Equilibrium, exploits and why the solver mixes

On this page

• Explain equilibrium as a strategy nobody can beat, not as the best play against everyone.
• Explain why a hand that mixes earns nearly the same with each of its actions.
• Judge a decision by its EV loss, not by how often the reference picks it.
• Choose the simpler action when the EV difference is small.
• Treat equilibrium as the baseline that every exploit starts from.

Concepts used: GTO and exploitative play, EV, EV loss and how Rail grades

What equilibrium promises

An equilibrium is a pair of strategies where neither player can gain by changing theirs. Play your half of it and no opponent can beat you: the best they can do is not lose. That is what people mean by GTO.

It is not the same as the most profitable play against everyone. Against an opponent with a clear leak, a strategy that leans into the leak earns more, at the price of being exploitable itself.

Why a solver mixes

If one action were clearly better with a hand, the solver would take it every time. It splits a hand between actions only when they are worth the same, or nearly the same. So a mix tells you something useful: the actions are close in EV.

The classic case is a bluff-catcher on the river. When the pot is 10 bb and the bet is 7.5 bb, the bettor's equilibrium range bluffs 7.5 / (10 + 15) = 30% of the time. Calling then wins 17.5 bb 30% of the time and loses 7.5 bb the rest: 0.3 × 17.5 = 5.25 and 0.7 × 7.5 = 5.25. The call is worth exactly zero, the same as folding, so mixing costs nothing.

The grade measures EV, not frequency

Because mixed actions are close in EV, Rail does not punish you for taking the less frequent one. The grade follows how much EV your choice gave up, and frequency only decides between Perfect and Good when the loss is small.

Compare an option the reference plays only 2% of the time that gives up 1.5% of the pot: that is Inaccurate. One it never plays that gives up 5% of the pot is a Mistake. The bigger the EV gap, the worse the grade, whatever the frequencies say.

If you always pick the 40% option, no single hand will grade badly. The pattern shows up in your reports instead, where it belongs.

Simplifying costs little when the EVs are close

Nobody can reproduce a solver's exact mixes at the table, and you do not need to. When two actions are close, picking one of them every time costs only the small gap, multiplied by how often you are in the spot.

The same logic tells you where not to simplify. When one option is far ahead, always taking the other is a real leak, and your grades will say so in every hand.

A baseline to exploit from

Treat equilibrium as your default against players you know nothing about: it cannot lose to anyone and it profits from every mistake they make. Deviate when you have a real read, and deviate in the direction of their leak.

When you do, the grade tells you what the exploit costs if your read is wrong. A small cost for a big expected gain is a good trade; a large cost on a thin read is not.

Rules of thumb

• A mix means the actions are close in EV; taking the more frequent or the simpler one costs little.
• Judge a choice by its EV loss. Frequency only separates Perfect from Good.
• The cost of a simplification is the EV gap times how often you are in the spot.
• Start from equilibrium, and deviate only with a read, towards the opponent's leak.

When they break

• A mix in a solver output can hide hands that are not close at all: check the EV of your exact hand, not the range's average.
• Against a player with a big, proven leak, equilibrium play leaves money on the table; that is when exploiting pays.