Bluffing and Strategic Reticence in Prediction Markets

TL:DR Binary prediction markets are zero sum and not incentive compatible. Large players with inside information are not incentivized to trade (and therefore give the market that information) until the most opportune time, thus negating an important aspect of the efficient market hypothesis (that markets immediately incorporate all relevant market information). The existence of ‘noise traders’ can supply the incentive to trade for insiders (since the noise traders arguably offer the opportunity for profit).

The paper discusses the ability of using an AMM to provide that automated ‘noise trader’ who takes a loss from ‘rational traders’.