Quote:
Originally Posted by statmanhal
We know the definition of GTO as being the same as Nash equilibrium, where heads up, if both play GTO neither has any incentive to unilaterally deviate.
But we don’t know, at least for now, just what that strategy is. So this and future threads present some questions for heads up play that, for me at least, can clear up some things:
Q1. Is my definition correct?
I agree with your definition and believe it's correct in general game theory.
Quote:
Originally Posted by statmanhal
Q2. Does the following make sense for a way to describe and implement a GTO strategy?
We define an information set as all the factors known to a player at some decision point (everything except villain’s cards). We then describe a GTO strategy as a rule book stating what action is to be taken for every possible information set.
I think this is limited depending on what constitutes a unique information set and what the out put action looks like.
I am specifically thinking of situations where actions (check, bet, raise, call, fold) for the same information set (specifically same hand) because two actions (check, bet, raise, call, fold) might occur with mixed frequencies because they have the same EVs otherwised known as a mixed strategy.
If your definitions allow for this then I see nothing wrong with the way you choose to describe these things, though, some game theory terminology already in use may be more beneficial since it is accepted outside of poker.
Quote:
Originally Posted by statmanhal
Q3. Assume a poker game only allows a small number of bet sizes. Is GTO still applicable?
I am going to take some liberty with your question and assume that you don't mean "is a nash equillibrium possible with limited number of bet sizes" and what you are asking "is the nash equillibrium of a poker game with limited bet sizes equivalent to the nash equilibrium of the game with variable bet sizing".
I don't know that this question is answered or not.
There is work in subgame equillibriums but that seems to be relevant only to perfect information games.
I would have to err on the side of caution and say the solutions would be different.
Quote:
Originally Posted by statmanhal
If so, if you ran enough solvers a long enough time, with lightning speed computers, you could eventually achieve such a rule book. True or false?
True - I think some academics and maybe even commercial solvers are doing this.