Abstract
Insider trading (i.e., "informed market manipulation") use private information to illegally profit. However, it is not necessary to have trading related information to profit in a stock market. If a trader can generate false information in order to mislead other market participants, he or she can make a profit. This is commonly termed "uninformed market manipulation."
Modelling stock markets provides a means to test market manipulation theories. This thesis presents such market models to describe different aspects of complex behaviour in stock markets and stock manipulation. This research is shown to be the first to characterise trade-based manipulation in a single realistic model of a limit order market.
We discuss the characteristics of stock manipulation by considering trading related information, and present a simple framework for manipulation. Realistic market micro-structure models are presented to characterise stock market and trader behaviour. A belief structure of a stock trader is characterised. This model allows us to control and tune trader beliefs explicitly in order to analyse their learning processes and effects on the order book. Models are then used to explain real stock market behaviour. The impacts of heterogeneous trader types on these models are considered. Finally, stock manipulation scenarios are characterised as external processes and introduced to our computational market, thus allowing stock manipulation models to be built.
Using these manipulation models, simplified formal explanations for manipulation scenarios are presented. The resulting models are used to address current theories regarding manipulation and in explaining the profitability and detectability of manipulation in liquid and illiquid markets.