The end of Intrade

With a cryptic note to customers, the betting market Intrade has announced it is closing down operations. The site's traffic was badly hurt after it was forced to jettison its U.S. customers over a pending lawsuite by the U.S. Commodity Futures Trading Commission. Since its founding in 1999, the Irish site has provoked quite a bit of debate over its supposed predictive powers.  In 2007, for instance, a group of prominent scholars including Daniel Kahneman, Robert Shiller, and Justin Wolfers argued that the United States should deregulate markets like intrade as they have the "potential to stimulate innovation in the design and use of prediction markets throughout the economy, and in the process to provide information that will benefit the private sector and government alike."

InTrade-like prediction markets certainly seem to have a future provided they avoid the attention of the wrong regulators. Competitor Paddy Power, which also offers non-sports markets on outcomes like elections and the Nobel Peace Prize, seems to be raking in record profits. John Authors argues that such markets still have utility in politics:

In 2004, Intrade correctly predicted the vote each of the 50 states made for president. In 2008, Intrade’s one mistake was to fail to predict that Mr Obama would pick up one electoral college vote in Nebraska. Last year, it showed it was not flawless by getting Florida wrong. Mr Obama held the state, but Intrade had put a 69 per cent chance on a Republican victory there.

The record, however, is far more successful than that of polling, which helps explain why the concept has been around for a while. The University of Michigan economist Justin Wolfers, who has made a special study of the field, found records of election betting in Wall Street going back to 1884. The average betting turnover per election in the early years of the 20th century, in 2002 dollars, was $37m. Over history, they performed better than the Gallup poll, with an average error in presidential elections of 1.4 per cent against more than 2 per cent for Gallup.

They are also surprisingly hard to manipulate. When the long-shot presidential contender Pat Buchanan encouraged his supporters to buy Buchanan futures on the electronic futures markets run by the University of Iowa, they failed to move the markets for more than a few hours – once the price was too high, sellers emerged and pushed it back down again. And again, attempts to manipulate the markets go a long way back; Mr Wolfers found that the Tammany Hall machine that controlled New York politics at the turn of the last century attempted to skew the markets.

Fine, but the fact that Intrade outperformed the likes of Dick Morris and Karl Rove isn't all that impressive.  The 2012 election may have been the most extensively prepolled political outcome in history. Taking an unbiased view of the numbers in the weeks leading up to the election, it was clear that the smart money was on an Obama victory. The main uncertainty was which swing states he would take, which by missing Florida, Intrade failed to predict. 

When the number of decision-makers on a given political outcome is even smaller, and information on their preferences is less available, Intrade was even less useful.  Intrade notably blew it on the Supreme Court's decision to uphold the Affordable Care Act, giving the bill a 77 percent chance of being struck down only two days before the decision.

As Barry Ritholtz has tracked, Intrade has also badly missed other decisions including the 2004 Iowa Caucus,  and the 2005 Michael Jackson trial. As Ritholtz put it, "Futures markets are really a focus group unto themselves: When the group is something less representative of the target market, they get it wrong with alarming frequency."

The behavior of smaller decision-marking groups like juries, the Supreme Court or even Iowa caucus-goers are much more difficult to predict than national elections where polling data -- if you aggregate enough of it -- is a pretty good predictor of the outcome. Intrade was most useful in the situations where we didn't really need it.