Quote:
Originally Posted by MarkD
you are arguing to argue I think.
I am not sure what monikrazy used to generate his sim, as this is a tough spot to simulate due to the straddle and limper (solver don't limp). I did a sim with CO open BTN 3bet 150bb stacks and we get similar pot size. It aggrees we are checking back flop and then it's a different hand. Let's say hero bets 1/2 pot (only option I can check) villain calls, turn checks through, villain bets pot on river now I get a different result where solver is folding 75%, calling 16, raising 8 of range. But now I see, Monikrazy is looking at the specific hand within the strategy after taking this line (which solver couldn't get there with) and now yes, it wants to fold 75% and raise 25%, but it is not expecting to get called by worse one pair hands at all. This is a bluff raise with a negative EV. Solver is not calling with worse one pair hands. It is folding 55% to our all in and calling 45% and the percentages make sense because when it bets pot it was polarized between mostly nuts and bluffs. For instance it's folding almost all AQT8ss hands on the river (bottom straight).I can't find a single example of a one pair hand that he would call our raise with. He is calling with the nuts, and very rarely with very specific combinations of T8.
I am absolutely not arguing for the sake of arguing, I am arguing because I.... disagree.
Surely it should be fairly easy to compare outputs? All you need to do is establish the inputs he used, and if you use the same and they come out with different answers, beyond what should be a fairly small margin for error, well, what does that say about them?
I am not against the solver. I am against how it is commonly used. In my mind, the way to use the solver is to have it show its working, and then we use that working to guide our own ideas as to how to play. This is not what is being done by most, it seems, who just want a simple answer to the question: is this a +ev play or not? If we could extrapolate the rules for why the solver does or doesn't like certain plays, we could improve our games. That we (I'm sure some do do this) don't have verification that different solvers agree is fairly worrying. I see a huge number of people accepting solver outputs as gospel, and then bashing people who question the answers. What if they're wrong? They could be, right? How do we know they're not? Wouldn't it be disastrous if we took a play that's +ev against all human populations and replaced it with an unexploitable (i.e. breakeven) play against bots? Where's the protection against that possibility? History is full of tech advances that make our lives better, but 100x as many techs fail, and we ignore that history at our peril.
Anyone want to buy my collection of NFTs?