Construction And Improvement Of Monte Carlo Simulation Sampling Model And Measurement Of Internet Financial Risk

In recent years,the rapid development of Internet finance has made research on its risk measurement increasingly important.This article selects the representative CSI Internet Finance Index as the research object and explores the Internet financial characteristics of its data.Based on this,the basic sampling model for measuring Internet financial risk is determined and improve the model in terms of sampling model selection,accuracy,Monte Carlo error and convergence.Finally,the Bayes-GARCHQMC model is established to measure the Internet financial risk.The research in this article is divided into three parts:The first part is the theoretical analysis of Internet finance and the research on the characteristics of Internet finance of the yield data.First,theoretically analyze the operating mechanism of Internet financial products and the factors affecting the rate of return.Second,from the perspective of data,based on the kernel density and quantile method to mine the tail features of the data.Compare the statistical characteristics of the CSI Internet Financial Index with the traditional CSI Financial Index and calculate the quantile values of 90%,95% and 99% for the two respectively.It is concluded that the quantile values of the CSI Internet Finance Index are greater than the corresponding CSI Financial Index.It shows that the data of Internet finance’s rate of return presents the characteristics of thicker and longer tail data.The second part is to determine the basic sampling model of Monte Carlo(MC)simulation method.Based on the Internet financial characteristics of the yield data.Comparing the MC simulation method based on geometric Brownian motion with the GARCH wave model,from the perspective of quantile,risk measurement and model improvement,it is concluded that the MC simulation method based on geometric Brownian motion is more suitable for the basic sampling model.The reasons are as follows: From the perspective of quantiles,Because the GARCH model cannot directly reflect the quantile information,it is not convenient to compare with the Internet financial characteristics of the data.The MC simulation method based on geometric Brownian motion,by simulating the path that the Internet financial index may occur many times in the future,calculate the quantile of the mean value of the simulated path at each moment and compare it with the quantile of the actual data to test the rationality and accuracy of the model.From the perspective of risk measurement,return test the model.It is concluded that the failure rate of MC simulation based on geometric Brownian motion is lower than that of GARCH model.From the perspective of model improvement.Once the GARCH model is established,it is not easy to repeatedly improve,and the MC simulation method based on geometric Brownian motion has a lot of room for improvement.Improvement from multiple angles makes the improved model more reasonable.Therefore,from the three perspectives of quantile,risk measurement and model improvement,determine the basic sampling model for measuring Internet financial risk.The third part is the multi-angle improvement of the basic sampling model by MC simulation method and the measurement of Internet financial risk.First,select a more general ?Ito process to improve the sampling model,introducing the GARCH model to describe the volatility in the ?Ito process.Establish GARCH-MC model.Through the return test,the improvement improves the accuracy of the model.Secondly,the accuracy is improved based on the Bayes method,determine the prior distribution of the sample likelihood function and parameters,and then derive the joint posterior distribution of parameters and the full-condition posterior distribution of each parameter,,use the Gibbs sampling to estimate the parameters of the GARCH model,and establish the Bayes-GARCH-MC model.Through the return test,the overall accuracy of the model improved again is further improved.Again,the Quasi-Monte Carlo(QMC)simulation method is selected to improve Monte Carlo error and convergence,generate quasi-random numbers with uniformity instead of pseudorandom numbers,obtain quasi-random numbers following standard normal distribution by inverse function method,and establish Bayes-GARCH-QMC model.The comparison shows that the Monte Carlo error of the model is lower and the convergence speed is faster.Finally,verify the reliability of the Bayes-GARCH-QMC model,and measure Internet financial risk based on the model,forecast the Va R and CVa R for the next 20 days.