Rake changes the game you are solving

A fold example worth checking

Start with a 5 bb matched pot. One player bets 5 bb and the other folds. At 5% rake, with a cap above the resulting charge:

The same hand history now has different terminal payoffs. This is a worked arithmetic example, not a new benchmark run. PulseGTO defaults to matched-pot rake. “Rake uncalled bets” enables the full-pot fold rule for rebuilding a model that uses it. A blueprint re-solve retains the blueprint’s rule.

Percentage, cap and eligibility

For an eligible pot, the basic charge is the smaller of the configured percentage of the raked base and the cap. At 5% with a 3 bb cap, a 40 bb base gives 2 bb; an 80 bb base would give 4 bb before the cap and therefore charges 3 bb. Express the pot and cap in the same units.

No flop, no drop controls whether a hand ending before a flop is eligible. A minimum-pot setting adds another eligibility condition. Do not assume two “5% cap 3” exports agree if their fold rule, no-flop rule or units differ. For the model of a real room, consult that room’s published rules and hand histories; this guide does not assert that every room or reference solver uses the same rule.

Why subtracting one fixed cost from every hand fails

A hand that folds early, wins a small pot or reaches a large showdown does not encounter the same rake charge. Strategies also adjust to the changed payoffs. Consequently the difference between a cEV solve and a raked solve is not a hand-independent fee you can subtract from a chart.

Nor can you infer the effect from total game EV alone. A higher cap may not bind in a shallow heads-up pot, while it matters in deeper or multiway pots. Uncalled-bet rake affects fold-winning lines and may change action frequencies even when showdown accounting is identical.

Comparing two solvers fairly

1. Match blinds, starting stacks, antes and dead money.
2. Match the preflop actions: calls, squeeze sizes, jam calls and raise limits.
3. Match postflop betting sizes, donk/probe permissions and all-in conversion.
4. Record rake percentage, cap, units, minimum-pot rule, no-flop rule and uncalled-bet rule.
5. Compare the same decision and EV baseline. A fold-relative value and a net value including the blind are not the same number.

Even matched rake does not make differing action trees interchangeable. PulseGTO’s preflop reference measurements state the specific trees and rules. Their results are evidence for those configurations, not all cash-game trees.

Keep the rule with the result

Configure rake when you create the game, then preserve its tree configuration with the checkpoint and blueprint. Changing rake makes a new game; do not silently continue an existing run under different payoffs. For a local postflop spot, supply the rake appropriate to that exact pot along with its ranges and stacks.

Before judging a small EV difference, distinguish historical blueprint estimates from values of a converged local re-solve. It is also unsafe to derive one player’s raked EV by simply negating another’s: money leaves the players’ combined payoff through rake.