Discrete-time stopping games under the risk probability criteria
Аннотация
This paper investigates discrete-time stopping games under the finite-horizon and infinite-horizon risk probability criteria. In this game model, each player can influence both the transition laws and the termination time. Unlike previous studies that focus on expected reward criteria, we aim to minimize or maximize the probability of accumulated costs/rewards up to the termination time controlled by the players, which reflects the reliability of each player’s income. For the finite-horizon risk probability criterion, we establish a backward iteration method and prove the existence of the value of the game. Furthermore, we derive stopping rules and strategies that form a saddle-point equilibrium. For the infinite-horizon risk probability criterion, we construct a sequence of value functions and demonstrate that its limit corresponds to the value of the game. Finally, we construct optimal strategies and stopping rules, which together form a saddle-point equilibrium in the infinite-horizon setting.