The PokerTableRatings.com database has access to billions of individual hand histories and has compiled over 400 individual statistics on millions of online poker players. Using this database, we have developed a unique method of detecting groups of automated Poker Bots based on statistical analysis of these players.
Bot detection is a 4-step process:
| 1. | PTR has collects hand histories from the major poker sites (over 8 billion and counting). The hands are stored in our massive database cluster. | |
| 2. | We use a mathematical grouping algorithm based upon select stats that are highly indicative of a particular play style. This process requires intensive processing in order to compare large numbers of players to one another over long periods of time. Each player is analyzed and placed in a group. During the process, very few groups grow beyond a single player. Some groups, however, contain many players which is quite abnormal even when analyzing huge numbers of individuals. We call these "suspect groups". | |
| 3. | Each suspect group is analyzed based upon a wider range of stats than originally used to group them to verify that the suspect play is consistent over all statistics we record. Any players whose play is shown to be different by this analysis are removed from the suspect group. | |
| 4. | We observe individual hand histories of these accounts in order to find abnormal tendencies that are consistent across the group of accounts. This provides additional proof that the suspect accounts are being operated by the same computer program. |
We have found this process to be extremely accurate in identifying groups of robotic players. For more information, please feel free to contact us here.
