AI in games

Every year we learn about astonishing progress in ML and AI. New algorithms beat humans in more and more complex games. In 2016 we saw the milestone of Go game was reached. This year the time has come for poker.

The goal of this competition is to bring attention to ML applications with high information assymetry and high uncertainty. Approaches used in this competition can find fruitful applications in various business problems like pricing, customer value management and many others.

Modern Artificial Intelligence should not only help us with developing algorithms for rational processes, but also facilitate us when modelling irrational behaviour of agents such as found in market competition, or in our case poker playing agents.

Alexander Vedyakhin
Vice President Sberbank

ML problem

Participants are invited to develop a bot-agent that can play Texas Hold’em Poker.

Bots have to make gaming decisions in real time, running in a Docker environment.

Bots play against each other in regular tournaments with hundreads of matches. Leaderboard reflects these tournaments.

Data about all previous matches and tournaments is regularly updated and available for analysis and learning.

Starter code in Python and C++ is available at our Github page, as well as submission instructions.

Sumbission platform Github repo