Prediction Markets: the Virtual Super 12 and 2004 U.S. Election

Abstract:     
This paper speculates about reasons why prediction markets work so well. Some common political arguments opposing prediction markets are discussed with reference to predictive information sources surrounding the 9/11 terrorist attacks (in hindsight). The paper then analyzes two internet prediction markets. First, the virtual super 12, where over 100,000 participants predict the results of rugby games. The wisdom of crowds is confirmed, as it is found that the consensus pick performs in the top 1% of participants. Next, Tradesports contracts on the US presidential election are analyzed. On the day of the election, Tradesports' prices reacted to early (misleading) exit poll information but were quicker to dismiss these results than media coverage. The prediction market demonstrated that Ohio was the key state that drove the election result.

Keywords: Prediction, efficient markets

JEL Classifications: G13, G14, G22

Accepted Paper Series