1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

JV: Poker Bot

Discussion in 'Joint Ventures' started by callmelucid, Oct 2, 2009.

  1. callmelucid

    callmelucid Regular Member

    Joined:
    Feb 15, 2009
    Messages:
    487
    Likes Received:
    446
    Hey guys. I'm a 100nl cash game player coming up from 2nl cash games earlier this year. i run @ 27.7bb/100 at 5NL cash games over 20,000 hands. If you know anything about poker, 5bb/100 is considered crushing a stake at poker so that should tell you how easy 5NL is. what I want to do is this...

    CREATE A POKER BOT THAT CAN UTILIZE A PROGRAM LIKE POKERTRACKER3 OR HOLDEM MANAGER TO CRUSH 5NL CASH GAMES

    i beat 5nl for 27.7bb/100 because it is really easy. all you have to do is play straightforward. this means that my strategies converted to code can have us making some serious dough on autopilot.... literally!

    Think about it:

    Say we make a bot that only makes 5bb/100. We will have 15 tables running for 10 hours a day on two computers. I am assuming each table averages 100 hands/hour which isnt hard to do.

    .05 x 5 x 15 x 10 x 2 = $75 a day doing nothing = $27k+ a year

    Now this is assuming a low winrate and few tables for a few hours on two computers. You can see how this can easily be upscaled.

    Email me if you are interested in designing this code.

    I will write for you the pseudo code on how I play all situations (which will take a while) and you just convert this into code.
     
  2. callmelucid

    callmelucid Regular Member

    Joined:
    Feb 15, 2009
    Messages:
    487
    Likes Received:
    446
    got a couple of pm's.

    bumping for people who might be interested but havent seen this yet
     
  3. voidale

    voidale Jr. VIP Jr. VIP Premium Member

    Joined:
    Sep 29, 2008
    Messages:
    583
    Likes Received:
    176
    just find the other bot that shared here and you can make your own profile... your own tactic
     
  4. callmelucid

    callmelucid Regular Member

    Joined:
    Feb 15, 2009
    Messages:
    487
    Likes Received:
    446
    yeah i tried it out and it was completely moronic w/ the default settings.

    i went to modify it and its not nearly as customizable as i would like.

    i think it at best it would only make a tiny profit. what i'm looking for is something that integrates with a program like holdem manager and plays based on position and hand ranges
     
  5. iglow

    iglow Elite Member

    Joined:
    Feb 20, 2009
    Messages:
    2,080
    Likes Received:
    856
    Home Page:
    i used to play poker a lot and been also looking for such tool. if u suceed it might be a bestseller ;)
     
  6. proexploit

    proexploit Guest

    My only issue:

    Why should we use you as the way the bot plays? Do you think you have perfect play? Surely some of your skill has to be human intuition and not something you can teach a bot. Or does the bot bluff etc as well? The permutations for every situation just seem impossible to actually work out.
     
  7. misterajc

    misterajc Registered Member

    Joined:
    Aug 19, 2008
    Messages:
    71
    Likes Received:
    28
    As soon as it became a bestseller it would stop making money. Programs have predictable strategies, and are therefore easy to identify and beat. The best way would be to keep the program to yourself, ramp up slowly, and don't use it often enough to let others identify a pattern. A friend of mine who plays Internet poker professionally keeps copious notes on all the opponents he sees, so when he plays them again he know how they behave. Any poker bot would have to change identities frequently.

    In developing this, beware of creating a Martingale, that is, a system that offers small steady gains at the risk of occasional huge losses. These are very common in gambling and financial market trading systems. Sometimes they are sold deliberately as a scam, but it is possible to create them accidentally in home made systems.
     
  8. callmelucid

    callmelucid Regular Member

    Joined:
    Feb 15, 2009
    Messages:
    487
    Likes Received:
    446
    bluffing at 5nl is stupid in most cases. people dont fold.

    i'm no poker pro but i have the fundamentals down and play at higher stakes and beat it pretty well.

    no I am not perfect obviously but beating 5nl for >27bb/100 is technically crushing the game.

    the thing about 5nl is it is so straightforward (maybe 10nl or 25nl too but that is when you start seeing more and more competent players). as long as i can put in basic fundamentals that i have learned and studied (which takes a while but is doable), the bot should be able to maximize upon the losing players of that level.

    i dont expect the bot to crush thinking competent players. only a human could do that imo. what my plan to do was play straightforward poker to maximize on bad opponents mistakes, make a slight profit at best against tight opponents (or the few TAG opponents), and the rest is capitalizing on rakeback plus bonuses.

    i have a straightforward strategy that i play with that could be converted to computer code. of course i'm a human so i was able to maximize it and avoid some situations which gave me a >27bb/100 winrate but even at half that, a bot would still be crushing that. even if the bot was 1/8th as good as me using my strategy, it would be beating the game more than most people. now multiply that by 15 tables on a computer times however many computers are running.

    of course once we make the software, it will not get released (or get released to a select few for big $$$ so poker sites dont catch on).

    the spot is still open if anyone is interested. got a few pm's but a few people havent responded back to me and a few people didnt quite seem to have what i was looking for.
     
  9. callmelucid

    callmelucid Regular Member

    Joined:
    Feb 15, 2009
    Messages:
    487
    Likes Received:
    446
    ya the bot would obviously be exploitable lol. i would help counter it slightly by varying the bet sizing and a few other things but ultimately it is exploitable. that is part of my decision to run it at 5nl or 10nl. i dont think it is possible to run a home made bot against a table full of competent or agressive players. if there are 1-2 competent players at a table and the rest are either rocks or calling stations, then a bot is going to make money.

    as for the martingale situation. that is something that i was thinking of. i have already thought of a few ways of stopping it (which is methods i use such as pot control, players never bluff on the river at micros, etc). of course it will decrease the bots variance but in the end it will stop big losses. thanks for the recommendation!
     
  10. heiska

    heiska Junior Member

    Joined:
    Dec 5, 2008
    Messages:
    138
    Likes Received:
    169
    All I can say here is that you might want to re-read PPL manual included in the download.

    The bot has played based on your position and hand ranges since version pre-alpha beta.
     
  11. callmelucid

    callmelucid Regular Member

    Joined:
    Feb 15, 2009
    Messages:
    487
    Likes Received:
    446
    i'll relook that over. the biggest thing i want is it to work off of players statistics too (vpip/pfr/af/3b%/Fold3b%/etc)

    example:

    you have 8s9s on button

    an 8/4 with a fold 3bet% of 0 i'm calling
    a 20/17 with a steal % >30 and a fold to 3bet% >60 i'm 3 betting

    so maybe if anyone knows how to integrate the current bot w/ Holdem manager, PT3, or a custom statistics mysql database
     
  12. heiska

    heiska Junior Member

    Joined:
    Dec 5, 2008
    Messages:
    138
    Likes Received:
    169
    The problem is (concerning <NL10 tables mostly) is that most of the fishes out there have no clue about poker and they play with whatever hands they feel like --> your suggestion wouldn't work.

    Also, on >NL10, most of the people can play poker and are mixing up their gameplay between LAG/TAG etc --> collecting data off players would again prove itself useless.
     
  13. callmelucid

    callmelucid Regular Member

    Joined:
    Feb 15, 2009
    Messages:
    487
    Likes Received:
    446
    what you said about <NL10 tables is a good thing though. every time you raise a limper and you have ak and they call with anything less, you make money. if they fold, you make money. even if they call with a hand like 66 but fold to cbet is high, you literally make money by just doing a 1/2 pot to 3/4ths sized bet every time despite board texture. of course you would have to set in variables dependent on their statistics but it is very simple (but tedious) if you know how to play on different board textures and vs different opponent types

    where you lose money is if the flop comes k24 rainbow and you get check/raised. bad players are calling here. the bot would need to understand board texture and equity (of course stats would have to take play into this, stack sizes, hand ranges, etc).

    its very complex but possible. its not just simple raise jj+ aq+ and get it all in every time. i'm skeptical that that makes money at all.
     
  14. iglow

    iglow Elite Member

    Joined:
    Feb 20, 2009
    Messages:
    2,080
    Likes Received:
    856
    Home Page:
    hey of course martingale sucks everybody knwos it ;)
    and indeed tool stops working as soon as its bestseller. bbut why would somebody who made money already + people who made money while it worked, care in the long run
    its like that with all techniques, they work on start later they get blocked
     
  15. vivalafeva

    vivalafeva Jr. VIP Jr. VIP Premium Member

    Joined:
    Sep 16, 2008
    Messages:
    1,481
    Likes Received:
    719
    ha if u wanna sell a few copies when u get it done, hit me up