Anti-collusion software

[Derek in NYC]

It occurs to me that datamining software could be written to analyze possible collusive play. Collusion online can take two forms: (1) sharing of hole card information via instant messenger or telephone, and (2) team play. The first type of collusion is likely to be harder to detect.

Both methods of collusion should be characterized by an abnormally frequent occurence of some of the following factors: (1) repeated or extended play by the same 2 or more players at the same table, (2) repeated involvement in common hands between the suspected colluders, and at least 1 third-party, (3) bet-raise-reraise patterns to trap the field; and (4) river, or possibly turn, folds in large pots.

Are there other patterns that would be relevant, and does anybody think it is feasible/worthwhile to develop such software?

[housenuts]

i assume the big poker sites already employ such software.

[Terry]

I'll buy it.

[stigmata]

I remember a certain untrustworthy person mentioning that they have hundreds of accounts (which was probably bull).

I guess this can be used to stifle collusion detection, as you can just keep mixing and matching accounts so that the same 2 rarely play togethor. If you were a very well organised high-stakes scamming operation, I guess you would go to such lengths.

Anyway just a thought
Not sure how you get round points 2, 3 and 4? Team play?

[Peter]

[ QUOTE ]
(2) repeated involvement in common hands between the suspected colluders, and at least 1 third-party, (3) bet-raise-reraise patterns to trap the field; and (4) river, or possibly turn, folds in large pots.


[/ QUOTE ]
These points all assume both players will stay in the hand together more often than two random players will. There are subtler ways to collude without any of the 3 points above being true.

Peter