Poker-AI.org

Hi everyone,

this is my first post here and I hope its in the right topic. I am currently working on my master's thesis and since I played poker semiprofesssionally for some years and my major is machine learning, poker AI seemed like an interesting choice. Basically I want to do what ibot has done in his paper, but focus more on the opponent modeling part using different ML algorithms. For evaluation, I want to have my bot compete against several other bots.

However, I searched the academic literature, the web and this forum and found quite an overwhelming stack of different approaches, but nowhere the practical implementation of a bot in some programming language is reasonably explained. I found the meerkak API and also opentestbed, but I'm not so familiar with Java and could not get it running yet.

Therefore, my question is what are the first steps you take when you create and test a new bot? In which environment do you develop and run the simulations in? Further advice is highly appreciated!

Thans in advance!
cco

I used Poker Academy to start. They've gone bust but there might be cracks about. There is also http://www.poker-genius.com/

I use my own environment now

I'm told there is a bot in opentestbed There are some here too https://github.com/corintio/opentestbed

When writing a bot the first thing to do is decide the approach the bot should take. see http://jonathan-rubin.com/files/CPReviewPreprintAIJ.pdf for an overview of approaches.

It doesn't sound as if you are interested in writing a competitive bot so you could just use something freely available such as MCTS or INOT, or even the PA bots.