More About Me

How I Found the Largest AI Bot Ring In Online Poker History

In 2014, I analyzed my poker database to find several players cheating in Pot Limit Omaha games online. My experience finding and bringing this bot-ring to light exposed me to an exciting world of analysis and investigation. I have a talent for unearthing suspicious behavior (one that I now pursue as a data scientist) and this was my biggest discovery.

Let's back up a bit. I had been playing poker professionally for quite some time in 2014. Black Friday had happened years before, and as a result, the online poker games had become extremely challenging. It was an interesting point in the history of poker because it prefaced CMU's Poker Bot.

I knew many of the players because we regularly played each other - we'd dueled it out over the years, and formed lots of mutual respect. We also had a shared interest in keeping an eye out for suspicious players. Yet, everyone everywhere was looking for an edge, and the game was growing increasingly conservative when it came to game selection.

During this time, I noticed some new players from the same location putting in some serious hours, and doing quite well over a few weeks. This was rare because most players slowly move up in stakes. It takes players a while to get accustomed to the better players and higher stakes. Furthermore, these players from the same location had an extremely similar playing style.

What was particularly alarming was that they never exclusively played each other, or played together in games that had very few players.

My intuition was definitely kicking in - alarm bells went off in my head! I figured it'd be wise to take a closer look before I continued to play in games with these suspicious players. Although cheaters in the past were frequently caught (and players reimbursed), this seemed... different. I should also state that I guarantee collusion and cheating affected my bottom line much more than that of the poker sites.

I had caught cheaters in the past because they were fairly obvious about it. They were usually connected socially, didn't play very well, or cheated in rather obvious ways. Note: it is much more suspicious for terrible players to win huge over time than it is for poker pros because losing professional poker players don't last long!

These circumstances seemed different to me because none of these more obvious clues applied. So, I did extensive research into their statistics (I created a database with every statistic you could imagine on my opponents). My jaw dropped as I poured over the numbers - their statistics were eerily similar. Granted, if there exists a game theory optimal Nash Equilibrium which all poker pros attained, this would be likely.

However, the game here was PLO which is extremely far from being solved.

I knew players could still have similar statistics, but it was a stretch. I decided to message the poker site and fellow players. The poker site conducted a 'thorough investigation', but insisted players have similar statistics all the time. I agreed, but persisted there is a significant different between nearly identical and similar.

I provided other insights, and found that the players I voiced my suspicions to were also beginning to investigate and voice their concerns. At this point, my hands were tied.

Eventually, in 2015, my initial suspicions resulted in a collaboration on a popular poker forum. Shortly afterward, this was heavily covered by pokernews and eventually the site made some attempts to admit the reality of the situation. This was the largest AI bot-ring in online poker history and despite very early suspicions from myself and other reputable players, it went unresolved.