Computer Poker Bots

This page contains several automatically playing poker agents, which were created during past practical courses (Computer Poker Challenge SS 08 and Computer Poker Challenge SS 09) held at our group. They are all based on the Pokerserver framework by the Computer Poker Research Group of Alberta, which is also the official framework for the annual Computer Poker Competitions.

Some selected (ro)bots in the following list participated in the annual Computer Poker Competitions and achieved in total the 2nd place in the 2008 6-Player Limit Competition and the 3rd place in the 2009 3-Player Limit Competition.

FAQ - Poker in AI research (in German)


All bots are programmed in Java and require the Pokerserver Framework (version until 2009 or newer versions), .

If you are unfamiliar with the framework, you should have a look at it, first.


Below you can find various different bots. For each bot, two packages are provided. Besides the full source code, the "Bot Package" contains only the executable for the bot, which can be simply put in the bots directory of the Pokerserver Framework.  Furthermore, a short description of the used approach is stated.

Please note that for some bots the documentation may be incomplete or missing. In addition, some bots are somewhat premature due time constraints in the past for the competition and especially bots which rely on an offline learned model could have been significantly improved with a longer training phase and greater computational power. However, the bots might be still useful for benchmark purposes.

6-Player Limit Texas Hold'em (2008)

In our first practical course Computer Poker Challenge SS 08 in 2008, we focused on the newly introduced 6-Player Limit Competition, such that following bots are only tested and optimized for this format.

Bot name
Bot Package Source Code Description
HokusPokus HokusPokus.tar Pre-flop play is determined by a starting Hand-Chart combined with a learner.
Post-flop play uses Opponent-Modelling by estimated Hand-Ranges and action frequencies combined with Monte-Carlo Simulation.
aethon aethon.tar uses various adaptions of the UCT-algorithm to Poker
mcBotUltra mcBotUltra.tar uses various adaptions of the UCT-algorithm to Poker
bbhardbot bbhardbot.tar based on reinforcement learning
BrainBot BrainBot_pkg.tar Monte-carlo simulation with opponent-modelling by regression and neuronal networks
AKI-Realbot AKI-RealBot.tar Monte-carlo simulation with opponent-modelling by action frequencies. Explicit focus on dynamic exploiting adaptions

Credits go to: Timo Bozsolik, Bastian Christoph, Thomas Görge, Benjamin Herbert, Michael Herrmann, Stefan Lück, Alexander Marinc, Lars Meyer, Arno Mittelbach, Kamill Panitzek, Immanuel Schweizer, Michael Wächter, Claudio Weck and Marian Wieczorek.

3-Player Limit Texas Hold'em (2009)

In our second practical course in 2009, the game of choice was 3-Player Limit Texas Hold'em. This time, each group implemented two bots. The first one was to implement a simple mathematical fair bot, i.e. to play according to some known concepts like pot-odds without any opponent modelling, in order to get used to the framework. For the second bot, every group could follow own directions, though we restricted ourselves to simulation-based approaches.

Bot name
Bot Package Source Code Description
flapyourwings flapyourwings.jar Monte-carlo simulation with opponent modelling using "effective hand ranks"
allineq allineq_player.jar "mathematical fair bot"
akuma akuma.jar Monte-carlo simulation with opponent-modelling by hand-ranges and action frequencies.
dpp dpp.jar "mathematical fair bot"
mabuse_sim mabuse_sim.tar hybrid system: monte-carlo simulation and mathematical fair actions guided by a strategy switcher
rebot_player rebot_player.tar "mathematical fair bot"

Credits go to: Damian Czarny, Thomas Hartmann, Sebastian Kasten, Steffen Remus, Daniel Schumann and Markus Zopf.

Terms of Use

All software on this site is freely available.​ Neverthe­less, we would be glad if you would cite this site ( if you use in some way the provided bots. Since all bots are based on the pokerserver framework, please notice also their copyright statements.

Further Material and Links

  • Annual Poker Competition Homepage  until 2010 and from 2011 on
  • Computer Poker Resarch at University at Alberta
  • You can find some relevant literature regarding Computer Poker on the pages of the practical courses (2008 & 2009).
  • A description of the 2nd Place Bot from 2008 AKI-Realbot:
    Immanuel Schweizer, Kamill Panitzek, Sang-Hyeun Park and Johannes Fürnkranz, An Exploitative Monte-Carlo Poker Agent, in: Proceedings of the 32nd Annual German Conference on Artificial Intelligence (KI-09), pages 65--72, Springer-Verlag, 2009
    A Technical Report Version can be downloaded here



small ke-icon

Knowledge Engineering Group

Fachbereich Informatik
TU Darmstadt

S2|02 D203
Hochschulstrasse 10

D-64289 Darmstadt

Telefon-Symbol+49 6151 16-21811
Fax-Symbol +49 6151 16-21812

A A A | Drucken | Impressum | Sitemap | Suche | Mobile Version
zum Seitenanfangzum Seitenanfang