Optimal Execution Strategies: A Computational Finance Approach

Abstract

In today's competitive business environment, strategies relating to market forecasting, decision making and risk management have received a lot of attention. The empirical results reveal that the market movement is not neutral to large orders. This makes the investors suffer from prohibitive execution costs. Hence, effective optimal execution strategies that assist investors on controlling market reaction are desperately demanded. However, most existing methods for analysing these strategies suffer from a serious weakness in that they fail to consider the impact of large orders on market price. In this thesis, the analysis of optimal execution strategies is conducted from the perspective of agent-based computational finance.

This thesis introduces an artificial stock market composed of agents assigned with information sharing and trading strategies, and analyses the market impact and reaction when agents are assigned with optimal trading strategies, including minimum risk volume-weighted average price (VWAP) and implementation shortfall (IS) strategies. In addition, refinement has also been made to the IS strategy by replacing the linear temporary impact function with a quadratic one.