Quote:
Originally Posted by mikesfans
Does pokersnowie give a GTO solution for the spots? and can we alter opponents strategy/betsizings to let snowie come up with a counter strategy?
It provides what it thinks is the optimal strategy against all the possible strategies it has 'trained' against. In that sense it
approaches a GTO solution (poker is far too complex for a full solution), but the problem is that in "defending" against some bad strategies, it can lose (or at least lose EV) against good ones. A related problem applies to solvers like Pio. Pio can give you a near-perfect solution if you know the other player's range and strategy, but if that player
doesn't play how you expect him/her to act, the 'solution' isn't exactly useful. (It makes no exploitative sense to have a balanced river shoving range of value bets and bluffs if the actual villain calls 100% instead of the 48% or whatever Pio says they should call with, for example).
Unfortunately, Snowie can't be adapted with user-generated ranges, and it also gets "confused", by strange betsizes or action sequences. (It plays "weirdly" against limps or mindonks, for example). It can only make best-guesses based on what it's seen in the past. I find it very useful for pre-flop and the early streets, but its river play is sometimes quite alien in relation to how real people play, so its suggestions for that street are less useful. What I like about it most is its speed. Its "solutions" might not be perfect/optimal, but at least they are instant. Pio/GTO+ take too much time for my purposes, but other people (with faster computers than mine) have learned a lot from grinding hundreds of study hours with them.