GTO Opponent's Range & Strategy: I'm confused

I'm not sure if there's some missing context that redeems these quotes, but as stated, they mischaracterize GTO.

You don't need to predict your opponent's actions or starting range in GTO. But you do need to plug in GTO-approved ranges to ensure the output is unexploitable.

You can plug in assumptions about your opponent's range or strategy. But this is an exploitative exercise. You're trying to punish their mistakes not solve equilibrium.

Thanks, that's exactly what I'm thinking, too. I think I'm seeing what's going on. There are "solvers" (Pio, PeakGTO, etc) that haven't worked out solutions, so they need inputs. For example, from https://pokercoaching.com/blog/poker-sol...

"Step #2 – Input the Ranges
As we already explained, poker solvers don’t actually know how to play poker, and they don’t know the first thing about which hands you might want to play from any position. Therefore, it will be necessary for you to input your own range and your opponents’ ranges into the solver.

For your own range, you should enter a realistic range that you know you would play in a given scenario. By all means, be honest with yourself when doing this part.

Next, you will want to input a range for your opponent or multiple players. In this case, you will need to use experience or some premade hand ranges that best fit population tendencies.

Inputting the right-hand ranges is equally as important as creating a reasonable game tree, as both will have a big impact on the final outcome of your solution."

I guess this is what that book is referring to. So, with the calculation solvers, you find the unexploitable strategy versus your opponent's range and strategy. However, the value of this exercise seems quite limited. I can see how you could glean gameplay concepts from such a thing, but that would be its limits because the output in terms of gameplay feel like a bit of nonsense to me. Because if I can get his range and strategy, what on earth am I doing thinking of anything except exploiting?

So I guess something like GTO Wizards library of solutions is the "real equilibriums" - what an equilibrium for both me and my opponent looks like at a given branch.

Am I missing anything?

Edit: After some research, I'm seeing libraries for sale for PioSolver, so that makes more sense to me...

The likely explanation is that if a solver needs assumptions about your opponent’s range, that means it hasn’t fully solved preflop and instead is solving a sub-game that only begins when your opponent’s range is what you tell it.

This is probably an incompleteness in the solver. To my knowledge, NL is not solved or even close to it yet, so this makes sense to me.

I say the game is not close to solved mainly because preflop is not close to solved. If solvers had the power to solve preflop then probably they could solve almost all of the game or at least come close.

My understanding (though I’m sure others’ are better than mine) is that solvers are solving a sub-game of the overall game according to the parameters you feed it. So it may be the case that if an opponent takes an action outside of the parameters it could beat the solver strategy, but I am guessing that solver-backed strategy has been refined enough that this is very difficult.

There are plenty of preflop solvers. GTO Wizard, HRC, Monker just to name a few.

You're basically hitting the line between Nash equilibrium and exploitative play. GTO gives you the floor no matter what villain does, but when you plug in their actual range the solver is no longer finding equilibrium, it's finding max EV against that specific population, which should sit above the floor. The catch is that your read on their range has to be accurate or you're just solving for a ghost opponent and the output is garbage.

When talking about PF I think you may be talking about different things. Preflop is indeed fully solved in the sense that we know the Nash Equilibrium strategy (a MUCh better term than GTO, BTW, since in many situations there is absolutely nothing optimal about GTO). Think of a heads up NLHE game. I tell you that itÂ’s dumb to play anything but AA and that when you do have AA you should shove. If I am really playing that strategy, (which is very much non-GTO) you could counter with a lot of different strategies and make money, but the optimal one is likely to minraise with ATC and fold to my shoves. 220/221 hands you win the blinds; 1/221 you lose 2bb. I soon see though that my original idea was dumb and add KK to my shoving range - this reduces the EV of your optimal strategy you were using. Hence, by definition, that was not a GTO strategy. You can counter-adjust and gain EV. I can adjust to counter your adjustment, and so on back and forth. It may seem like we could go on forever with these adjustments and counter adjustments, but game theory math tells us that isnÂ’t true. We will eventually converge to a point where we are both using a strategy that will not permit our opponent to gain EV regardless of what adjustment he makes. This is the Nash Equilibrium, and it is known for preflop play (at least to a high degree of approximation).

When some people say preflop is not solved they mean that we don’t know the optimal strategy against all possible preflop ranges that an opponent might play. That’s likely true since there are nearly infinite possible opponents strategies to consider. This isn’t what we mean by “solved”, though. We mean that a Nash Equilibrium strategy is known.