$100,000 Poker Bot Tournament 356
Costa Galanis writes "The LA Times is reporting that a poker tournament will be held where engineers will be able to pit their automatic poker-playing programs against each other in a tournament similar to the upcoming World Series of Poker main event, with a 100,000 dollar cash prize for the winning program. The article mentions how the recent rise in popularity of poker has encouraged many to try and create the poker equivalent of chess' Big Blue, the chess playing computer program that defeated the world's top chess player in a widely publicized event, and also talks about how many engineers also are trying to make bots that are good enough to play and beat human players for money in online casinos."
Texus Holdum (Score:2, Interesting)
Heres hoping this doesnt ruin online poker (Score:2, Interesting)
Um... pokerbot will always win (Score:5, Interesting)
Re:erm.. WTF (Score:4, Interesting)
Have you ever watched poker on TV? Did you notice that the same few people seem to be at the final table a disproportional percentage of the time. It's because although the cards themselves are random, the game is not. Every bet and every action is a subtle piece of a conversation about your perceived strength of your own hand versus your perceived strengths of your opponents.
There's a lot of skill. It's not simply high card wins. The really good poker players willoften win without the best hands, because they know when their opponents are weak and will be willing to give up on a pot.
So, yeah, a poker bot could replicate this.
Re:Bluffing. (Score:5, Interesting)
If two bots were identical and one bluffed 3% of the time (and didn't bluff away all of his chips), in the long run the bluffing bot should win. Because the non-bluffing bot will believe the bluffing bot has a hand those extra 3% of hands, and thus the bluffing bot will in the long run win more than half of the hands and do better in the long run.
The interesting question is how often one should program the computer to bluff in what situations. . .
Re:Bot Training (Score:5, Interesting)
Against real players the primary way of determining this is through the unconscious betting patterns almost every player has. Bots with some AI could do well at this. Bots against other bots is potentially an even more difficult problem.
What I fail to see is why anyone who had a well functioning bot would enter this kind of contest. There is far more money to be made without getting yourself the undue notoriety of this sort of success.
No limit? (Score:2, Interesting)
Re:Bluffing. (Score:5, Interesting)
Not true at all. To be a successful player, you must detect patterns in your opponents' play so you can infer whether you are likely to be ahead of them, and how to maximize the pot when you think you're going to win. This is very difficult for computers to do, even with sophisticated learning algorithms.
Hidden markov models (Score:5, Interesting)
Re:Heres hoping this doesnt ruin online poker (Score:3, Interesting)
I, for one, as a blog and website operator am SICK AND FUCKING TIRED of comment/trackback/referral spam. Do they honestly think that by spamming my server logs I'm going to going to be interested in throwing my money at them? I seem to be missing something, but I'm guessing the people in charge of advertising and promotions for these sites aren't that far removed from Percy from The Green Mile.
I'm sure many will agree with me: die poker sites, die!
Re:Bot Training (Score:5, Interesting)
If you disagree, you're more than welcome to join our weekly game.
Re:Heres hoping this doesnt ruin online poker (Score:5, Interesting)
It may be that current bots can beat some of the worse human players, but it's not clear how many of the human players are that bad, and it's not clear how good the companies that run the servers are at detecting bot behavior.
One thing I'm still wondering about is human-human collusion. It's a big concern in breathe-the-same-air games between humans who don't know each other. Not sure about online poker, however -- do you get thrown in a table with randomly chosen players, none of whom you're likely to know? What about collusion between bots? E.g., you could be the only player at the table, not realizing you're playing against 6 bots, each of which knows what cards the others have.
Re:Um... pokerbot will always win (Score:2, Interesting)
Re:Chinook (Score:2, Interesting)
I am also an ulumni of the U of A. More specifically I did my research in the games department, although not specifically on poker. I was sort of suprised the article never mentioned Darse Billings who is one of the main people behind poker agents at the U of A.
Here is a relevant link: http://www.cs.ualberta.ca/~games/poker/ [ualberta.ca]
Re:Poker is Hard (Score:3, Interesting)
About your sig: it is highly unlikely that any form of prime factorization is NP-complete. The reason is that the problem is contained in both NP and co-NP. So if it's NP-complete, then NP=co-NP, which nobody seriously believes is true and would be very surprising.
Re:That's not the point of bluffing. (Score:3, Interesting)
I bluff a lot early in the night, then I start playing tight later in the night.
Then, when I get a really good hand, and bet hard, I get a few people to come in with me, because I've got them into the mindset that they have me figured out. They "know" I'm bluffing.
It's all about letting them think they have you figured out.
Who the hell modded this guy up? (Score:3, Interesting)
This is a ridiculous statement. You *might* be able to apply it to limit hold 'em, but certainly not no-limit games. The point of bluffing is to not only make it more difficult for your opponents to put you on a hand, but also (and more importantly, IMHO) to allow you to win pots when you do not have the best hand at the table. This is important in limit hold 'em and critical in no limit. If you think the point of bluffing is just to make you harder to read, I'd love to sit down with you at a table.
Re:Bluffing. (Score:3, Interesting)
Not (entirely) true. I have followed research in poker programs and I can tell you that programs that bluff play consistently worse than programs that don't. That is, if bluffing is interpreted as "rating your own hand as a lot better than it actually is, in the hope of getting your opponent to fold". The problem is that you are overbetting your hand when you bluff, and if you get called on it, you lose BIG.
The trick is not to just bluff, but to bluff when the hand is worth it, and when you are fairly certain your opponent will fall for it. This means that a good poker program is not one that bluffs, but one that is very good at opponent modelling.
Opponent modelling is in a completely different ballpark than chance bluffing. While there is some research into this, not much has been published. That is why research groups as the one of Jonathan Schaeffer (mentioned in the article) have a far better chance of winning the prize than individuals working on poker programs. I know where I bet my money.
In theory easy, reality difficult.. (Score:3, Interesting)
Making the bet, the bot should make a probabilistic bet around the "value" of the cards. He can bet much higher (bluff) or lower (hoping someone else will bluff) than their real value. Likewise, when calling/raising a bet, a bet is considered the result of many possible sets of cards.
To make an example:
Two pairs ->
$5 5%
$10 25%
$15 50%
$20 15%
$25 5%
On the opposing side, $15 ->
Nothing 5%
One pair 20%
Two pair 50%
Three equal 20%
Straight 5%
He will bluff exactly so often as is optimal. Likewise, he will call a bluff exactly so often as is optimal.
As an opponent:
call more often -> too many non-bluffs
call less often -> gets away with too many bluffs
bluff more often -> called too much
bluff too little -> fold too many hands
Obviously, it is quite a bit more complex than that (bet given the cards, call/raise given multiple other bets) but that is the basic idea.
Of course, this only finds the optimal strategy - i.e. one that will not lose. It is quite another thing to find a strategy to exploit opponents' sub-optimal strategy. One-on-one, it can't be beat. In a tournament (unless you play "to the bitter end" with only one bot standing, where the last round should be one-on-one) other strategies might be better for exploiting other bots and getting the most cash.
Kjella
Tip the Dealer, people (Score:2, Interesting)
-the pot was particularly big, or
-you won it by luck and are in a good mood
The dealer gets no part of the rake, but a wage. A kinda pitiful wage if you consider the training and experience a good dealer requires, and the kind of shit they have to put up with sometimes.
Tip. No, it does nothing to increase your chances, but neither does tipping a waiter
Re:Bluffing. (Score:3, Interesting)
Too late... (Score:1, Interesting)
Too late.
I started doing exactly this four years ago. I ran 4 clients at
It was fun while it lasted and got me the down payment on my house, but I wouldn't recommend trying to jump into the game at this point unless you have something really special.
If you really want to give it a go, you'll need about $3-5K to get started with one client. Find a good offshore hosting site, and expect to lose a chunk of money in testing and tweaking. There are a ton of stupid little gotchas that can't be hammered out offline.