On repeated zero-sum games with incomplete information and asymptotically bounded values

2015.

Sandomirskiy F.

We consider repeated zero-sum games with incomplete information on the side of Player2 with the total payoff given by the non-normalized sum of stage gains. In the classical examples the value of such N-stage game is of the order of N or square root of N as N tends to infinity.   Our aim is to present a general framework for another asymptotic behavior of the value observed for the discrete version of the financial market model introduced by De Meyer and Saley. For this game Domansky and independently De Meyer with Marino found that the value remains bounded as N tends to infinity and converges to the limiting value. This game is almost-fair, i.e. if Player1 forgets his private information the value becomes zero.  We describe a class of almost-fair games having bounded values in terms of an easy-checkable property of the auxiliary non-revealing game. We call this property the trigger property and it says that there exists an optimal strategy of Player2 that is piecewise-constant as a function of a prior distribution. Discrete market models have the trigger property. We show that for non-trigger almost-fair games with additional non-degeneracy condition the value is of the order of square root of N.

Keywords: repeated games with incomplete information, повторяющиеся игры с неполной информацией, error term, bidding games, trigger games, asymptotics of the value, игры торгов, ступенчатые игры, цена информации, асимптотики значения игры

Publication based on the results of: Interplay of Axiomatic and Algorithmic Game Theory (2015)