Manipulation Detection in Cryptocurrency Markets: An Anomaly and Change Detection Based Approach

Abstract

As a financial asset, cryptocurrencies innovated the financial industry in different ways. However, the lack of regulations and transparency in cryptocurrency markets is hindering the industry from reaching its full potential. There is a need for extensive technical analysis of the cryptocurrency market data to detect possible market manipulation attempts. Anomaly detection techniques can reveal information about abnormal activities in the market and provide insights on manipulation attempts. In this study, a robust unsupervised anomaly detection tool (ADT) is developed for this purpose. Experiments show that ADT outperforms a set of methods in detecting the anomalies in features extracted from the cryptocurrency exchanges data and on a set of benchmark data sets.