Incentive-compatible Online Opinion Polls

The basic idea behind such information markets is that the pricing scheme of the market encourages the participants to buy or sell shares in accordance to their private information. Consider, for example, the security that pays $1 if the winner of the 2008 US presidential elections is democrat. An agent who privately believes with probability p that US will have a democrat president in 2008, will find it rational to buy more shares until the price reaches p. Numerous experiments and studies have shown that the price in such prediction markets converges to reflect the true information available in the market, despite game theoretic negative results (e.g., the “no trade” theorem). Moreover, prediction markets have been known to consistently outperforms other traditional prediction tools. However, one important shortcoming of prediction markets is that they must be linked to a publicly observable event that is precisely defined. This strong requirement can create problems even for the simplest prediction markets: for example, had Al Gore won the recount of votes following the 2000 US presidential election, the security associated with G.W. Bush would have still paid