The Pot-Limit Constraint

The pot-limit rule is the defining structural feature of Pot-Limit Omaha. It bounds the action space at every decision and is the dominant determinant of bot policy shape.

The rule

The maximum bet permitted at any decision point in Pot-Limit Omaha equals the size of the current pot plus the amount the actor must call. A raise must therefore be at least one minimum bet and at most a pot-sized raise. The rule applies on every street and to every player. The action space is consequently a closed interval rather than the half-open interval available in No-Limit Hold'em.

Geometric stack progression

Because the pot grows in proportion to the bets placed into it, sequential pot-sized actions produce a geometric stack progression. Starting from a pot of one big blind, a sequence of pot-sized raises grows the pot by a factor of approximately 3.5 per action. A four-bet pot-sized sequence pre-flop typically commits stacks of around 50–60 big blinds; deeper stacks therefore exit the pot-sized line and require sub-pot raise selections to maintain stack-to-pot ratios useful for post-flop play.

This geometric progression is significant for bot design because it determines which decision points are practically reachable. The bot's policy table must allocate resolution to the spots that actually occur, and the pot-limit rule concentrates those spots into a relatively small number of stack-to-pot regimes.

Action abstraction in bots

Solver-based bots discretize the continuous bet-sizing space into a small set of fractions of the pot — commonly 33%, 50%, 75%, 100% — plus an all-in option where stack depth permits. The pot-limit ceiling removes the need to model arbitrarily large overbets that occupy a non-trivial share of the action space in No-Limit Hold'em. The smaller abstraction reduces the size of the strategy table and the memory footprint of online play.

SizingTypical use
33% potRange bets on dry boards, blocker bets out of position.
50% potMixed value/protection on coordinated boards.
75% potPolarised lines on dynamic boards, turn barrels.
PotHeavy-equity draws, nut-class value, leverage on rivers.

Coupling with equity

The pot-limit ceiling places an upper bound on the price a hand can be charged to continue. A draw with 45% equity cannot be made unprofitable in a single bet under the rule, because no legal bet offers worse than 2-to-1 immediate odds. Bot policies therefore weight equity denial less heavily than in No-Limit Hold'em and rely on multi-street pressure, blocker effects, and range advantage to extract value.

Consequences for policy

Three policy properties follow directly from the constraint. First, calling frequencies are higher than in No-Limit Hold'em at comparable equity levels, because the price laid is never punitive. Second, range realisation is higher in position, because the in-position actor can always cap the pot size on the next street. Third, post-flop check-raise frequencies are elevated, because the geometric escalation lets the raiser commit stacks within two streets without needing overbets.

Inquiries on policy design, sizing trees, or solver abstraction parameters are accepted through the project contact.

Sit at the pot
  1. Sklansky, D., Malmuth, M. (1999). Hold'em Poker for Advanced Players, on pot-limit structure.
  2. Brown, N., Sandholm, T. (2018). Depth-limited solving for imperfect-information games.
  3. MonkerSolver documentation: action abstraction parameters.