Solver

A solver is software built to compute the game theory optimal strategy for a specific poker spot. You give it the two ranges, the board, the stack and pot sizes, and a menu of allowed bet sizes, and it searches the resulting game tree until it finds the strategy pair where neither range can profitably deviate.

Under the surface, most solvers use an algorithm called counterfactual regret minimisation. It plays the spot against itself millions of times, tracking how much better each action would have done, and gradually shifts the strategy toward whatever it regrets not doing more. Over enough iterations the average strategy converges on equilibrium.

The output is two things for every hand in the range: how often it takes each action, and the expected value of each of those actions. That is almost always a mixed strategy rather than a single answer, because balance requires some hands to take more than one action across the many times that spot comes up.

Solving is compute-heavy, since the game tree for even one spot is enormous, which is why solved output is usually precomputed for study rather than generated live at the table.