GTO (game theory optimal)

GTO stands for game theory optimal. A GTO strategy is one that cannot be beaten by any counter-strategy — not because it wins the most, but because nothing an opponent does can make it lose.

GTO does not mean 'always winning'. It means your opponent cannot gain by changing their strategy against you, whatever they do. Against a perfect opponent that is the best you can achieve; against a weak one a purely GTO player leaves money on the table because they never punish specific leaks. Modern high-stakes players study solver output to learn the shape of balanced strategies — bet sizes, frequencies, which hands bluff — and then apply exploitative adjustments when they spot patterns. The opposite pole is exploitative play, where you deliberately play unbalanced because the opponent will not punish it.

What the term actually means

GTO describes a Nash equilibrium. In a two-player zero-sum game such as heads-up poker, there exists a strategy pair where neither player can improve their result by changing their own play while the other stays fixed. A player using that strategy is unexploitable: they do not need to know anything about the opponent, and no adjustment the opponent makes will beat them.

That is a defensive guarantee, not an offensive one. An unexploitable strategy breaks even against another unexploitable strategy, and it does not punish mistakes as hard as it could. Against a player who folds far too often, the GTO bluffing frequency leaves money on the table.

Why balance is the central idea

Most of what players mean by "playing GTO" comes down to balance: having both strong hands and bluffs in any line, in proportions that make an opponent's decision indifferent. If a bet is sized at the pot, a caller needs roughly 33 per cent equity to break even, so a betting range mixed to win exactly that often leaves the opponent with no profitable response — calling and folding earn the same.

This is why solver outputs are full of mixed strategies, taking the same hand down two different lines at some frequency. The mixture is what removes the opponent's ability to read the range.

Solvers, and what they do not tell you

Solvers compute approximate equilibria for a defined spot: a fixed stack depth, a fixed set of bet sizes, a fixed starting range. Change any input and the output changes. Two consequences follow. First, a memorised solver line is only correct for the tree it was solved in. Second, because the solver is told which bet sizes are allowed, it cannot tell a player that a size outside the tree would have been better.

GTO versus exploitative play

Against strong, unknown opposition, approximating equilibrium is a sound default: it limits the damage of being outplayed. Against identifiable weaknesses it is the wrong goal. A player who never folds should be bluffed less and value-bet more; a player who folds too much should be bluffed more. These adjustments are deliberately exploitable themselves, which is the trade — giving up the defensive guarantee in exchange for a larger edge against this particular opponent.

In practice most strong players use equilibrium as a reference point rather than a target: understand what balanced play looks like in a spot, then deviate from it on purpose and for a stated reason.

The multiway caveat

The theoretical guarantees of equilibrium play are clean in two-player games. With three or more players the mathematics is far less tidy — equilibria may not be unique, and the result can depend on what the other opponents do with each other. Confident claims about "the GTO play" in a multiway pot deserve more scepticism than the same claim heads-up.