Nowadays at the top level of play, most players are aware of the existence of tells, and do their best to combat it by becoming consistent in their motions. Polk: There certainly is if players are weak enough to have them. Navin: Is there a place for intuition - or even physical tells - in the future of Texas Hold Em No Limit poker? I understand it is important for universities to try and raise funding, but to take away from our accomplishment as players felt like we were being slighted. We beat it handily, winning the vast majority of the days as a team. My main issue was that the way that the event was presented to the media, was that it was an extremely close contest, which was not the case. While we did not win at a 95% confidence interval, we did win at a 92-93% confidence interval, so we were quite close to achieving statistical significance. I am on the record saying that it was only a matter of time before the AI eventually overtakes humans, and that appears to have happened this year.īut after our victory last year there were many phrasings that I found to be both misleading and not true. Polk: I felt a bit unhappy with the way that the match ended last time. Navin: At the end of the session, where your team had collected the most chips, the AI developers told the media that the session had been a "statistical tie." It almost sounds as if your winning was being de-legitimized. This means that while they can use a variety of sizes a human would be unable to, they cannot harness the true power of no limit hold'em (being able to bet whatever you want) because they need to pick certain sizes that they understand to prevent the amount of analysis from being too large.
#Holdem bot vs pro how to
No limit hold'em has far more game states than say, chess or backgammon, which makes solving the game a much more difficult task.Īnother issue that programs have, is that they have to figure out how to interpret different bet sizes. No limit hold'em is also not "unsolvable", it is just so complex to solve that we are nowhere near the solution. When playing limit hold'em, all it has to do is decide how to play against a bet (that has a specific bet size). This makes the game much harder to solve, as the computer has to decide what it would do against a $100 bet, $1,000 bet, or even $15,000 bet. In pot limit/no limit hold'em, you can choose what size you want to bet. In limit hold'em you can never choose what size you want to bet, only whether you would like to bet at all.
Polk: This is a bit of a misquote, AI has "solved" limit hold'em, not pot limit.
What is it about the betting limits involved that makes it impossible to solve? Navin: You've said that AI bots have recently "solved" pot limit hold'em, but that "no limit" is unsolvable. This time is spent more efficiently than a human could dream of, as it can run all scenarios out instead of just likely to occur ones. It is here that the program recalculates optimal strategy, so that it knows exactly what it is going to do facing all further action. At that point 50% of the hand is over, and 50% of the hand remains to be played. Polk: It would pause at the same point in every hand, which would be once the turn is dealt.