The Research Of Internet Financial Risk Measurement Based On Least Squares Support Vector Machine

The development of Internet finance not only brings new impetus to China’s economy,but also provides many conveniences for people’s life.However,due to the lack of effective regulatory mechanism,different technical levels and varying risk tolerance of enterprises in the industry,in recent years,frequent risk events have affected the healthy development of Internet finance.Therefore,it is particularly urgent to measure and control the risks in the Internet financial market.In this context,how to accurately measure the risk of the Internet financial market has become the focus of this paper.This paper first takes the P2P business mode as an example to elaborate the risk status quo of China’s Internet finance,and emphasizes the change of government’s requirements for the development of Internet finance to clarify the significance of quantification and risk prevention and control of Internet finance.With the development of artificial intelligence technology and theory,it brings new opportunities for volatility prediction and risk measurement in financial market.This paper will measure the risk of Internet financial market based on least squares support vector machine(LSSVM).In the empirical part of this paper,based on the close price of Internet financial index,the initial volatility model is selected by comparing the volatility prediction effect of traditional GARCH models and LSSVM model.On this basis,we further consider the impact of other economic variables on the volatility of Internet financial market,and conduct secondary modeling of relevant economic variables and the volatility estimated by the initial model,so as to improve the accuracy of volatility prediction.After obtaining the best estimated volatility,we estimate the risk level of China’s Internet financial market using Value at Risk(VaR)and Expected Shortfall(ES)methods based on variance-covariance method,historical simulation method and monte carlo simulation method respectively.It is found that the return rate of Internet finance index has the characteristics of autocorrelation,fluctuation and agglomeration.Compared with the traditional GARCH model,LSSVM method can effectively improve the accuracy of volatility prediction.The introduction of relevant economic variables(such as exchange rate and panic index)further improves the prediction accuracy of volatility to a certain extent.Monte carlo simulation method has the best prediction ability in measuring the risk of China’s Internet financial market.