Poker program Cepheus is unbeatable

My research group has been developing computer poker applications for many years so it was with considerable interest to see a headline in the Guardian newspaper that read "Poker program Cepheus is unbeatable, claim scientists." Researchers at the University of Alberta in Canada, who have long been the leading research group in computer poker, have used "four thousand computer processors, each handling six billion hands every second. With each game Cepheus [the name of their system] played, the program built up a database of cards dealt, betting decisions and outcomes. At the end of the marathon training session, the database contained 11 terabytes of information on calls, raises and folds for every hand a player could have." It will be interesting to see if Cepheus competes in the annual Computer Poker Competition and if it shows a step change in performance. However, we will have to wait until 2016 for the next competition.

Learning to play poker...

The Game AI group at the University of Auckland have been playing poker since 2007. Over the years a case-based Texas Hold'em poker bot, called Sartre, has been improving its performance both against other poker bots in the Annual Computer Poker Competition and people via our online poker bot.  The online poker bot retains each hand of poker played as a new case and so over time it learns to play against people better as it's experience of the game grows.
    When you play against Sartre you may win several hands in a row, you may even win a considerable number of hands. But, on the whole Satre wins and the graph below proves it. Sartre is now consistently on average winning 0.35 sb/h. It is consistently in profit. Of course this comes with a big caveat - we're not playing for real money and consequently some players tend not to fold when the should because they want to see the outcome of the hand. However, there are plenty of trusted sites like Jackpot Capital online where you can risk you money and test you skill should you want to.