On-Chain Analysis-Based Detection of Abnormal Transaction Amount on Cryptocurrency Exchanges

Cryptocurrency exchanges play an indispensable role in the cryptocurrency market. However, some exchanges are suspected to be involved in various abnormal or malicious behaviors while providing services to users, such as money laundering, wash trading and even running away. Besides, these behaviors are reported to be often accompanied by an anomalous increase in the transaction amount. Therefore, it is a topic worthy of study to detect whether the abnormal transaction amount occurs in the exchange and when it occurs. This paper uses web crawler tools to collect a relatively complete dataset of exchanges and then conducts a correlation analysis to obtain the most important factors that influence the transaction amount of different exchanges. Then, the prediction model of the influence of various factors on the transaction amount is obtained based on deep learning. The deviation between the predicting transaction amount and the actual transaction amount is calculated to provide a basis for abnormal transaction amount detection. Finally, through a case study on the detection results, some abnormal transaction amounts are related to policy changes and industry events, while the others are suspected to be related to illegal behaviors