Hi,

I am doing some research before I start playing online. I have a lot of experiecnce in live games, and would like to move my skills online.

I am very interested in poker bots. My educational background is in computer science and mathematics so this type of thing would very much appeal to me on various levels. I am sure that most players hate bots with a passion.

I would be interested in what bots are the best, and if the source code is available for the best bots (or any bots), and where to find this. I would love to be able to take some existing code and tweak it to be able to make the bot very dynamic so it could be tuned into each game based on my observations. Maybe this capabilty already exists.

I know this won't be too popular, so if you do not like what I have to say please excuse me and ignore me. If you are a fellow computer scientist please ring in with your thoughts.

Thanks,

Superstar

There was a scammer here recently, you could check out his scam limit bot.

It doesn't come with source code, but you could study its behaviour and figure its weaknesses.

Are you interested in Limit or NL? Most bots seem to be limit.

__________________
ICM / EV Calculator and other free poker software

HBAHAHAHAHAHHAHAHAHAHAHAHAHHAHAHAHAHAHAHAHAHA

*takes a breath*


BAHAHAHAHHAHAHHAHAHAHAHAHAHAHHAHAHAHHAHAHAHA


On another note. Not only do u say ur awsome at poker, but u also say ur a professional athelet, AND NOW A COMPUTER SCIENTIST.

Please stop, please please please please please leave before u make urself look any more idiotic

For my sins I now teach CS at University.. I'm a physics PhD by training.. so my maths is strong.

The "definitive" work was done by the University of Alberta poker group. their system (Poker academy) not only has their own bots (with ten years research into them) but is also open to bots ppl construct themselves. These are the strongest bots on the net. The system retails at $129 ... it is serious software.... in as much that most pros like to have it to test out stuff (the analysis is very good) and the preflop set ups can be played time and time again. BUT - no one who knew what they were doing would set it up in a real game...

I'm not a limit player. I was beating the limit bots within 2 days on the highest setting... This is the bots "homeground" this is where the decade of research focussed..

I am a No-limit ring player - the bots they have for this... never got near me..............

__________________
See me playing $10/$20NL like it was play money

http://video.google.co.uk/videoplay?...405&q=xxdemexx

Doberman: "but Sarge, isn't poker gambling and just luck?"
Sgt. Bilko:" not the way I play it"

I was under the impression there are no bots for NL, as obvioulsy they cannot read a players move (bluffing etc)

There are bots for NL?

bots are garbage, they play strictly percentages and win a few cents to a couple of dollars every hour. In order to win money you have to let them run for hour and hours at a time, this is where poker rooms catch them when they play a session for 36 hours. By no means is there any guarantee, you could run a bot for 17 hours and end up a few dollars behind or a few dollars ahead. Bots are easier to beat than players, you just wont make much money off them

There are about 20 NL bots in PA. half and half for tournament and ring. You can modify thier game slightly even without programming experience (tighten them/loosen them). They are position sensitive but they really struggle with limits.. Some will go all the way with mid pair, few put down top pair... they are prone to underbetting strong hands... basically they are too predictable EVEN though the AI in them adapts to your game... In fact it looks like they delibrately let you win the first 200 hands to see how you play sets, tptk etc etc...

You can also set bots against each other for thousands of hands. and see how certain plays work out..... nice toy...

__________________
See me playing $10/$20NL like it was play money

http://video.google.co.uk/videoplay?...405&q=xxdemexx

Doberman: "but Sarge, isn't poker gambling and just luck?"
Sgt. Bilko:" not the way I play it"

To CS guy,

hmmm, yeah, I saw that University of Alberta stuff after a google search (I should have done that first) . I am surprised by the fact that no one thinks there is a good NL bot out there.

I mean if it has the math down, and then you run the algortithm through some kind of speed of play engine to vary the bot, give it a nice set of rules about draws and hand values... Maybe a Rosenblat or better learning algorthim similiar to something I learned in undergrad computational molecular biology.

Main thing... I think it is important that it not be known you were playing a bot. You knew you were, and being CS you would have some insight into maybe beating it.

I think there has to be ((I play like one myself) (no emotional value in hands))
It may just not be for general public consumption.

To whomever cares (hopefully nobody, god), I have CS/ECON degrees, but am currently employeed as a professional tennis player. I also am building a web development startup company. It's a F&*&in great way to pay the bills. I don't REALLY work!

Superstar

winholdem

__________________
I get more ass than a toilet seat. All shapes/colors/sizes.

caution:
http://girlvideos.blogspot.com/

Just to go on...

With chess we knew it was only a matter oftime before the best player was a computer - it happened before 2000.

With backgammon the best player was a bot 5 years earlier (ish)

With "Go" (the Japanese game) I think that there are computer 'players' but te challenge of patterns is beyond what analytical mathematical methods allow.

For example the "travelling salesman problem"- a varient of the p=np problem) - is not really solvable other than by examining all the pathways.. As soon as your nodes get large.. goodbye computer.

Poker shares these feature - recognition of patterns of play and a p=np type dispostion. (i.e lots of ways of doing things for any one hand)..

It is tempting to look at hole cards, position and "rep" (given a mathematical likelyhood) and put it all together and bash out the answer - but actually this sort of problem doesn't yield to this kind of expectation value approach - it looks like it should --- but it doesn't...

This has some equivalence to throwing dice.. there's a feeling held by some that if you knew exacly how hard you throw a die, you knew the ground friction, the size of the die, etc etc - then you should be able to forcast what number will be thrown.. the reality is that you can't... the errors overtake the measurements.

Maths only gives us 5% of the answers - as Fermi said - there is NO reason to beleive that all mathematical problems are linear and can be solved - the reality is that you can set a million and one maths questions that can't be solved.

The "tradgedy" of maths teaching is that it often leads students to beleive that maths has all the answers (there's no point in trying to solve unsolvable questions is there?).. the reality is that this is mathematical propaganda.. maths just aint that good

__________________
See me playing $10/$20NL like it was play money

http://video.google.co.uk/videoplay?...405&q=xxdemexx

Doberman: "but Sarge, isn't poker gambling and just luck?"
Sgt. Bilko:" not the way I play it"