Wash Trading Detection Techniques for Centralised Cryptocurrency Exchange Services

Abstract

Centralised cryptocurrency exchanges are often a mandatory first point of entry for most blockchain users. This means that such services have to compete for user attention, often by boosting their attractiveness through questionable behaviours. One of such practices is wash trading, i.e. injecting fake trades to artificially boost the service statistics to portray an unreal users engagement. To protect inexperienced users from dishonest services it is then paramount to develop a set of techniques to identify fraudulent behaviours. This is why, in this paper, we propose a set of automated, yet intuitive analysis that may hint at possible misbehaviour. The goal of these techniques is to be intuitive enough to be understandable by inexperienced users. To this aim we present our experimental results on the real world data of two exchanges, one considered honest and the other suspicious. The outcome is encouraging, as it reveals how dishonest behaviour can be macroscopically detectable, even by the considered intuitive tests.