yes but don't I need to input my opponent's range in a solver and how can it be reliable if I don't know the exact range and the exact way my opponent is going to play
Here’s an example that shows what GTO is and more importantly is Not.
Imagine an old lady who sits at the table and knits. She folds every hand except Aces, which she goes all-in pre-flop.
The exploitative strategy that makes the most money is to raise with any two cards and fold to that occasional all-in (except with AA).
The GTO strategy would be calling with aces, kings, queens maybe 80% of the time, AKs 80% of the time, Jacks 50% of the time, and every hand probably down to 87s some small portion of the time.
In this extreme example the exploitative strategy is going to make WAY more than the GTO one.
Now pretend we have two poker bots. One plays the exploitative strategy and the other plays the GTO.
But the lady switches her strategy. She now goes all-in with every hand she gets.
The GTO bot is now winning big because it’s playing its good hands and folding its bad ones.
The exploitative bot is now folding 100% of hands (except aces) because the player is countering it.
answer : solver can solve for pre-flop ranges in HU spots
thus we can know GTO regardless of what your opponent does, since the gto solution starts pre-flop and isn't based upon any unknown variables/range assumptions
ps i think monker does multiway pf solutions
Last edited by LordPallidan12; 12-08-2021 at 01:20 AM.
Just to extend garicasha’s example (which is on the right track, but incomplete): let’s now suppose the old lady isn’t dumb, and she realizes that her opponent is exploiting her “shove with AA, fold everything else” strategy. She can counterexploit her opponent, perhaps by adding some garbage hands, knowing that her opponent will fold (usually) to her raise. The opponent in turn can exploit this by adding more calls to his range, and the old lady can again make an adjustment to this adjustment.
It might seem that this string of adjustments and counter adjustments might go on ad Infinitum but it actually does not. Both player’s strategies will begin to converge to ones where further adjustments will have an increasingly small effect on the profitability of the strategy. Eventually both will arrive at a point where no possible adjustment can increase their profitability. This is called a Nash Equibrium, which is the technically correct term.
It’s often referred to, though, as game theory optimal (GTO), which is unfortunate terminology IMO because it implies something that isn’t true, namely that it is the absolute best strategy in all situations. It is not optimal in that sense; it’s only optimal in the sense that an astute opponent cannot cause you to lose more money by adjustment of his strategy. In short, GTO is the only strategy that is unexploitable, which is why it makes no difference how your opponent plays.
It’s often referred to, though, as game theory optimal (GTO), which is unfortunate terminology IMO because it implies something that isn’t true
Indeed. The game is still about exploiting opponents. If your opponents have nothing to exploit, the "GTO" move is not to play as you'll all just lose to rake
yes but don't I need to input my opponent's range in a solver and how can it be reliable if I don't know the exact range and the exact way my opponent is going to play
In short, it's not.
That's why solvers are not beginner tools.
If you input wrong/bad ranges or don't node lock correctly, your output could be wrong or even plain bad in some cases.
If I understand correctly, once we input a range, the solver calculates the GTO play against that range.
But if in reality your opponent has a completely different range, you could very well be losing with your "GTO" strategy against that range.
If you however take GTO ranges for both players, you will run into a few spots here and there where your GTO play is -EV.
But overall and over the longer term, it's going to make money even against unknown non-GTO ranges.
Rake has a huge impact on this, but GTO is probably winning enough so that it can beat even the worse rakes in online poker.