Poker News

After over a decade of effort, scientists at the University of Alberta believe they have achieved their goal of creating the “perfect” poker playing program, one that doesn’t make any bad moves and is a winning “player.”

According to a study released in Science magazine, those University of Alberta scientists think they have built a program named Cepheus (a former king of Ethiopia whose daughter, Andromeda, figures prominently in Greek mythology) that is perfect in its play of the game of poker. Using statistical analysis and pre-determined strategies, Cepheus reportedly could play heads-up poker against another player for a lifetime and come out, at the minimum, a break-even player. “We’re not saying that it’s guaranteed to win money on every single hand,” says Michael Bowling, a computer scientist at the University of Alberta and a co-author of the study. “What we’re saying is that, in the long run, if you looked at all the hands that could happen and you averaged all of those, then the computer can’t be losing, at a losing rate — it has to be either breaking even or winning.”

The study points out that many games, such as chess, checkers or tic-tac-toe, have a finite number of outcomes that can occur. These “perfect information” games are fairly easy to build a computer program around that can, in essence, defeat a human every time. With poker, however, other factors come into play. Poker, as an “imperfect information” game with such moves as bluffs in play and not knowing what cards the opponent holds, is more difficult to build a computer program around. “We had this training phase where the program started off playing uniform random against itself, meaning that (Cepheus) had no idea what it was doing other than following the rules of the game,” explains Michael Johanson, another computer scientist also at the University of Alberta and Bowling’s co-author of the study.

As time went on, however, Cepheus began to update the information in its databanks and improved its play. Instead of making random plays, Cepheus apparently began to incorporate poker strategy into its processes. Depending on the point in a hand, Cepheus could allegedly calculate what effects a raise would have – instead of a random play – and whether it was positive expected value (+EV) or less (-EV). By utilizing its experience – something that the scientists called “regret value” at not making a particular move – Cepheus has apparently become a computer program that will always make “the perfect play,” according to Bowling.

If you’re thinking about putting together Cepheus for yourself, you’d better have a large basement. The study indicates it took nearly 70 days for the program to reach its “perfect” phase, requiring 200 computers with 32 gigabytes (GB) of RAM and 24 central processing units (CPUs). The scientists indicate they could continue to train Cepheus but, at this point, it would be useless as the program has reached as close to perfection as can be expected. “Even if the program spent a lifetime in training, it is so close to perfect now that it wouldn’t have much value other than academic novelty,” Bowling stated.

On another note, Cepheus was able to demonstrate academically and mathematically something that has always been a basis for profitable poker play. During the study, Cepheus demonstrated that the dealer has the advantage over its opponent, but the advantage is miniscule. “We can now prove that the dealer has an advantage of what we call ’88 millablinds’ per game,” Johanson said. “That’s .088 of a big blind per game,” Johanson reported.

For those worried that online poker will be invaded by Cepheus-playing poker bots, there isn’t much concern at this time. Cepheus is only programmed for Texas Hold’em and, at that, it is the Limit game, long known as much more of a mathematical game. It also only works in a heads-up situation, meaning the six or nine-handed tables that cash games take place at (the program wasn’t built for tournament play) seem to be safe for now.

The scientists at the University of Alberta have been working on the concept of a “perfect” poker playing program since 2003. After some trial programs failed to defeat their human opponents, the Alberta team came up with one in 2008 that was a breakthrough. That program, named Polaris, did lose to professional poker players Ali Eslami and Phil Laak but, after some tinkering by the scientists, came back a year later against the same two men and won three of six matches and tied once for an overall record of 3-2-1.

Where the “perfect” poker playing program goes from here is up to the masterminds at the University of Alberta. Could their next challenge be “cracking the code” of No Limit Texas Hold’em? Or will they move onto a full-table challenge rather than heads-up matches?

Leave a Comment

Your email address will not be published. Required fields are marked *