Research On Risk Measurement Of Shanghai Crude Oil Futures Market Based On GARCH Family Models And MCS Test

On March 26,2018,as the first futures product opened to the outside world,my country's crude oil futures were officially listed on the Shanghai Futures Exchange.Its launch will have important strategic significance for the development of my country's energy industry.Shanghai crude oil futures quickly opened the crude oil futures trading market with the help of differentiated competition in trading varieties and convenient conditions such as RMB settlement,providing favorable financial tools for Chinese crude oil industry chain related companies to actively participate in the crude oil market,hedge and avoid risks.The price of crude oil is extremely susceptible to huge fluctuations due to changes in the international situation,changes in supply and demand,and changes in transportation conditions,which in turn will cause huge risks to my country's crude oil futures market.In view of this background,the risk measurement of my country's Shanghai crude oil futures market and an effective risk measurement method are of great significance for guiding investors to invest and for the regulatory authorities to formulate risk prevention and control strategies.Based on the normal distribution,student t distribution,and generalized error distribution,this paper introduces the skewed normal distribution,the skewed student t distribution,the skewed generalized error distribution,the generalized hyperbolic distribution and the Johnson SU distribution and constructs the GARCH model,EGARCH Model,GJRGARCH model,APARCH model and CGARCH model.The current most mainstream VaR method is used to measure the risks of the Shanghai crude oil futures market.Using the rolling time window prediction mechanism,the out-of-sample dynamic VaR is predicted one step forward based on the results of in-sample parameter estimation.Using the more robust MCS test method proposed by Hansen and select the asymmetric linear VaR loss function considered by Gonz'alez-Rivera.Investigate the influence of the update interval of in-sample parameter estimates,the VaR quantile level,and the out-of-sample VaR prediction span length on the VaR prediction ability and characterization accuracy of the risk measurement model.The VaR prediction performance of different distributed GARCH family models is evaluated and the excellent model set(SSM)is obtained.And the model suitable for the Shanghai crude oil futures market risk measurement is selected.Then estimate the parameters of the relative optimal model and make relevant analysis.Through the above empirical research,this article finds:(1)The Shanghai crude oil futures yield sequence has the characteristics of sharp peaks,thick tails,left-biased,and significant volatility aggregation;The impact of bad market news is greater than the impact of good news,Showing obvious characteristics of leverage effect;(2)No matter which parameter update interval method is used,the number and types of models in the surviving excellent model set are almost the same,and the size of the loss value function is not much different,that is,the parameter update interval method has almost no impact on the VaR prediction accuracy of the model;(3)Under the high-risk quantile level,the EGARCH-snorm model is more suitable for measuring the risk of the Shanghai crude oil futures market;under the medium-risk quantile level,EGARCH-norm performs better Excellent;and the APARCH-norm model has the best risk prediction effect on the low-risk Shanghai crude oil futures market;(4)According to the discussion under different prediction spans,it is found that there is a prediction span critical value H=300.When the prediction span is less than or equal to 300,only the loss value is used to judge the prediction ability of the risk measurement model,the EGARCH-snorm model is the relatively optimal model for measuring the risk of the Shanghai crude oil futures market;when the critical prediction span H=300,Different from the previous conclusions under the setting of the forecast span,the EGARCH-snorm model no longer has relatively optimal risk prediction capabilities.At this time,the EGARCH-ghyp model and the EGARCH-sged model can more fully and accurately reflect the Shanghai crude oil Risks in the futures market.