Minimax regret in the 11-20 money request game

Arad and Rubinstein’s 11-20 money request game nicely triggers level-k reasoning. Yet we show that mixed-strategy minimax regret, in a general class of money-request games, mimics a level-k reasoning, at least if the number of players is supposed to decrease in the depth of reasoning. We also show that, in this class of games, the minimax regret probability distribution is the exact reverse of the mixed-strategy Nash equilibrium distribution, and that minimax regret leads to a larger expected payoff than the Nash equilibrium payoff.