Detecting collusive shill bidding

Abstract

Shill bidding is where spurious bids are introduced into an auction to drive up the final price for the seller, thereby defrauding legitimate bidders. Trevathan and Read presented an algorithm to detect the presence of shill bidding in online auctions. The algorithm observes bidding patterns over a series of auctions, and gives each bidder a shill score to indicate the likelihood that they are engaging in shill behaviour. While the algorithm is able to accurately identify those with suspicious behaviour, it is designed for the instance where there is only one shill bidder. However, there are situations where there may be two or more shill bidders working in collusion with each other. Colluding shill bidders are able to engage in more sophisticated strategies that are harder to detect. This paper proposes a method for detecting colluding shill bidders, which is referred to as the collusion score. The collusion score, either detects a colluding group, or forces the colluders to act individually like a single shill, in which case they are detected by the shill score algorithm. The collusion score has been tested on simulated auction data and is able to successfully identify colluding shill bidders.