>> Bots that play purely mathematically optimally on expected value aren’t effective or interesting. Interesting is up to you, but effective is definitely wron

>> Bots that play purely mathematically optimally on expected value aren’t effective or interesting. Interesting is up to you, but effective is definitely wrong. ICM-perfect bots crush small tournaments, which do not take into account opponent behavior - merely modeling the gamestate. The faster the blinds and the smaller the stacks, the better, but even normal structures get killed by these so-called "expected value" only bots. Game Theory Optimal (GTO) attacks are incredibly effective at all levels of the game. The AI need not incorporate opponent feedback to be a winner. It can make it better, but it is not at all required.