Private Information, Overconfidence and Trader Returns in Prediction Markets

In lab experiments on the value of information in financial markets, groups of “insiders” are randomly chosen to receive perfect information. However, in typical (non-experimental) financial markets, investors often engage in extensive fundamental analysis, a process which may result in over-confidence in one’s private information. In this study, we examine trading volume, prices and trader returns in a set of four real money prediction markets where the values of securities are tied to a movie’s box office performance. Before the markets opened, every trader submitted a detailed forecast of the movie’s future performance. Therefore, all traders have self-generated private information, the accuracy of which can only be known ex-post. As expected, the volume and timing of trading were consistent with over-confidence. In three of the four markets, contract prices were consistent with the prior information equilibrium. In those three markets, traders whose forecast was associated with the winning contract had significantly higher returns than traders whose forecasts suggested that another contract would pay off. In the other market, there were no significant differences in returns across trader groups. This research suggests that when traders are overconfident and trade accordingly, there can value to being better informed if the information is accurate.