A Dynamic Trading Strategy Based on Conditional Value-at-risk

This thesis studies two risk measurement methods, Value-at-Risk (VaR) method and Conditional-Value-at-Risk (CVaR) method. The concepts, prop- erties and calculation methods of VaR and CVaR method are introduced. On the basis of CVaR method, the mean-CVaR model is established. The thesis focuses on modeling a dynamic CVaR portfolio optimization problem based on the dynamic programming method. Moreover, a VaR constraint is added to the model, which strengthens the dynamic CVaR portfolio optimization. Through numerical analysis, the investment risk loss value and the portfolio investment ratio under the relevant confidence level can be obtained. From the results based on the real stock data, it is concluded that the risk of multi-stage portfolio investment is much smaller than that of single-stage investment. Finally, two other methods were selected for comparison. In summary, the CVaR method has a considerable rate of return and moderate risk. The thesis is wrapped up with a conclusion and future work.