Research On Value At Risk Of Open-End Fund Based On Monte Carlo Simulation

With the rapid development of the Internet,the number of people choosing openend funds as financial management tools has increased year by year.Therefore,the measurement of open-end fund risk has become a research hotspot.In recent years,the value-at-risk VaR has become a common metric to portray the value-at-risk of funds.The common methods to calculate the VaR value of funds include historical simulation method,Monte Carlo simulation method and GARCH model parameter method.The traditional Monte Carlo simulation method calculates the VaR with the default time series obeying normal distribution,but in reality the fund return series do not necessarily obey normal distribution,and most of the time there are spikes and thick tails,so the traditional Monte Carlo simulation method needs to be optimized and improved.On the basis of Monte Carlo simulation,this paper introduces MC-GARCH-VaR model,which capitalizes the GARCH model to characterize the volatility aggregation of fund returns and eliminates ARCH effect to improve the accuracy of the model.In order to optimize the parameter estimation of GARCH model,Markov chain Monte Carlo method is introduced to construct MC-GARCH-VaR model,which not only improves the accuracy of VaR calculation compared with maximum likelihood estimation method,but also solves the defects of Monte Carlo simulation method in high dimension and static state.The details are as followsx:The return rate of "China Merchants Liquor Index(LOF)(161725)" fund from March 28,2018 to March 24,2020 is taken as the training set to analyze the fluctuation nature of its return rate.Then the data from March 25,2020 to January 15,2021 is used as the test set.The historical simulation method,traditional Monte Carlo simulation method,GARCH model parameter method,MC-GARCH-VAR model and MCMCGARCH-VaR model are used to estimate the VaR value of "Investment China Certificate Liquor Index(LOF)(161725)" fund held for different consecutive days and conduct Kupiec test.The results show that: the number of consecutive days of holding funds will affect the accuracy of VaR calculation,when the number of holding days is too small,the model results are not reliable.When holding the same long days,the failure days of the improved mc-garch-var model and mcmc-garch-var model are reduced to a certain extent.Through the empirical analysis,we get the conclusion that MC-GARCH-VaR model and MCMC-GARCH-VaR model have improved the accuracy of the VaR of the Baijiu liquor index(LOF)(161725),and the MCMCGARCH-VaR model has higher accuracy,which provides investors with more accurate fund risk information so as to choose a more suitable investment plan.