Strategic-Continuation Optimization Outperforms ICM in Tournament Poker

A new paper on arXiv (2608.09586) introduces Strategic-Continuation Optimization (SCO), a policy-construction method for tournament poker that outperforms the Independent Chip Model (ICM). The paper argues that ICM, which converts tournament chips into prize equity based solely on stack sizes, omits crucial factors such as action order, blind obligations, seat rotation, and the elimination pressure big stacks exert on short stacks. These omissions can alter successor-state contrasts that determine optimal moves. SCO addresses these gaps by enumerating current-hand outcomes, mapping them to successor states, pricing those states with continuation values computed from a finite tournament model, and optimizing the current-hand policy. The comparison policy uses the same optimizer but prices successor states with analytic ICM, isolating the effect of pricing. The paper evaluates the resulting policies, demonstrating SCO's superiority in tournament strategy. The research is relevant to game theory, artificial intelligence, and poker strategy, and could have implications for AI in imperfect-information games.