Texas Hold'em Poker in Haskell

Every hand, ranked

The evaluator covers the whole ladder, royal flush down to high card. Ranking is only half of it: two players can hold the same category and still not split the pot, so every category carries its own comparison rule. Above that sits winner determination across several tied players at once.

Four ways to play

Four distinct strategies ship with it: Random, Passive, Aggressive and Smart. A human can sit at the same table, and human input is validated before it reaches the game. The deck is shuffled and dealt with proper randomisation, which is one of the places the program has to step outside pure functions.

There is no mutable state to fall back on, so the whole game is a value passed forward and returned changed.

Blinds, bets, elimination

The table runs the full loop rather than one hand in isolation. Small and big blinds post each round; fold, check, call and raise each carry the constraints that make them legal; chips are tracked per player and the dealer rotates. Players are eliminated when they run out, and rounds continue on top of that state, so the game runs to a result.

State without mutation

There is no mutable state to fall back on, so the whole game is a value passed forward and returned changed. Shuffling the deck needs randomness and asking a human to act needs input, so those parts sit in a monadic layer while the rules underneath stay pure functions over immutable data.