4 ms·

People are working on it (of course, since it could be extremely lucrative). However, the problem is 'hard' because No Limit Hold 'Em is really a game of bluff

by 727374 12y ago

People are working on it (of course, since it could be extremely lucrative). However, the problem is 'hard' because No Limit Hold 'Em is really a game of bluffing and not math. The best human players alive have mastered the skill of determining the cards their opponents hold based on their moves in the current hand, the current game, and previous games. Online poker lacks the physical 'tells' that occur in a live game, like making an obvious fake reaction to some event. But, there are also some online 'tells' such as reaction speed. I used to play a lot of online poker.

panorama 12y ago
This is largely incorrect. NLHE can be more about "bluffing" because there's greater variability between ones chance of winning a hand and the potential bet size. But it still comes down to a mathematical decision, just that it's more difficult to accurately calculate the best play given the wide range of variables: your opponent's likely holdings, the bet size, the pot size, your hand. In Fixed Limit, the bet size and pot size are all controlled and can be pre-determined, leaving fewer permutations to solve. It's easy to calculate that if a bet is a constant, then you can profitably call with holdings from XY and "higher" in a given situation, but it obviously gets harder when the bet size can vary greatly. To answer OP, Fixed Limit is considered solved (or close to it? Haven't been following the news on this) from a mathematical and game theory perspective, and so AIs have been built to play game theory optimally. NLHE is still far, far from being solved, but there are bots that have been created to churn out small, albeit automatic, wages short-stacking (to mitigate bet-size variability) smaller stakes games where its tendencies aren't as openly exploited. Poker is very much a mathematical game. And physical tells, reaction-time tells online are pretty insignificant in the longrun.
- 727374 12y ago
  Sorry, but you can't just casually mention 'your opponents likely holdings' like it's trivial to determine. Putting an opponent on a hand is The Skill that top pros have mastered and if anything probably comes from great pattern matching and a strong memory, . But I don't know nor do you because nobody has built a computer that can do it well.
  - panorama 12y ago
    You speak of these 'top pros' and their hand-reading abilities as if they're an enigma. Hand-reading is mostly a mathematical exercise. It's also infrequently about placing a person on a specific hand but rather a range of hands, with different weights assigned to different potential holdings based on sure, 'pattern matching', understanding player psychology, etc. But it still goes back to basic combinatorial math. Surely a computer can do this better than a human. Source: Was a 'top pro' for many years online
- splonk 12y ago
  2 player fixed limit is not solved, but the best in class are generally considered to be better than the best human players and likely have been for several years. I'm not up to date on the general consensus on the state of 2 player no limit, but personally I'd bet on the best bots over the best humans even at 100 blind stacks. http://www.computerpokercompetition.org/ http://www.computerpokercompetition.org/ has some results from previous competitions. You can probably find some of the relevant papers on the subject from there as well.

by 727374 12y ago

panorama 12y ago

This is largely incorrect. NLHE can be more about "bluffing" because there's greater variability between ones chance of winning a hand and the potential bet size. But it still comes down to a mathematical decision, just that it's more difficult to accurately calculate the best play given the wide range of variables: your opponent's likely holdings, the bet size, the pot size, your hand. In Fixed Limit, the bet size and pot size are all controlled and can be pre-determined, leaving fewer permutations to solve. It's easy to calculate that if a bet is a constant, then you can profitably call with holdings from XY and "higher" in a given situation, but it obviously gets harder when the bet size can vary greatly. To answer OP, Fixed Limit is considered solved (or close to it? Haven't been following the news on this) from a mathematical and game theory perspective, and so AIs have been built to play game theory optimally. NLHE is still far, far from being solved, but there are bots that have been created to churn out small, albeit automatic, wages short-stacking (to mitigate bet-size variability) smaller stakes games where its tendencies aren't as openly exploited. Poker is very much a mathematical game. And physical tells, reaction-time tells online are pretty insignificant in the longrun.

727374 12y ago
Sorry, but you can't just casually mention 'your opponents likely holdings' like it's trivial to determine. Putting an opponent on a hand is The Skill that top pros have mastered and if anything probably comes from great pattern matching and a strong memory, . But I don't know nor do you because nobody has built a computer that can do it well.
- panorama 12y ago
  You speak of these 'top pros' and their hand-reading abilities as if they're an enigma. Hand-reading is mostly a mathematical exercise. It's also infrequently about placing a person on a specific hand but rather a range of hands, with different weights assigned to different potential holdings based on sure, 'pattern matching', understanding player psychology, etc. But it still goes back to basic combinatorial math. Surely a computer can do this better than a human. Source: Was a 'top pro' for many years online
splonk 12y ago
2 player fixed limit is not solved, but the best in class are generally considered to be better than the best human players and likely have been for several years. I'm not up to date on the general consensus on the state of 2 player no limit, but personally I'd bet on the best bots over the best humans even at 100 blind stacks. http://www.computerpokercompetition.org/ http://www.computerpokercompetition.org/ has some results from previous competitions. You can probably find some of the relevant papers on the subject from there as well.