Agent-based modeling of stock market manipulation: an experiment in China

Abstract

This study proposes an Agent-based sTOck Manipulation (ATOM) model based on heterogeneous traders and their behaviours to investigate the stock manipulation scenarios. In the ATOM model, the fundamental traders, technical traders, a manipulator and followers make orders based on the respective trading. The manipulator employs fake limit bids during price pumps to drive up stock prices, followed by market asks during dumps to profit from high prices. Followers track large orders in the order book, with varying sensitivity to trading quantities. In the empirical analysis, we use ATOM model to simulate the manipulation behaviours disclosed by China Securities Regulatory. We also investigate three stock manipulation regulations including order size's upper limit, order submission-cancellation interval's lower limit and soft close. The results show that stricter order size and cancellation limits hinder manipulator profits during pump phases, while a soft close extends trading time but increases uncertainty, impeding manipulator gains.

Keywords: Stock Manipulation, Agent-based Modeling, Experimental Finance, Market Microstructure JEL Classification Codes: G4