How Do Poker Bots Decide? Ranges, Styles and Randomness
A plain-English look inside an AI poker bot: how it thinks in ranges, how a personality changes its play, and why its randomness makes it hard to read.
A poker bot doesn’t “look” at its cards and get a feeling. Every decision comes from three layers working together: a strategy, a personality and a dice roll.
Layer one: a strategy built on ranges
A range is every hand a player takes a particular action with in a particular spot. “Opens 25% of hands from the cutoff” is a range. A good bot never thinks about its hand alone; it thinks about all the hands it could have here, and all the hands you could have.
That is why each bot’s profile shows a 13×13 chart. Pairs run down the diagonal, suited hands sit above it and offsuit hands below. A coloured cell means the bot plays that hand; a cell that is only partly coloured means it plays it some of the time.
Layer two: a personality
Starting from a balanced strategy, a personality layer shifts the bot’s frequencies:
- Grinder keeps them close to balanced, so it has no big weakness to attack.
- Sticky turns most raises into calls and almost never folds once it is in a pot.
- Maniac turns calls into raises and bluffs far more than it should.
Each change is a known leak. That is what makes the bots useful for practice: a recognisable type of opponent, with mistakes you can learn to punish.
Layer three: randomness
When a strategy says “raise this hand 50% of the time”, the bot rolls for it every hand. This matters more than it sounds. If a bot always raised its strong hands and always checked its weak ones, you would learn the pattern in an evening. Mixing makes its bets carry the same message whether it holds the nuts or air, exactly like a strong human player.
Counting combos
To estimate a range, count combinations, not hand names. Every pocket pair has 6 combos, every suited hand 4 and every offsuit hand 12, for 1,326 in total. That is why “plays 25% of hands” means about 330 combos, and why offsuit hands make a range much wider than their number of cells suggests.
Why this helps you
Once you know how a bot decides, you can play against its range instead of guessing its hand. That habit carries over directly to games against people.