System and method for detecting collusion in online gaming via conditional behavior

A method for detecting collusion in online poker involving a plurality of players, comprising the steps of:

• collecting data on every action in every game for every player and sending the data to an online poker database;
• storing hidden game information data on every action in every game for every player in the online poker database;
• storing game action data on every action in every game for every player in the online poker database;
• performing a player action analysis of correlated actions between a pair of online poker players and building a user action database with data obtained in the player action analysis;
• creating two or more Bayesian Network graphical models including at least one collusional model to represent forms of collusional behavior and at least one non-collusional model to represent forms of non-collusional behavior;
• employing the created Bayesian Network graphical models to determine a likelihood of conditional behavior between the pair of online poker players;
• after determining the likelihood of conditional behavior, computing individual scores for the Bayesian Network graphical models using a Bayesian statistical technique;
• comparing the computed score of a collusional model with that of a non-collusional model;
• after comparing the computed scores, performing a collusion threshold analysis using a thresholding scheme to determine whether the collusional model indicates collusion for the pair of online poker players;
• and
• notifying an appropriate party of unfair collusional play if it is determined that the collusional model is appropriate for the pair of online poker players.