PTR Uncovers Bot Accounts on PokerStars

Following a 2+2 post, PokerTableRatings.com (PTR) discovered 10 PokerStars accounts being bots. These have been frozen by the online poker room upon evidence provided by PTR.

PTR announced that, according to their investigation, the programs played around 8,320,000 hands in $0.25/$0.50, $0.5/$1 and $1/$2 no-limit games, generated a rake of $186,572 and a profit of $57,839 in total. The site based its assumptions on their own analysis of the statistics of the 10 accounts (7emenov, bakabar, crazier, mvra, nakseon, kozzin, demidou, koldan, Daergy, feidmanis) which showed more than suspicious similarities in the changing of stakes as well as “a unique and easily identifiable betting pattern from the big blind in unraised pots.” To see their full analysis, visit PokerTableRatings.com.

Stats of the 10 bots on PokerStars (source: PokerTableRatings.com)

Researching poker player computer programs has been under ways for some time now, most notably by the Computer Poker Research Group (CPRG) of The University of Alberta, the same institution that brought us Chinook and Deep Blue, the bots that ‘solved’ checkers and chess. As opposed to these games, however, poker “is a game of incomplete information,” as The Economist put it. This makes it way harder for a program like CPRG’s Polaris to compute the possibilities and also limits the game types it understands (in Polaris’ case, heads-up limit and no-limit Texas Hold’em).

Although CPRG head Michael Bowling stated that the real goal was to “study how to build computers to make decisions in difficult circumstances, especially where there is missing information,” the research focuses on building a “winning poker program.” However, if anyone wanted to put such a thing into use in an online poker room, it could prove to be even more challenging. Was it not hard enough to create a bot capable of standing up against top-level players (as Polaris did against Phil Laak, for example) as well as shifting smoothly between various game types, programmers also had to face the difficulty of bypassing online security systems that seem to work just well enough. In the end, it all comes down to what Laak said: “anyone smart enough to put a bot down would make way more money operating above board.”

Read more about Polaris at The Economist.