Machine Learning Based Approach for Anomalous Pattern Detection in Stock Market

Abstract

Stock price manipulation refers to the activities of traders who carefully designed trading behaviors to manually push up or push down the equity prices for the purpose of making pro ts. Manipulation is an unraveled issue for both developed and emerging stock markets. Since prices of the stocks must be determined by the market without any interference, detection manipulations is really important for a fairly & orderly market. The existing methods in the industry for detecting fraudulent activities in securities market rely heavily on a set of rules based on expert knowledge. The securities market has deviated from its traditional form due to new technologies and changing investment strategies in the past few years. The current securities market demands scalable machine learning algorithms supporting identi cation of market manipulation activities. Here we have proposed a novel approach for detecting stock market manipulation based on a machine learning techniques. Our focus relies on detecting Pump & Dump manipulations, which can be considered as one of the most occurring type of manipulation. For achieving this target, we use supervised learning algorithms to identify suspicious transactions in relation to pump & dump manipulation in stock market. We monitor indexs average return, volatility of the price for a pre-de ned monitoring period and calculate a ratio related with small buy order & large sell orders. We use a data sample, which contain nearly 100,000 of tradings, & which compressed into 125 of feature vectors. We adopt Naive Bayes and SVM for clas- si cation of manipulated samples. As an optimization, we apply bootstrapping techniques for enhance the results. Empirical results show that Naive Bayes out- perform SVM in Sensitivity, F measure & ROC area achieving 100%, 88% & 98% respectively & SVM outperform Naive Bayes in Precision & Speci city achieving 90% & 97% respectively. i