While the financial market is becoming more and more volatile, the risk underlying is becoming more and more complicated as well and people have to reconsidering the importance of financial risk. Hence financial risk management is the critical issue of modern economy and finance research. The purpose of financial risk measure is to evaluate the degree of risk in the future numerically based on the financial time series, in order to forecast, control and avoid financial risk. As the fundamental part of risk measure, the axioms is getting improved in recent years. However there are defects in two aspects: first, the is rarely research about the real risk experience of investors in the risk measure axioms; second, the lack of comprehensive consideration of nonlinear transaction cost will result in the misevaluate of risk.Considering the insufficient above, the dissertation firstly focuses on improving the risk measure axioms. Based on the analysis of current risk measure axioms new axioms are proposed which require the risk measure can reflect the asymmetric risk attitude toward different distribution, and can reflect the impact of nonlinear transaction cost on the integrated risk evaluation. The new risk axioms represent the asymmetric risk attitude of investors with local subadditivity and local superadditivty, and represent the impact of nonlinear transaction cost with super homogeneity.Secondly, this dissertation puts forwards two new risk measure function under the new risk measure axioms formerly proposed. On one hand, this disseratation propose a new liquidity evaluation method to adjust the asset return, which is the basic statistics of risk measure. On the other hand, relevant functions are adopted to figure out the bilateral deviation around the reference point in order to connect the asymmetric risk attitude of investors to final result of risk measurement.Thirdly, based on the new risk measure function, new portfolio optimization models are constructed and corresponding solving method are proposed as well. Then the solution sensitivity anlysis is carried out with different paramter settings, according to which the impact of risk attitude, investment volume etc. on the optimized portfolio is investigated. And the research above also indicates that the axioms and corresponding risk measure functions proposed in this dissertaion are feasible and practical.Fourthly, the impact of risk attitude of investors, reference point and investment volume on portfolio efficient frontier. When investment volume is large enough, the transaction impact on asset return should not be neglected, hence the analysis is carried out with numeric method. Other wise, the transaction impact can be ignored with less investment volume. In this situation the portolio optimizaion models reduce to simple forms with explicated solutions, which are more conveient to analyze and compare with other portfolio frontiers.Finally, a new asset pricing model is reduced according to the solving process of mean-risk model. The new asset pricing model is based on the risk measure function proposed formerly so as to reflect how asymmetric risk attitude of investors and liquidity deciding the asset return. The following emprical research of asset pricing model indicate that the asymmetric risk attitude and liquidity can explain the asset return effectively, which proves that the risk measure axioms and function proposed in this dissertation are reasonable and practical. |