A little while back, Akhbar looked at the poor performance of the prediction markets for presidential politics this year. The basic problem is that the market lags the polls, so a tool that was supposed to give us more predictive power seems to give us no new information. Here is a more detailed look, via FlowingData.
Short recap: the authors seem to think that inTrade outperformed the polls. Looking at the data myself, it seems neither the polls nor inTrade were clearly superior. The inTrade market bit hard on McCain’s postconvention bounce, for instance. Much more than did the polls.