoted the Hollywood Stock Exchange (which only uses play-money) as the most accurate predictor of box-office receipts. He named the IEM and Intrade as the best sources for political predictions, saying, “a decision market’s fundamental characteristics—diversity, independence, and decentralization—are guaranteed to make for good group decisions.” Prediction markets weren’t just a fun trend in gambling, they represented a big idea: mobs can be more clever than individuals.

But there are some anomalies in prediction markets. Al Gore’s prices in the 2008 Democratic nomination markets were higher than announced candidates like Bill Richardson and Chris Dodd. As late as last fall, he had a one in five chance of taking the nomination, though he wasn’t running. Wolfenden explains, “There seems to be a group of traders who are Al Gore fans. Some of them are still holding out the possibility that he may be the Democratic nominee.” Gore’s inflated stock reflects a similar phenomenon in sports. A Giants line can almost never be trusted because there are so many New York football fans who feel they can’t bet against the home team. Intrade’s clients, made up of foreigners and rich, college-educated American males who have offshore accounts, have serious man crushes on Gore.

Despite small blips like this, high-ranking officials may want to periodically check in on their job security. As the 2006 Congressional election approached, rumors that Secretary of Defense Donald Rumsfeld was on his way out had died down, so his resignation the day after the election shocked the major news media. If network executives had checked the predictions markets, they would have seen traders bid Rumsfeld’s departure up by several factors in the 24 hours before he resigned.

Intrade takes investments on more than political outcomes. High-volume traders make several thousand dollars a day betting on whether American financial markets will finish up or down. The market also allows people to bet on whether Scooter Libby will receive a pardon by the end of Bush’s term (currently trading high at 65 cents a share) or whether the first Category 3 storm of the next hurricane season will make landfall in Florida, Texas, or Alabama. You can even bet on bird-flu. Would Intrade take bets on when the pope will die? Wolfenden says no. “For a proposition to be traded it has to fall within the bounds of good taste. We try not to get into death contracts or that sort of thing.” The market’s principals don’t want to encourage anything untoward, but they do take suggestions. “The craziest one we’ve been asked for was: ‘Jesus to return in next ten years, or Rapture, or apocalypse,” Wolfenden reports. How Intrade would pay out if the apocalypse occurred was not specified.

Prediction markets also furnish us with a prospectus for peace. The numbers are grim. There is only about a 5 percent chance of Hamas recognizing the right of Israel to exist by Sept. 30. And according to market wisdom, America is nearly twice as likely to “launch overt military action against North Korea” by New Year’s. A second Korean War would probably rate higher if the Bush administration weren’t rattling its sabers at Tehran. Traders are giving war with Iran a one in five chance by the close of the third fiscal quarter. And investors are not bullish on the chances of catching Osama bin Laden. Wagers on his capture by 2009 are trading at just over a dime.

Those who have qualms about betting on war and peace still have the more conventional 2008 horserace. Jim Webb is trading in the vice presidential markets at almost the same price as Hillary Clinton. But nearly one third of investors have put their money on “the field,” believing Barack Obama will choose a relatively unknown candidate. On the Republican side, Minnesota Gov. Tim Pawlenty is the one in five favorite, and Mitt Romney’s recent grinning and groveling are moving his veep shares up quickly.

But if the numbers hold, it’s doubtful that either will see high office soon. The IEM has the Democratic share of the presidential vote around 50 percent, and the Republican share at 47 percent. The smart money says President Obama.