Research on Portfolio Risk Prediction Based on Copula-GJR-Skewt Model

For risk prediction of diversified investment portfolio, we use the thick tail and the biased characteristics of GJR-Skewt model to depict a single asset and using Copula model to depict a diversified investment portfolio non-linear correlation structure, simulating the random distribution of financial assets with Monte Carlo method and combining with rolling time window method to conduct the sample dynamic forecast for the future portfolio risk.The empirical results show that Copula-GJR-Skewt model can achieve satisfactory results of predicting the risk of asset returns.For the VaR forecast performance, we use the GJR-Skewt model as the edge distribution functions and even if there is a system error, it can also achieve optimal prediction.