GTO Is a Losing Strategy: The Rake Math Nobody Emphasizes

I want to give you the second technical fact, because the first one — that GTO is unexploitable, not maximally profitable — is bad enough on its own, but the second one makes the picture much worse. Here it is, and I want to repeat it because it is one of the most underappreciated facts in modern poker: GTO in a rake environment is theoretically a losing strategy.

Read that again. In a rake environment, the Nash equilibrium between two GTO players results in both players losing money to the house. Not one of them. Both. The strategy that is marketed to you as optimal is, in the real environment where it is actually being played, a strategy for going slowly broke.

Nash assumes a game that doesn't exist

Go back to the definition of a Nash equilibrium. It is a pair of strategies in a two-player zero-sum game such that neither player can improve by deviating while the other stays put. The phrase doing the quiet work there is zero-sum. Zero-sum means the chips one player wins are exactly the chips the other player loses. Nothing leaves the system. Every chip stays with the players.

The real game of poker is not zero-sum. It is negative-sum, because the rake comes out of the pot before it goes to either player. The house takes its cut on every hand. Chips leak out of the system on every pot. The Nash equilibrium is computed against the abstract game in which all the chips stay with the players. The real game is a different game, one in which a piece of the pot disappears into the house's tray every time you play a hand.

The equilibrium does not account for this. It cannot, because it was never asked to. The math was solved for a frictionless game. You are playing in a game with friction. The solver hands you the answer to a question that is not quite the question you are sitting at the table asking.

Two perfect players, both down

So follow the consequence through. Take two players who both play the equilibrium perfectly against each other. By the definition of the equilibrium, neither can do better than break even against the other. Their edge over each other is exactly zero. That is what the equilibrium guarantees: a fixed point where neither side gains.

Now turn the rake back on. Every pot, the house takes a percentage. Over a long enough sample, both of these perfect GTO players end up below their starting bankrolls. The rake takes more than the equilibrium edge, and the equilibrium edge between two equilibrium players is zero. So the rake takes everything it asks for and neither player has any edge to cover it. Both lose.

This is one of the most ironic results in applied game theory. The strategy that gets sold as optimal is, by structure, a vehicle for going slowly broke in the environment where it is being used. This is not a bug in your application of GTO. You did not execute it wrong. This is a feature of GTO itself. The equilibrium was never going to make money in a raked game. It was never designed to. It was designed to be unexploitable, and unexploitable in a raked game means both players pay the rake equally and both go down.

The rake sets a minimum skill differential

Here is the way I want you to hold this, because it reframes what your edge actually is. You are supposed to make money in poker by being better than your opponents. But "better than your opponents" has a precise meaning in a raked game: it means winning enough chips from them to overcome the rake.

The rake creates a minimum skill differential below which both players lose. If the gap between you and your opponent is smaller than what the rake takes, you both end up down, no matter who is technically the better player on paper. You have to clear the rake before you clear anything. The rake is a tax you pay before your skill edge even starts counting.

So where does the skill differential come from? Not from the GTO baseline. Two players sharing the same baseline have a differential of zero. The differential comes from the deviations — from the exploitative adjustments one player makes against the specific leaks of the other. The exploitative deviations from GTO are what create the skill differential. Without them, you and your opponent both pay the rake equally and both end up down. With them, you take chips from your opponent at a rate that exceeds what the rake takes from you, and you net positive over time.

I'll say it as plainly as I can: the exploitative deviations are the only source of profit in a raked game. The GTO baseline alone cannot be profitable. It mathematically cannot. The baseline is the part of your game that produces a guaranteed long-run loss once the house takes its cut. The profit lives entirely in the part the baseline does not contain.

What this does to the "optimal" framing

This is, I think, the most important fact in this whole conversation. The thing you have been told is optimal is, by structure, a guaranteed loser in the environment where you are using it. If you were reaching for the equilibrium as your goal, and you were playing in a raked game — which is every online game and almost every live game in existence — you were reaching for a strategy that produces guaranteed long-run losses.

The industry has not been hiding this fact, exactly. But it has not been emphasizing it either. And the emphasis is the entire difference between a player who reaches for GTO as a goal and a player who uses GTO as a tool for finding deviations. The marketing of GTO as the optimal strategy has been, in a precise technical sense, the marketing of a known loser. Sold honestly, the product would be described as "the unexploitable baseline that is theoretically a loser in your actual game." That is a true description. It is not a description that sells subscriptions.

I am not telling you GTO study is worthless. I am telling you that GTO study without the corresponding exploitative work is reaching for a strategy that cannot make money. The full toolkit is GTO plus exploitation. GTO alone is incomplete. It is the defensive floor. It is the thing that keeps you from getting destroyed. It is not the thing that makes you a winner, because in a raked game it cannot be.

Do the math on your own stake

I want you to make this real rather than abstract, so do the arithmetic. Look up the actual rake percentage at the stakes you play. Calculate, roughly, how many big blinds per hundred hands the rake takes from you. That number is the minimum win rate you have to clear just to be break even. Hold that number in your head and notice that two GTO players grinding each other would both lose exactly that amount.

Then ask yourself the question the optimality framing has been keeping you from asking: where is my edge over my opponents coming from, in big blinds per hundred, and is it bigger than the rake? If you cannot articulate where your edge is coming from, you do not have one in any reliable sense — and the GTO framing has been the thing keeping you from noticing.

The edge is in the deviation. It was always in the deviation. The rake math is the proof that it cannot be anywhere else.

---

This article is drawn from the audio lesson The GTO Illusion.