Note to users. If you're seeing this message, it means that your browser cannot find this page's style/presentation instructions -- or possibly that you are using a browser that does not support current Web standards. Find out more about why this message is appearing, and what you can do to make your experience of our site the best it can be.

Science 30 July 2004:
Vol. 305. no. 5684, p. 603
DOI: 10.1126/science.305.5684.603b
Prev | Table of Contents | Next

Random Samples

A century ago, betting on the presidential election was a popular quadrennial pastime. Now, thanks to the Internet, so-called prediction markets are booming again--and getting the attention of economists.

Two articles published this month in the Journal of Economic Perspectives suggest that betting markets can outforecast the polls. In one, Paul Rhode and Koleman Strumpf of the University of North Carolina, Chapel Hill, relate that from 1884 to 1940, when election wagering was wildly popular, bettors correctly picked the winner in every year but 1916, when Woodrow Wilson beat Charles Evans Hughes with a last-minute upset in California. After 1940, because polls were perceived as more scientific, the gambling markets faded from public consciousness, says Strumpf.

But the current betting markets--led by Web sites such as www.intrade.com, tradesports.com, and fairbet.com--may outperform opinion polls, according to a paper by Justin Wolfers of the University of Pennsylvania and Eric Zitzewitz of Stanford University. In the last four presidential elections, the University of Iowa's Iowa Electronic Market has averaged an error margin of ±1.5% in the week before the vote, compared with ±2.1% for the Gallup polls. More recently, traders picked John Edwards as John Kerry's running mate 2 months before Kerry did.

One reason markets are better at prediction, Strumpf says, is that polls are time-bound snapshots of the public's mood. Thus, for example, candidates always surge in popularity after a convention, but canny bettors know it won't last.




To Advertise     Find Products

Science. ISSN 0036-8075 (print), 1095-9203 (online)