Risk adaptive trading using technical indicators and exponential decay

Risk taking behaviours perform much better than risk averse behaviours in rising market conditions, while the inverse is true in falling market conditions. Applying on the stock market, these behaviours can be modelled using risk sensitive rein- forcement learning techniques. These modelled behaviours are called risk models. Because market conditions do not stay constant, individual risk models do not produce consistent performance through an extended period of time. However, the same could not be said if these models are used interchangeably. This is due to the fact that each model excels in specific market conditions. The objective of this research is to propose a risk adaptive trading system that is capable of identifying the market condition and selecting the most suitable risk model to perform trading. A novel method of combining technical indicators to increase the reliability of identifying market conditions is presented. Based on this, the most suitable model is selected to conduct trading. Experiments were conducted and results have shown that the proposed risk adaptive trading system has the potential for significant returns. The risk adaptive trading system is also shown to be accurate in its selection of the most suitable model. Furthermore, using existing finance theories to validate the proposed method of combining technical indicators, this research then presents topics that could be explored further.Bachelor of Engineering (Computer Science