"Unexploitable" Is Not "Optimal"

There is a word that has been doing damage to your game for years, and the word is optimal. You have been told GTO is optimal so many times, by so many sources, in so many videos, that the equation between the two has gone invisible to you. You hear GTO and your brain reads the right way to play. You hear optimal and your brain reads GTO. The two are bonded together in the modern poker vocabulary, and the bonding, in a precise mathematical sense, is wrong.

I want to be careful before I start, because this is going to sound at moments like an attack on solver work, and it is not. The tool is good. The framing is bad. If you can keep the tool and update the framing, your play improves. If you keep the framing without examining it, you keep bleeding the way the framing has been making you bleed. So let me say the thing plainly, and then spend the rest of this dislodging it: GTO is not optimal.

What GTO actually is

GTO stands for "game theoretically optimal," which is the marketing translation, or Nash equilibrium, which is the technical name. In a two-player zero-sum game like heads-up poker, a Nash equilibrium is a pair of strategies — one for each player — such that neither player can improve their result by changing their own strategy while the other keeps theirs the same. That is the entire definition. It's a fixed point. Both players are doing the best they can given what the other is doing, and neither has any incentive to move.

That definition is beautiful, and it's real. Nash won a Nobel partly for proving these equilibria exist. The modern solvers are the practical fruit of decades of that work. None of it is fake. But look at what the definition is — and what it is not. A Nash equilibrium is a defensive property. It tells you what to do if you assume your opponent is also playing the equilibrium. The moment your opponent deviates from it, your equilibrium strategy is no longer the best response to them.

Unexploitable is not maximally profitable

Here is the swap the word "optimal" has been hiding. GTO is unexploitable, and unexploitable is a different thing.

Unexploitable means your opponent cannot do better than break even against you over a long sample. That's the whole guarantee. Notice what it does not say: it does not say you can't do better than break even against them. You almost certainly can. The equilibrium is leaving money on the table whenever the opponent deviates — and the opponent always deviates, because real humans are not playing the equilibrium.

So the central fact is this: GTO is unexploitable, not maximally profitable. Against an opponent who isn't playing GTO — which is every opponent you have ever faced — the GTO strategy is leaving expected value on the table. You can do better than GTO against any specific opponent. The thing that lets you do better is exploitative play, and exploitative play is, by definition, a deviation from the equilibrium. (I pull these two apart in more detail in GTO vs. Exploitative Poker.)

If you take one thing from this: the word optimal, in the context of poker against real opponents, should point at exploitative play, not GTO play. The industry trained you to use it the other way. The gap between what you've been aspiring to and what you should be aspiring to — that gap is the leak.

Feel it at the extremes

The abstract version won't land until you see it applied, so picture one opponent who calls every single bet you make, every street, no matter what. Always call. What is optimal against him?

It is not GTO. The optimal play is to never bluff and value-bet anything with equity over his calling range. That generates enormous EV against this specific man — far more than the equilibrium would. The equilibrium would still have you bluffing at some nonzero frequency on certain textures, because the equilibrium is built to be unexploitable against an opponent who can fold. This opponent cannot fold. Every bluff is a pure loss. The equilibrium is wrong against him.

Now flip it. An opponent who folds to every bet, no matter what. The equilibrium still has you value-betting at some frequency, because it's built to defend against someone who can call you light. This opponent never calls. The value bets are wasted; he's already gone. The optimal play is to bluff every time, because every bet wins the pot uncontested. The equilibrium is wrong again.

Most real opponents live between the extremes. They call a little too much, or fold a little too much, or bluff at the wrong frequencies in specific spots. The equilibrium does not adjust to any of it. It plays the same against every opponent in every spot — it is, in a real sense, blind to the human in front of you. Exploitative play has eyes. It asks where this specific person is wrong, and how to take advantage. The looking and the taking advantage are the actual edge. They are not in the equilibrium. They are in the deviation from it.

The part that makes it worse: the rake

The first fact is bad enough on its own. The second one makes the picture much worse, and almost nobody emphasizes it: in a rake environment, GTO is theoretically a losing strategy.

A Nash equilibrium between two zero-sum players assumes the game is zero-sum — every chip stays with the players. The real game isn't zero-sum. It's negative-sum, because the house takes rake out of the pot before it reaches anyone. The equilibrium edge between two equilibrium players is zero. The rake is greater than zero. So two perfect GTO players grinding each other for a long enough sample both end up below their starting bankrolls. The strategy that's marketed as optimal is, in the real environment in which it's being played, a strategy for going slowly broke.

This is not a bug in your application of GTO. It's a feature of GTO itself. The equilibrium was never going to make money in a raked game. You're supposed to make money by being better than your opponents — winning chips from them faster than the rake takes from you. The rake sets a minimum skill differential below which both players lose. The exploitative deviations are what create that differential. Without them, you and your opponent both pay the rake equally and both end up down. The deviation is the only source of profit in a raked game.

So put the right thing in the slot

I don't want to leave this on the cynical note, because there's an upgrade path and it's available right now. GTO is not optimal — but it is an excellent starting point for finding optimal play. The baseline tells you what a perfectly defending opponent would do, which lets you ask the most useful question in poker: how is the actual person in front of me deviating from that baseline? Once you have the question, the answers — and the money — become available.

That reframe is small. The consequences are large. The player with GTO baselines in his head who uses them to find deviations is, almost by structure, a better player than the one who has the same baselines and tries to execute them. Same baselines, different relationship to them. The relationship is the whole skill. The harder question — when and how to leave the baseline — is the one I take up in When to Deviate from GTO.

So this week, stop calling GTO optimal. When the word comes up in your thinking, pause and ask: with respect to what model? If the answer is "the Nash equilibrium in a frictionless, rake-free, two-player game," then notice you are talking about a model, not the real game. The pause is the practice. Over months it dissolves the bonding between GTO and optimal, and once it's dissolved you can use the solver as a tool without being captured by its own description of itself.

---

This is drawn from the audio lesson The GTO Illusion — hear the whole argument.