Simulating Trader Manipulation in a Limit-Order Driven Market

Use of trading strategies to mislead other market participants, commonly termed market manipulation, has been identified as a major problem faced by present day stock markets. Although some mathematical models of market manipulation have been previously developed, this work presents a framework for manipulation in the context of a realistic computational model of a limit-order market. In this work, the Maslov (2000) limit order market model is extended to introduce a manipulator and technical traders. The Maslov model simplifies the concept of a market place for a particular stock by constructing a limit-order book contents which are manipulated by an infinite pool of uninformed traders. The developed model is considered in terms of market dynamics, overall profit, and detectability of the manipulation strategy. Concepts such as the role of supply and demand in determining price direction, how technical traders extract information from trading, and how manipulators use these principles to alter prices are presented via the model properties. Uninformed traders in the original Maslov model do not consider behaviour of other market participants when taking trading decisions. Because of this exogenous nature of uninformed traders, manipulation is not possible in the Maslov model. In contrast, technical traders base their trading on the behaviour of informed traders that are believed to be trading in the market. These technical traders utilise a Bayesian learning model to extract information from informed trading and use that information when making their buy/sell decisions. This behaviour of technical traders makes manipulation possible in our model. The manipulator pretends to be informed and misleads the technical traders in performing a “pump and dump” manipulation scenario. We divided this manipulation strategy into three periods termed, “ignition period”, “momentum period”, and “call-off period”. In the ignition period, the manipulator buys at higher prices (i.e., only market orders) giving an appearance of a higher activity causing the price to go up. This false signal affects technical trader behaviour and their demand further raises the price in the momentum period. The manipulator finishes his manipulation strategy in the call-off period by dumping his shares in a rapid rate causing the price to collapse. The presence of technical trading adds a persistence behaviour (momentum) to the Maslov prices. The manipulator profit is higher when there are more information seekers (i.e., technical traders) in the market and the presence of manipulation reduces the technical trader profits. We also show that the level of information asymmetry between buying and selling makes the manipulation possible and more profitable. Moreover, when technical traders believe that there exists more informed trading in a market, the manipulator effort required to mislead the market is comparatively low and hence the manipulator profit is higher. In future, we are planning to find evidence of real stock manipulation characteristics such as high volatility and low market efficiency in our manipulation model. There is also an opportunity to extend our model to study the dynamics of other price manipulation mechanisms such as “marking the close” and “painting the tape”. Moreover, measures such as volatility (ARCH/GARCH effects), properties of price return distributions, and market efficiency measures such as the Hurst exponent could be used to detect the manipulation period.