Research On VaR Model Based On Adaptive Data Decomposition Method

With the acceleration of economic globalization,the financial market becomes more and more open and the risks in the financial market continue to increase,which makes the risk measurement become more and more important.How to construct a scientific and effective risk measurement method has been a hot and difficult issue concerned by many scholars.In this paper,a VaR estimation model based on adaptive decomposition method is proposed to build a value-at-risk model.In this paper,the daily closing price of standard & poor's 500 index is used as the data source,and empirical mode decomposition,overall empirical mode decomposition and local mean decomposition are used for data processing,and GARCH model is combined.Meanwhile,it is compared with the traditional GARCH model.Through comparative analysis,it is found that the VaR estimation effect of the arimagarch model based on LMD and subject to generalized error distribution of residual sequence is optimal.The specific work of this paper is as follows:Firstly,the original data was processed with logarithmic rate of return.Then,EMD,EEMD and LMD were respectively used for data sub-processing,and multi-scale analysis including descriptive statistics,normality and periodicity were carried out for each component after decomposition,laying a foundation for the subsequent establishment of a reasonable VaR model.Secondly,the ARIMA-GARCH model that makes each parameter significant is established for each component on the premise that the residual sequence obeys the normal distribution,t-distribution and generalized error distribution respectively,so as to obtain the conditional variance sequence of each component.Next,quantiles at different confidence levels are obtained to obtain the conditional variance of integration,and the VaR value of the S&P500 index at the significance levels of 1%,5% and 10% is calculated.Finally,the failure rate of each VaR sequence was backtested and compared with the results of the classical method.The empirical results show that for the S&P 500 day's closing price of logarithm yield data,when the residual error of each model are assumed to be under the condition of generalized error distribution,based on decomposition and LMD in residual error sequence to obey generalized error distribution under the condition of ARIMA-GARCH model VaR estimation model accuracy is higher than other two kinds of processing methods,but also higher than that of classical time series model.In summary,when it comes to risk management,the measurement of risk can be saidto be a very important link,and the accurate resolution of financial risk depends to a great extent on the accurate measurement of risk.Through research and comparison,it is found that the VaR model based on the adaptive decomposition method proposed in this paper canfit the data very well and has an accurate VaR estimation effect.Therefore,the model given in this paper can also be generalized and applied to the risk measurement of carbon prices,currency funds,Internet virtual currency and so on.