Poker Wiki

Geometric Bet-Sizing

Equal fraction per street to go all-in on the river.

Reviewed by StackedPoker Theory TeamUpdated 1 min readadvancedThe Language of Bet Sizing

What is Geometric Bet?

Geometric sizing is betting the same fraction of pot on each street so that you end up all-in on the river. Example: With pot 10bb and stack 90bb across 3 streets, bet roughly the same fraction each street to commit all chips by the river.

bet_fraction = (final_pot/starting_pot)^(1/streets) - 1

Why it matters

The geometric bet size ensures the pot grows at a constant rate relative to both players' stacks. It's optimal for the player with a polarized range who wants to commit all chips efficiently.

Going all-in on the river after geometric betting is often the goal when you have the nuts or are executing a multi-street bluff.

Nuance and caveats

Formula: R = (final_pot / starting_pot)^(1/streets) - 1 where final_pot = starting_pot + 2 × effective_stack. With deep stacks, geometric bets are often overbets — this is theoretically correct for perfectly polarized ranges but impractical because real ranges are never perfectly polar.

In practice, smaller geometric bets (50-75% pot) are used to build the pot over multiple streets while preserving range complexity. The concept matters most in 3-bet pots with low SPR, where the pot grows quickly and all-in on the river is natural.

Key takeaway

Equal fraction per street to go all-in on the river.

Frequently asked questions

What is geometric sizing in poker?

Geometric sizing is betting the same fraction of pot on each street so that you end up all-in on the river.

How do you calculate geometric sizing?

bet_fraction = (final_pot/starting_pot)^(1/streets) - 1.

Why does geometric sizing matter?

The geometric bet size ensures the pot grows at a constant rate relative to both players' stacks.

Concepts, lessons and tools that build on this page.

Source: Definition, explanation, formula and examples are quoted from the StackedPoker poker-theory concept registry (lib/theory/concepts.json).