The topic of my thesis is the development of a poker agent. Contrary to popular belief poker is not a gamble if somebody plays enough games. I have chosen telesina poker because it is quite simple and not very well known so there is no related software, bot or education material on the internet.
My goal was to develop a competent agent with which the game is a challenge. Playing with the bot we can develop our own strategy which hopefully will be then profitable in the practice as well. But also planning and creating the program itself was among my goals so that I can gain experiences.
I describe the game theory background of poker. Through a highly simplified poker game I demonstrate this approach and calculate the optimal strategies.
The optimal strategy is exploitive but is only maximally exploitive if the opponent is playing optimal as well. That is why it was not my goal that the agent plays optimally in the sense of game theory. The goal is rather that the agent can adapt and take advantage of the faults of the opponent. This so called exploitive strategy is demonstrated and used for the development of the agent. In connection with this I introduce some important terms like „pot odds” and show their use in open card situations.
I have created four components:
• Winning chance calculator: calculates the winning chance of the players who are still in the game only based on their cards supposing that they go all-in.
• Telesina manager: creating statistics and charts.
• Simulator: carries out the games of the bots against each other.
• Bots: poker agents which are connected to the simulator through an interface.
I have created five different bots with different, more and more subtle strategies: from the random player to an agent which is realizing so called third level thinking. I have tested the bots in a lot of games against each other and created charts about the results. Finally in the last chapter I discuss the possibilities of further development of the agent.