Hi,
I have put online a HoldEm odds calculator:
http://www.jbridge.net/jimmy/holdem_sim.htm
The problem is that the betting advise is over optimistic, due to the assumption that every opponent has a random hand. If a player keeps raising, he is more likely to have As & Ks than 2s & 3s.
Any ideas to compensate for such factors?
Computers are notoriously poor poker players. If you are trying to come up with an algorithm to tell a player when to bet and when to fold, you are getting into the cutting edge of artificial intellegence. AFAIK, no one has done this well.
What you could do, that would be quite useful IMO, is give a place enter a hand that the oher player 'may' have and have it calculate your odds against that hand and the odds of the player being dealt that hand. Or, you could list hands the player could have and the odd against that hand and of being dealt that hand.
I don't think you should muck with the raw odds. That's useful info that the machine is good at calculating. Let the human do what he or she is good at.
As dubwai says, this is cutting edge research, but if you
are prepared to read research papers and have access to
clusters, it is possible to beat the average player. Humans
is now number two in Chess (1997 Deep Blue verses Garry
Kasparov), and poker will be next.
Texas Hold'em is a lot harder then Chess, but there has
been at least one success story so far
http://the-raw-prawn.blogspot.com/2004/10/will-bots-destroy-online-poker.html
So play on-line while it lasts. :)
There are a lot of areas where the human brain is still
superior. One of them is pattern recognition. The game
Go (http://gobase.org) is probably the last game where
computers will beat us and this game is all about pattern
recognition. Scientists are using Go as a play ground to
learn more about AI.
In the end, I think that computers will pass us by, for
better or for worse.
> Texas Hold'em is a lot harder then Chess, but there
> has
> been at least one success story so far
> http://the-raw-prawn.blogspot.com/2004/10/will-bots-de
> stroy-online-poker.html
> So play on-line while it lasts. :)
Interesting article thought I'm curious as to how you determine whether a player (or bot) plays poker at a 'master level'. I'm not aware of any objective measure of a poker player's skill. Generally these assesments are based on winnings, tournaments won, and people's opinions.
I think there are good bots and then you have Vex Bot
which is in its own class. The link I sent did not tell
the whole story about Darse Billings bot. Gautam Rao is
one of the best on-line players in the world, and he
had a hard time to beat it. He lost until he adapted his
game and after 7000 games he won, but not with ease.
See the whole story, second half of this link.
http://www.cs.ualberta.ca/~games/poker/Press/nyt.html
The boot is apparently built on a branch of math created
by John Nash. You know, the movie 'A Beautiful Mind'.
> I think there are good bots and then you have Vex Bot
>
> which is in its own class. The link I sent did not
> tell
> the whole story about Darse Billings bot. Gautam Rao
> is
> one of the best on-line players in the world, and he
Never heard of him before now.
> had a hard time to beat it. He lost until he adapted
> his
> game and after 7000 games he won, but not with ease.
Several != Seven
And the way I read it, once he figured out the computer's strategy, he dominated it. That's where computers lack skill. They don't adapt well to changes. A human can determine that he is losing and adjust. That's the big difference. It sounds very similar to the case I give above, the computer is not intimidated so a non-aggresive strategy is needed.
I imagine bots will play pot odds very close to the statistical probablities. This means that if you have the nuts or close to it, you can basically force the software to call all your bets by making the pot odds such that it cannot fold.
> The boot is apparently built on a branch of math created
> by John Nash. You know, the movie 'A Beautiful Mind'.
John Nash did not create game theory. He was one of it's pioneers but Von Neumann wrote a book about poker and game theory before Nash was into it.
> And the way I read it, once he figured out the computer's
> strategy, he dominated it.
According to the more detailed article in
http://www.illustreradvetenskap.com (not on-line), Rao Gautam
was first down 1500 dollars, then up 2800 dollars, then down
3000 dollars, and then finally after 7030 games Rao was
3230 dollars up. That is not how you dominate a game. However,
I agree that no conclusions can be drawn from one incident.
> That's where computers lack skill. They don't adapt well to
> changes.
When a computer can not adapt, it is just because the new
needed strategy has not been included into the model, not
because the computer lack the ability to adapt. It is just
a question of how big the 'space' of different strategies is.
Darse Billings is convinced that a future version of his
program will put humans in the back seat of poker. I find no
good reason to doubt him.
> John Nash did not create game theory. He was one of it's
> pioneers but Von Neumann wrote a book about poker and game
> theory before Nash was into it.
My mistake. It sound interesting and as of now I put the
best of their work on my to-read-list.
> When a computer can not adapt, it is just because the
> new
> needed strategy has not been included into the model,
> not
> because the computer lack the ability to adapt. It is
> just
> a question of how big the 'space' of different
> strategies is.
> Darse Billings is convinced that a future version of
> his
> program will put humans in the back seat of poker. I
> find no
> good reason to doubt him.
I don't have any reason to trust him. The problem I see is that fundamentally poker is a game of you think that I think that you think but you think that I think that you think. It's a lot like the scene in the Princess Bride with the poison. I'm not saying that it's not possible for a computer to play this game effectively but it's not easy and it really comes down to making a computer able to out-smart a human.