BLOG

Bet sizing in solver trees: what 33%, 75% and 125% mean for your line

GTO MAX AI team · 4 October 2026 · 6 min read

Every solver output you have ever seen was computed with a short list of bet sizes, and that list shaped the answer more than most players realise. The sizes are not arbitrary: each does a job, and knowing the job tells you what the solver is thinking when it picks one. This post explains the standard sizes, why trees are small, and what happens when a real opponent bets a number that is not on the list.

1. Why a tree has three sizes and not thirty

A solver's work grows with every action it must consider at every node. Allowing ten bet sizes on every street makes the tree hundreds of times bigger for an answer that barely changes, because near sizes play nearly the same. Practitioners settled on a few sizes per street that cover the strategic purposes: a small bet, a large bet, an overbet where the board justifies it, and all-in. GTO MAX AI's live tree is 33% and 75% of the pot on the turn, 33%, 75% and 125% on the river, raises at 75% of the pot, all-in always available; flops in the library use one third of the pot. The exact numbers matter less than the roles.

2. What each size is for

A third of the pot: the range bet

A small bet is what a range with an equity advantage but no nut advantage wants: it bets everything cheaply, denies equity to the hands that would fold, and loses little when raised. Dry flops after a preflop raise are the classic case: the raiser bets a third with the whole range, because the big blind cannot profitably raise or call wide against it. A small bet is not weak. It is a bet by a range that is ahead everywhere by a little.

Three quarters of the pot: the polarised bet

A larger bet comes from a range that is split: strong hands that want to build a pot and bluffs that want folds, with the medium hands checking. The size is chosen so that the opponent's bluff-catchers are indifferent: three quarters pot needs 43% folds to break even as a bluff, and the equilibrium bluffs just enough that a correct defender cannot gain by calling or folding. Wet boards, turns that favour the aggressor, rivers where the medium hands have resolved: these are large-bet streets.

The overbet: nut advantage

When one player can have the nuts and the other cannot, the player with the nuts can bet more than the pot, because the opponent's best hands are bluff-catchers and the price can be set anywhere. Overbets are how a solver extracts maximum value with the top of a range on boards where it holds all the strongest combinations, backed by bluffs that need a very high fold frequency and get it because the opponent faces a terrible price. An overbet from a player who understands this is a polarised signal; from a player who does not, it is usually a hand that is scared of being outdrawn.

All-in

Always in the tree, because at some stack-to-pot ratio it is the only sensible large bet, and because leaving it out would let the solver pretend a player cannot be put to a decision for everything.

3. What a tool does with a size not in the tree

An opponent bets 55% of the pot and the tree has 33 and 75. The tool maps the bet to the nearest tree size, here 75, and reads the strategy at that node. For most spots the answer is the same; the frequency of a call might shift a few points. For spots on the boundary (a size exactly between two tree sizes against a mixed defence) the mapping can change the top action. A tool that publishes its tree lets you recognise those cases. One that does not is asking you to trust a mapping you cannot see.

The same applies to your own sizing. When the advice says "bet 75%" and the client's slider gives you 72%, the difference is nothing. When your preferred size is 50% and the tree offered 33 and 75, you are playing a strategy the solver did not evaluate, and the frequencies you were shown do not quite apply to it.

4. Sizing against a specific opponent

The tree's sizes are the equilibrium's. Against a real player they are a starting point. A station who calls three quarters of the pot as readily as a third is a player you bet bigger against with value and never bluff; a player who folds to any bet over half pot is one you bet exactly half pot into as a bluff and small with value. This is where the exploit line in GTO MAX AI changes the size as well as the action, and says why: the profile's fold frequency against each size decides it. How the profile is built: player profiling.

GTO MAX AI: the RTA that knows the player.

GTO line, exploit line and the reason, live, on up to four tables, in Chrome. GGPoker, Natural8, ClubGG, PokerNow, and PokerStars in beta. Try it for 30 minutes on Telegram.

Get a 30-minute trial

Coaching and study tool. Use it where real-time assistance is allowed: study, home games, private clubs, play money. Most online sites forbid real-time assistance. Advice can be wrong; every decision is yours. Terms of use.

More from the blog

Longer reads: the poker RTA guide and all guides.