Stock market manipulation detection based on trade data

Abstract

Stock market is a mechanism by which investors trade for stakes or other financial securities of publicly traded firms. While the ideal and fair market would be where the prices and thus the investors' profits are determined by supply and demand of shares according to the fundamental information which is publicly available to all investors. However, overtime many activities have been identified which are carried out with an intent to deceive other investors and earn profits through that. Such an activity is termed manipulation. It's the responsibility of regulatory bodies like SEBI in India to identify and penalize the manipulators to provide a fair market for investm ent. With the advent of highly advanced algorithmic trading, the issues of being able to detect manipulative actions has become much more significant. Through this project we aim to understand the manipulative actions and classify them based on the way it cheats the fair market mechanism. Having got an overview, we chose Cycle trades in particular and study the procedure used to detect such manipulation. Understanding major steps of the process, we present the outputs of the implemented python script used over one day trade logs of INFIBEAM script. Finally we comment on a few recommendations to improve certain aspects ofthe process.