Gaming prediction markets: equilibrium strategies with a market maker (Q603926)

The paper is oriented to strategic behavior of risk-neutral and incompletely informed traders in a prediction market. The attention is focused on the interaction between traders and logarithmic market scoring rule in a Bayesian extensine-form game with incomplete information, where finite-stage and infinite-stage game models are considered separately. The main results show that different information structures can lead to radically different strategic properties. Namely, in the case of conditionally independent signals, the truthful betting is the unique perfect Bayesian equilibrium, meanwhile if the signals are unconditionally independent, the perfect Bayesian equilibrium need not (in the case of infinite-stage model, it cannot) exist in such a game.