There are two structures that comprise an opponent model, the weight array and the action frequencies, and both are updated by the learning portion of the opponent modeling system. The weight array is actually a model of the opponent's hand; it is reset at the beginning of each hand and contains information pertaining to what cards the opponent may be holding. It is used by the hand evaluation system to give more accurate estimates of hand strength. The action frequencies are inter-game statistics on how many times the opponent has taken each action in a given situation. They are used by the learning system to appropriately update the weight array, and they could be used by the betting strategy system.