Fund Risk Measurement Of Time-Varying Copula Model Based On MCMC Method

For financial time series is often appear time-varying,peaks,fat tails and volatility clustering features,in order to accurately describe the dependent relationship between financial time series,accurately measure the risk of financial assets of time series,this paper first using GARCH model and ARMA model for a single fund yield sequence modeling,respectively using ML method and MCMC method to estimate model parameters by AIC criterion to choose relatively optimal GARCH(1,1)-ST model;Measure the VaR value of fund sequence based on ML method and MCMC method;Furthermore,the optimal time-varying Copula model is constructed by parametric method and semi-parametric method to accurately describe the dependency relationship between fund sequences.This paper constructs the time-varying Copula model based on MCMC method through two key steps.Firstly,the closing price data of 1546 trading days of the four funds were selected,and the logarithmic return rate was taken as the research object for statistical analysis.It was found that the series presented time-varying,peak,thick-tailed and volatility clustering characteristics.The logarithmic return rate data of the four funds were selected.Through the test of normality,it is found that the four groups of fund return rate series do not obey the normal distribution.Then the optimal Copula model is obtained by comparing the marginal distribution with the empirical distribution function or the kernel distribution estimation and the marginal distribution with the parameter estimation.ARMA model and GARCH model were used to model the yield sequence of a single fund,and the model parameters were estimated based on ML and MCMC methods respectively.The VaR value of the fund return sequence is estimated,and the conclusion that MCMC method is superior to ML method is obtained.Secondly,in order to accurately describe the dependent relationship between fund yield sequence,using common static copulas connect and time-varying copulas connect model two dependency relationship between the fund series modeling and comparison analysis,it is concluded that the time-varying Sjc-copulas connect model based on MCMC method relatively optimal,to depict the tail dependence relation between fund yield sequence of t-copulas connect model has good performance and can describe the dynamic process of copulas connect model is superior to static copulas connect the conclusion of the model.The empirical study shows that the time-varying Copula model based on MCMC method can achieve better results in describing the dependency relationship between fund return sequences and measuring the VaR value of fund risk.