Construction Of Bayesian ASV-T-POT Model And Risk Measurement Of Internet Financial Product Returns

In recent years,the rapid development of Internet finance makes the research on its volatility and risk measurement more and more important.This paper takes the return rate of Internet financial product Jingdong Small Treasury as the research object to explore its Internet financial characteristics.After the selection of volatility model categories,it constructs Bayesian SV model to study the volatility of the return rate of Jingdong Small Treasury.After multi-angle comparative analysis,the Bayesian ASV-T model is selected and the POT model is further constructed on this basis to measure the volatility risk of the return rate of Jingdong Small Treasury.The research content of this paper is mainly divided into the following three parts:The first part: the Internet financial characteristics of the return rate of Jingdong Small Treasury and the type selection of the volatility model.Firstly,taking the return rate of the representative Internet financial product Jingdong Small Treasury as the research object,the Internet financial characteristics of Jingdong Small Treasury are described from three perspectives of basic statistical analysis,basic analysis of volatility and graphical analysis of return distribution.It is found that the return rate has strong volatility persistence and peak thick tail,and its peak is much higher than the normal distribution and t distribution.Secondly,GARCH model and SV model are compared and analyzed from four aspects of model structure,data characteristics,prediction ability and parameter estimation.Finally,based on the above analysis,the model category is set as SV model.The second part: the construction of the Bayesian SV model of Jingdong Small Treasury and the research on the volatility of its return.First of all,based on data characteristics and considering that it is impossible to intuitively see from the sequence itself whether the rate of return has volatility characteristics such as "leverage effect",so construct separately Bayesian SV-T model with thick tail characteristics and Bayesian ASV-T model with leverage characteristics.After determining the conditional likelihood function and the prior distribution,and deriving the full conditional posterior distribution of each parameter of the model,the parameters are estimated based on the MCMC algorithm,and the constructed model is obtained.Secondly,according to the volatility analysis of the above models,it is found that the volatility of the return rate has a strong persistence.The two models are compared and analyzed from the four perspectives of parameter estimation,volatility analysis,DIC criteria and out-of-sample forecasts.It is concluded that Bayesian ASV-T model is better than Bayesian SV-T model in both fitting effect and prediction effect.Finally,according to the Bayesian ASV-T model,this paper studies the volatility of Jingdong Small Treasury and concludes that the highest volatility level is0.0588 and the lowest is 0.0007,which can reflect more volatility outliers.The parameter estimation results show that the return rate of Jingdong Small Treasury has the "inverse leverage effect",which reflects the immaturity of the Internet financial market and investors have irrational investment behavior.At the same time,it also lays the foundation for the third part of constructing the POT model to measure the rate of return risk of Jingdong Small Treasury.The third part: constructing the Bayesian ASV-T-POT model to measure the risk of the return rate of Jingdong Small Treasury.Based on the results of the Bayesian ASV-T model constructed in the second part,the residual sequence standardized by the return rate is obtained,and the POT model is further constructed.After the threshold selection and parameter estimation,the Bayesian ASV-T-POT model is obtained.According to the risk measurement of this model,the Va R values of the return residuals at 95%,97.5% and 99% confidence levels are 5.6335,10.3493 and20.3651 respectively.On this basis,the Va R value of the return rate of Jingdong Small Treasury is further calculated,and the Kupiec return test and prediction of Va R are carried out.The results confirm that the Bayesian ASV-T-POT model can effectively measure the risk of the return rate of Jingdong Small Treasury.