PREDICTIVE POWER

As I watched last week's political equivalent of ``Monday Night Football'' -- this one a match-up between George W. Bush and the (uninvited) Iraqi insurgents -- I pondered public reaction. It didn't take long to find out how the president's speech went over, thanks to an Information Age oracle: the prediction market.

Opinion polls wouldn't be out for days, but the prediction market was buzzing almost immediately, with die-hard political speculators casting bets about whether President Bush would be re-elected and whether the United States really would turn over sovereignty to the Iraqis on June 30.

Like a smaller-scale version of the stock market, the prediction markets aggregate thousands of individual opinions into a single price. But instead of betting on the future of a company, these let you bet on the outcome of current events. Often, these prediction markets turn out to be as good as or better than scientific polling.

Among the questions that have drawn listings on these markets in recent years: whether the California recall would succeed, whether Kobe Bryant would be found guilty and when Osama bin Laden would be captured.

It took the Internet to get these markets going in earnest, and it's no surprise that they have gained a real following, given the information they could provide. Consider some of the successes other financial markets have racked up over time. After the 1986 explosion of the space shuttle Challenger, it took months to officially pinpoint the cause. Yet within hours of the disaster, the stock market had correctly singled out engineering firm Morton Thiokol for punishment among the major contractors.

And just before Hong Kong reverted to Chinese rule in 1997 amid worries over a possible sudden end to its vibrant economy, both the stock market and real estate market rose. The transfer turned out, in fact, to be peaceful.

The accuracy of those markets -- as well as the prediction markets that allow betting real money -- probably results from the fact that they pool the beliefs of many people. These are cold, hard financial transactions based not on what people want to see happen, but on what they really believe will happen.

The prediction market can also be driven by players who have inside information that allows them to bet more accurately. For instance, a staffer for Politician A, with inside information that A is in trouble, can anonymously purchase a political future on the opponent, Politician B. By increasing the demand for B futures, the staffer has driven up the price, thereby reflecting to the world that B is more likely to win. (Can the market be manipulated? Sure, Politician A can ask people to drive up his stock. But that won't happen easily if these markets grow, making it harder for a small group to manipulate it.)

First prediction market

One of the best-known prediction markets is the Iowa Electronic Market, started in 1988 by the University of Iowa. Leading up to the 1996 election, for example, the market offered a ``gamble'' on whether Colin Powell would be the Republican nominee. Given a $1 payoff per contract, someone who paid 25 cents was expressing the belief that Powell had at least a 25 percent chance of making it. (Powell's perceived chances rose when he kicked off his book tour, but dropped temporarily after the O.J. Simpson verdict, possibly because people anticipated widespread racial discord.)

Now after experience with many elections, the Iowa market at www.biz.uiowa.edu/iem claims that its predictions are statistically better than the polls. The Iowa market is regulated by the Commodity Futures Trading Commission (CFTC) and limits total bets per individual to $500. It is run by the University of Iowa's business-school faculty as part of its research and teaching.

A popular market

One of the most popular prediction markets is run by TradeSports (www.tradesports.com) and was founded in 2002. It is based in Ireland, which allows it to take bets in real money without running afoul of gambling prohibitions in the United States. TradeSports serves as a clearinghouse, making 4 cents per $10 contract traded. Its most active future is PRESIDENT.GW BUSH2004, of which about 300,000 contracts have traded hands, representing over $1.6 million.

Another market, NewsFutures ( http://us.newsfutures.com/home/home.html.), uses virtual dollars, but ultimately allows traders to bid those virtual winnings on real prizes. NewsFutures also offers software that enables companies to run such markets internally, allowing people in the know within the firm to place educated bets on such things as how long the company will take to bring a new product to market.

One of the more interesting prediction market trades in recent years allowed speculators to bet on whether weapons of mass destruction would be found in Iraq. At the time of the invasion, TradeSports' traders thought there was an 80 percent chance the weapons would be found. But the market headed down several months before administration officials began to admit that there was even the possibility that they had been wrong.

The market plunge also preceded the first report by then-chief U.S. arms inspector David Kay, who said nothing had been found to back up the administration's claims that the weapons existed.

Another example of clear-eyed wagering was in the California gubernatorial recall election. From August 2003 on, TradeSports participants were pretty certain Gov. Gray Davis would be ousted. The best Davis ever did was in September, when the price indicated that he stood about a 40 percent chance of keeping his job. This dropped quickly to 12 percent thereafter.

Of course, the markets aren't always right. There was a TradeSports contract that would pay $10 if Saddam were captured by a specific date. When his sons were killed in July, the Saddam future shot up to $6.50. But just before his actual capture it was as low as 30 cents.

Still, the markets have done well enough that even the government was interested in trying out its own version. In July 2003, the Defense Advanced Research Projects Agency (DARPA) announced its Policy Analysis Market to harness free markets to ``focus on the economic, civil, and military futures'' of various Middle Eastern countries. No sooner was the market announced, however, than public outrage forced cancellation of the program.

A creative idea

I thought that market was a creative idea, by government standards. After all, one of the enduring mysteries of the Sept. 11 tragedy was a sharp market spike in Wall Street bets against United and American airlines just before the attacks.

There is still disagreement as to whether people were profiting from prior knowledge of the attack, or whether the betting just reflected that airlines were in trouble before the attacks. In any event, smart intelligence experts watching stock and prediction markets should take heed of any such worrisome trends.

What I didn't understand about the DARPA plan was why the government would waste taxpayer money on operating a prediction market when private ones already existed. In fact, in an ironic twist, one of those markets immediately floated a wager paying $10 if the project's main proponent at DARPA, John Poindexter, managed to hang on to his job after the brouhaha. He didn't.

That particular wager gyrated for weeks, reflecting the frenetic pulse of the Washington rumor mill. This, of course, is one of the main points of these markets: to instantly capture the opinions of people on the inside, who might never be interviewed by the news media.

So how about markets for predicting acts of terror?

The upside is that people in the know, perhaps the distant cousins or barbers of terrorists, might whisper what they knew to people who could place the bets to drive up the market -- and alert intelligence agencies that an attack might be in the offing.

The downside, as Stanford Law School Professor Joe Grundfest told Stanford Magazine last year, is that the markets could be misleading. ``Suppose you were a terrorist.'' he says. ``If you wanted to blow up a bridge, you would try to get everybody to look everyplace else by buying the futures that would say, `No, what we're going to do is blow up a power plant.' ''

Nonetheless when the investors are all motivated by making money instead of blowing themselves up, prediction markets hold great promise.

A market on Iraq's future

A few months ago, when I was fed up by the portraits of Iraq being painted by both the politicians and the media, it occurred to me that we needed some sort of ``progress in Iraq'' future. I suggested this to Justin Wolfers, an expert on prediction markets at Stanford's business school, who in turn mentioned it to someone he knows at TradeSports, and IRAQ.TRANSFER.30JUN was born. It pays $10 if the coalition turns over sovereignty on schedule. Otherwise it expires worthless.

Now back to Monday night's political football game. Before the Bush address at the War College, IRAQ.TRANSFER was up 30 cents to about $8.80. This was no doubt because the market correctly anticipated that Bush would hold his course and not announce a change in plans for the transfer of sovereignty. Also, before the speech, the market for Bush keeping his job fell by 16 cents to about $5.20, in anticipation perhaps that he might not come off well in front of the cameras.

The next day, IRAQ.TRANSFER remained unchanged, as there were no revelations in the address. However, Bush, apparently exceeded market expectations with his performance; his re-election future regained almost exactly what it had lost the previous day.

Certainly according to the IRAQ.TRANSFER price of $8.80, it is a good bet that the coalition will turn power over on time. I only wish TradeSports had been more specific as to exactly what was being turned over to whom.

SAM SAVAGE is a consulting professor in management science and engineering at Stanford University, and president of AnalyCorp Inc., a firm that develops executive education programs and software for business analysis. He wrote this article for Perspective. For links relating to this article, see Prediction Markets at www.stanford.edu/dept/msande/faculty/savage.