Risk Measurement And Dynamic Correlation Analysis Of Portfolio Based On Time-Varying Copula

With the rapid development of the capital market,investors have become increasingly aware of investment and wealth management,and financial products have become more and more popular.The current international situation is undergoing complex and profound changes.The trend of multipolarization and globalization has become more pronounced.The negative effects of economic globalization have gradually become prominent,and the financial risks in the investment process have also increased dramatically.Portfolio theory holds that investors can reduce risk by investing in different financial assets that are less relevant.Therefore,the analysis of the correlation between different financial assets has important implications for selecting an effective investment portfolio and reducing financial risks.The financial rate of return series has the characteristics of time-varying and volatility.Copula can be used to connect marginal distribution functions to form a flexible and diverse multivariate distribution function,and it is widely used in the field of financial income sequence correlation and risk measurement.At present,experts at home and abroad have conducted a variety of explorations and researches,mainly through the establishment of various types of yield models to understand the relationship between investment assets,thus calculating VaR to measure financial risk.In this paper,we select an asset portfolio consisting of three representative financial assets: stock,bond,and gold,using a different time-varying Copula function to measure the risk of the asset portfolio,and comparing the results of the fitting.At the same time,this paper analyzes and studies the dynamic correlation between the three financial assets.Firstly,according to the thick-tailed,asymmetrical characteristics of financial asset returns,the GJR(1,1)-biased t model was chosen to fit the marginal distribution function of the rate series.Then use the binary Gaussian-DCC Copula function,the binary t-DCC Copula function and the binary time-varying SJC Copula function to describe the correlation structure between two financial assets.After analyzing the fitting effect of different models,it was found that the binary t-DCC Copula model had the best fitting effect on the yield series correlation of stock-bond and bond-gold;As for the stock-gold yield series correlation,the Gaussian-DCC Copula model has the best fitting effect.Observing the dynamic correlation coefficient,we can see that the stock-bond yield series,the bond-gold yield series are negatively correlated,and the stock-gold yield series is positively correlated,and the correlation between bond and gold is stronger than the other two.Then we use the Gaussian-DCC Copula function,the t-DCC Copula function,and the C-Vine Copula function to characterize the related structure of the portfolio of three financial assets.According to the AIC and BIC criteria,it is concluded that the t-DCC Copula function has the best fitting effect on the portfolio.Finally,using Monte Carlo simulation to randomly generate random vectors obeying the t-DCC Copula model.According to the definition of VaR,calculating the VaR value of asset portfolio under 95% confidence levels,and the conclusion that the portfolio can effectively reduce investment risk has been proved.