This thesis mainly discusses the dynamic portfolio of financial markets. A single-cycle dynamic log-optimal portfolio model was established by taking magnification-risk function as the revenue target, value at risk (VaR) control function as the risk indicators. Besides, A multi-cycle dynamic log-optimal portfolio model based on (VaR) risk control was also set up by applying dynamic programming methods, which analyzes the nature of the model and proves the unique existence of the model's optimal solutions. Empirical study about the two models has been done respectively. At last, a conclusion will be made that multi-cycle model are superior to single-cycle model in terms of the VaR risk and returns by comparing and analyzing single-cycle and multi-cycle models. |