Optimal action + exact EV for any Ultimate Texas Hold'em spot, computed in your browser.
The solver models Ultimate Texas Hold’em as a finite expectimax problem, exactly enumerating all compatible future deals. At each decision state, it computes the expected value of every legal action by evaluating terminal showdowns and averaging outcomes under a uniform distribution over unseen cards. The recommended action is the one with the highest computed EV.
The implementation uses a board-first dynamic program: each complete five-card board is evaluated once, then reused across every compatible flop and decision state. The computation runs entirely client-side, with the state space partitioned across parallel Web Workers and the performance-critical evaluator implemented in Rust and compiled to WebAssembly.