A method that repeats self-play to approach equilibrium.
It repeats self-play, raising the frequency of actions with large regret. It is what runs inside a solver. Example. If Fold kept turning out better than Call in a spot, Fold gets chosen more often from then on.