|
Poker Bots and Real-Time Game Theory |
 It wasn't long ago that poker bots didn't use any real-time game theory at all. Their method of operating consisted of a series of if/then statements and basic starting hand lookup tables. In fact, many of the poker bots out there today still rely on this primitive method. Setting up rules and conditions can be effective on some levels, but can also be exploited very easily. The inability of poker bots to adapt to opponents is what makes them very unreliable. This is especially true for cash game bots. Over time, their strategy becomes very obvious and predictable. Sadly this gives other players the opportunity to capitalize on any minor errors in a bot's play. Wait, There's Hope! Fortunately there is a type of game theory that goes beyond straightforward guidelines and actually makes its calculations based on the strategy of its opponents. Oh, how the times have changed for poker bots! The Man Who Made It Possible  Have you seen A Beautiful Mind (2001)? Russel Crowe's character, John Nash, is the man responsible for adaptive game theory as we know it. Nash's theory, Nash Equilibrium, is essentially the balance of every possible decision in face of every possible response (it gets more complicated, but this captures its purpose). When applied to poker, the theory produces a strategy that is inexploitable; every time your opponent makes a mistake, you are necessarily awarded with more equity (profit). The larger the blinds are in relation to stack size, the more effective this strategy becomes. Since cash game poker bots are placed in situations where stacks are composed of 100 big blinds, they cannot take advantage of this mathematical goldmine. Why Tournament Poker Bots Have The Advantage  The most important part of sit & go tournaments (especially turbos) is play when blinds are high. The difference between a losing player and a consistent winner is not early or even middle game ability. The majority of tournament equity (profit) comes from how you play when you have 10 big blinds or less. Some players are very talented and adept when it comes to the early blind levels, but are verified losing players because they do not understand how to make mathematically accurate pushes/folds. We designed ICM-Bot to excel during this stage of the game. While our poker bot may appear passive and ultra-conservative during the early phases, it earns its profit as the blinds rise and game theory calculations come into play - situations where it has an edge over all but the most skilled (we're talking world-class) human players.
|
