What is GTO?

The longer version

What each scene is saying, spelled out. Tap a note to jump to its scene.

1. 0:00What is GTO? The letters stand for Game Theory Optimal.
2. 0:03It is a strategy that can’t be exploited. Take a pot-sized bet on the river: if you bet two value hands for every bluff, your opponent risks one pot to win two, so they need to be right one time in three, which is exactly how often you are bluffing. Calling and folding earn them the same. They have no reply that wins more.
3. 0:08It mixes. The same hand in the same spot takes different actions at fixed frequencies: sometimes a big bet, sometimes a small one, sometimes a check. If you always did the same thing with the same hand, you would be predictable, and predictable can be exploited.
4. 0:13How does a solver find such a strategy? Step one: it maps every decision. Check or bet, then fold, call or raise, for every bet size and every card that can come.
5. 0:17Step two: it plays that tree against itself, over and over. After each pass it measures regret, meaning how much better another action would have done.
6. 0:20Step three: it shifts toward the actions that would have done better, and repeats. With each pass, the strategy becomes harder to exploit. This method is called counterfactual regret minimization.
7. 0:24What’s left is an equilibrium: no single change makes either player better off.