Beginner's Poker Course Part 4: Game Theory Optimal (GTO)
We are all well aware now of the concept of EV, and we've been taking a look at our first charts, which showed correct play for several preflop situations.
That is great. Because just by understanding EV, you can start playing poker and develop an intuition about what the highest EV actions are. It is already a solid baseline to build your game upon.
The concept of GTO (Game Theory Optimal) is quite a bit more advanced mathematically, however. And I want to be honest, there are very successful poker players who don't even care much about this concept, yet are still able to pick the highest EV actions 'all the time', and win.
'Now stop the long intro!' You may demand of me. And you're right. Let me start by giving a definition and explain to you why it is, or can be, important.
My own definition of GTO would be like this:
Playing a strategy that is best against an opponent that plays the exact same strategy.
The exact same strategy? Well, you see, if we strictly follow the charts from the last article, and all others from GTOWizard, then let's assume there are perfect charts for every possible situation in poker. What if our opponent does the exact same thing? What would happen?
Ever Changing Strategies
I understand if that is still a bit abstract, but at least we get a clear picture now! And it's not for nothing that I chose bots here. Because in real-life games, when would it happen that two humans are playing the exact same strategy? It is safe to say that that is (close to) never.
But why then still care about GTO? Well, just to reiterate, GTO stands for Game Theory Optimal. Game theory is simply the study of strategic decision-making, and 'optimal' of course means the best or most favorable. You might then conclude that playing GTO and following all the charts from GTOWizard all the time is the optimal thing to do in poker. Not quite, actually.
Trust me, I don't want to confuse you, but it can be. You see, GTO really just assumes that all players in the game are perfect, 'Gods', always being able to pick the perfect action, the one with the highest EV. Or call them super bots, perhaps; because you know that nowadays AI can achieve superhuman, or God-like intelligence, and performance in games.
Well, we are playing poker against real humans in our beloved online poker rooms, and luckily that is the case. Bluntly stated, GTO would only be best if we ourselves were playing against these superbots, or supergods. That is game-theoretical optimal poker!
It is truly safe to say that every human plays differently. Consider all the situations that can arise in poker—and remember, there are countless! Given all these scenarios, no two humans take the exact same actions for all of them.
What is important to comprehend is that we should play differently against each of them; 'optimal poker' looks different against every single one of those humans! Every unique opponent we could encounter in the real world (a human that is), requires a different strategy on our part!
I would then like to draw a clear distinction between 'optimal poker' and 'game-theoretical optimal poker'. Simply put, it's the difference between playing against humans and playing against gods.
Gods are playing so perfectly that they never have to change their strategy, they are all playing GTO against each other. Real humans' strategies, on the other hand, are ever-changing; we are ever-changing in general, whereas Gods are not.
We could then say that GTO is somewhat of an illusory concept in poker (I'm not calling God illusory here); the illusory part being that it assumes opposing strategies never change. Because that's what GTO tells us: a set of strategies that are so refined they are perfect at all times, and perfect against each other. Perfect strategies playing perfect strategies; and that means there is no need to ever change them.
Human strategies change; they are highly dynamic, whereas GTO strategies are static. A formal definition of GTO is then:
GTO is a strategy where no player can benefit by changing their tactics if others keep theirs unchanged. It represents a 'Nash Equilibrium' where each player's strategy perfectly counters the others, making the full set of NE strategies unexploitable when everyone plays perfectly.
Nash Equilibrium is a new term being thrown in here, I know. But, at least in poker, GTO and NE are used interchangeably. Though nowadays, most people just use GTO, probably because it has a better ring to it.
Why Care About GTO Anyway?
Nash Equilibrium in poker is really about the same set of strategies those 'gods' would play. Strategies that, again, never change; because there is perfect balance between them (hence 'Equilibrium' from the NE term). We would say, 'these gods are playing in a Nash Equilibrium' or 'these gods play GTO'. From now on, we will keep using the term GTO, however.
So if GTO strategies seem so unattainable for us humans, given our imperfect and limited minds, why is this such a big thing in poker, anyway?
You see, even though GTO is not necessarily something to (always) strive for, it can provide a really solid baseline to build your game upon. One reason could be if you're playing against a player who outclasses you. You are following a beginner's course, so let's assume this can happen in the games you're playing.
The thing is, even though playing GTO is not optimal against real humans (because they are not gods also playing GTO), it also assures that you do not lose against whatever strategy you're up against.
Playing GTO makes you unexploitable.
Why Not to Care About GTO?
Caring about GTO really comes down to becoming unexploitable, as mentioned earlier. But if you are not playing against God or someone better than you, there is some real merit in not caring about GTO at all. There is one caveat, though: you would have to be really adept at picking the highest EV action.
It feels nice to come full circle here, doesn't it? Poker is all about picking actions that have the highest EV, and that remains true even after learning GTO. What is important to understand now, though, is that GTO strategies also do that, if and only if they are playing against other GTO strategies.
What that also means is that GTO strategies do not always pick the highest EV action (although they often would) against humans who play strategies that vastly differ from GTO. That might be just the most important thing to remember about GTO and truly highlights its limitations.
This makes GTO strategies suboptimal against humans and optimal against gods, bluntly said.
Exploitative Play
Then what is optimal against humans, if not GTO? This is what's called exploitative play. The optimal strategies against humans simply involve taking the actions with the highest EV. And even though GTO does the same thing when playing against other GTO strategies, it does not change its strategy when the strategies of humans change.
You see, the EV of our actions depends heavily on how our opponents react to those actions we take. GTO doesn't adapt to our opponents, but we should.
That means if our opponent folds too many hands, we should raise more aggressively to try to pick up the pot—just an example. And perhaps we would have more incentive to pull off a big bluff on the river, in a big pot, figuring he would lay his hand down often enough for that bluff (like an all-in) to be the highest EV action at that point in the hand.
There are countless little adjustments we could and should make, given our specific opponent's tendencies. For instance, we have:
Our opponent has:
We see a flop of:
This is a terrible board for our hand, and we should likely check, at least most of the time, with the ace five here.
We don't exactly know our opponent's hand here. But imagine we have observed him or her for quite a while now. Given what we've seen of him until now, we currently predict that if he does hold a hand like the five four of spades, he will fold to a small bet (even though he shouldn't).
So we believe our opponent folds too often, even a reasonably strong hand like a straight here, due to the fear of a flush, let's say. And even though the GTO play would be to (mostly) check our hand in this situation, we might opt to now (always) bet instead; the EV of a bet just became higher than that of a check because our opponent folds too often.
Our opponent is not playing GTO with his flawed strategy of folding his lower straight to a small bet on the flop. That means we should deviate from GTO in turn; we should bet with a hand we 'normally' (in 'GTO land') wouldn't. The opponent folding the straight there is his weakness, and we're exploiting it.
The basis of exploitative play is then:
To pick up on how our opponents are playing differently than GTO, and exploit it properly.
This shifts back the notion that GTO is not so important to it becoming very important:
How would we know how to exploit our opponents if we don't know how they are playing differently from GTO? How would we know if they are calling or folding too much, for instance; and what are those limits, when are we doing something 'too much' in poker?
That actually makes GTO a very solid baseline to lean on. It is funny; we shouldn't play it against our opponents, but we could (and perhaps should) use it to effectively exploit them.
Well, should we use it? It can be a very handy tool. But remember, at the beginning of this article, I mentioned that there are poker pros who throw it to the wind; they are exclusively focused on exploiting their opponents. They may claim it's not useful, but what they really mean—something I mentioned earlier—is that we should never really play GTO against human opponents.
But useful it can be, as we just noted; knowing GTO can be a solid foundation to base your exploitations on. Though I have to tell you; GTO tools came onto the market around 2015, and I have been playing profitable poker long before that, only starting to use these tools since around 2019. Many others have been successfully exploiting their opponents without the use of modern-day GTO tools.
Perhaps, then, this is one of those classic 'difference between science and art' dichotomies. GTO is a science, and exploitative play is an art: an art that can be mastered through raw playing experience, mastering different kinds of heuristics and concepts off-table for use in-game, and, well, knowing baseline GTO play. Gotcha!
GTO vs. Exploitative Play
There has been a debate in poker about which is better: GTO play or exploitative play. I'd suggest you ignore this debate and just focus on where the value lies.
Exploitative play focuses solely on taking the highest EV actions against human opponents, who each have their own flaws. GTO poker is a theoretical approach that can be seen as a defense against such exploitation.
Because if you don't play GTO, you can theoretically be exploited. If your opponent doesn't play GTO, you can exploit them.
The value in exploitative play is clear: if we take the highest EV actions every time, we win the most money. It's really as simple as that.
The value in GTO is more nuanced, you could say. We talked about using it as a foundation from which to develop your exploits; though we also noted that this route isn't the only one, as poker pros have successfully exploited opponents for years without knowing GTO strategies.
As long as you now understand the distinction between the two:
GTO makes you unexploitable but misses the opportunities arising in-game to exploit your non-GTO playing human opponents.
Exploitative poker is all about deviating from GTO in the right way, the most profitable way, and again, that comes down to choosing the highest EV actions.
GTO is playing poker in a completely balanced way, and given the complexity of the game, it's unbelievably hard to achieve. We can work on it, though; we can study and work on making our game more balanced and build solid defenses against potential exploitations from (better) players. That is where the 'other' value lies.
This makes GTO both about exploitation and building defenses against it. We can 'know' GTO and intelligently deviate from it, enabling us to 'optimally' exploit our opponents. On the other side of the coin, we can 'make sure' that highly skilled players, who may be better than us in GTO and/or exploitative play, can exploit us only minimally.
With that being said, I feel it's time to move out of the theoretical realm and back to actual play, to usable strategies. After all, you're here to learn to play poker, and I understand that. I just wanted to build this foundation because I believe it's important to have a deep understanding of how the game works. I hope I have given you a solid grasp of the concepts of EV, GTO, and exploitative play. It will be helpful with our upcoming journey of evaluating a lot of interesting spots that can come up in the game.
So let's get practical soon. In the next article, we will explore flop play. There will be more hand examples. We will cover the notion of mixed strategies and how that relates to GTO. Hope to see you there!
Do you see the value in learning GTO?