Listen to the people who got the election right
By Geoffrey Colvin

(FORTUNE Magazine) – THE MAIN REASON PSYCHOTHERAPISTS ARE TREATING Kerry supporters for a new psychiatric disorder--post-election selection trauma, a Florida doctor calls it--is that on the morning of Nov. 2 many of them actually thought their man was going to win. I won't get into how they could have spared themselves the pain of losing, but I know how they could have avoided those painfully unrealistic expectations. The answer is so-called prediction markets, which this election has put emphatically into the mainstream. You'll now be hearing lots more about them. You may even find you could use them in your own business.

You've heard of prediction markets--online places like the Iowa Electronic Markets, (which operates also as,, and World Sports Exchange, where you can bet real money on the outcome of elections and many other events. What you may not realize is how uncannily accurate they have proven and how well they pre- dicted this election, one of the closest ever by the only measure that counts, electoral votes. Despite Bush's substantial popular-vote majority, his electoral margin was just 34 votes, making this race the second-tightest in 88 years (after the 2000 squeaker). That's a contest that's tough to call.

Yet on election morning the Iowa Electronic Markets continued to predict a Bush victory, as it had since July. Same story at, where you could also bet on the outcome in each of the 50 states. If you had examined the tally that morning, you could have gone to bed early that night, for it correctly predicted the winner in every state. That is, it foretold not only that Bush would win, but also by precisely how many electoral votes.

That performance beat any poll, which is no fluke. Economists who study prediction markets find that on the whole, they predict election outcomes far more precisely than polls do. It seems that lots of people, each trying to be smarter than the others, with real money at stake, are collectively tough to beat--in fact, as far as anyone can tell, impossible to beat.

One's immediate thought is that such power could be applied to more problems than forecasting the next President or Super Bowl champion. That was the reasoning behind the infamous Policy Analysis Market developed under Admiral John Poindexter of the Defense Advanced Research Projects Agency (DARPA), which would have taken bets on various geopolitical indicators and events. Congress and the public went nuts when they heard about it, envisioning terrorists wagering on their own horrific acts, and the project died. But the idea can't be killed and is being applied in all kinds of imaginative ways.

For example, a pair of economists studied a market that was set up within Hewlett-Packard to predict printer sales. The result was more accurate than forecasts done the usual way. Siemens established an internal market to predict whether it would meet its deadline on a software project. Managers said yes. The market said no way, which turned out to be right.

Got any future events you'd like foretold? Certain situations are more susceptible to the power of prediction markets than others. A pair of Stanford economists, Justin Wolfers and Eric Zitzewitz, point out that it's best to ask a highly specific question, such as who will win the World Series. They note that a market may not develop if the question is one on which a few insiders possess extremely reliable information that everyone else lacks, since outsiders would find little motivation to trade. Contracts on the future of the papacy, for example, have generated little interest.

As the accuracy and convenience of these markets become more apparent, especially after their stunning performance in this election, their use will mushroom. That DARPA project may prove unnecessary as publicly accessible private markets pose the same questions DARPA wanted answered. Right now, for example, you can track the markets' odds of Iraqi elections beginning by the end of January (recently 81%), a Palestinian state being established by the end of next year (18%), or Osama bin Laden being captured by year-end (11%).

The lesson increasingly seems to be that on questions of importance or interest, prediction markets will develop--and will speak with brutal honesty. In a bit of "live by the sword, die by the sword" irony, when the DARPA imbroglio erupted, TradeSports immediately introduced a contract on whether Admiral Poindexter would resign. Weeks in advance, the market predicted accurately that he would.

GEOFFREY COLVIN, senior editor at large of FORTUNE, can be reached at Watch him on Wall $treet Week With FORTUNE, Friday evenings on PBS.