Design And Application Of The Residual Control Chart Of Skewed Autocorrelation Process

In the process of financial time series,the samples often appear autocorrelation and fluctuation clustering,which violates the independence assumption of shewhart control chart.In this paper,the sample data is fitted by arma-garch model to obtain the independent residual sequence,so as to construct the residual control chart.Usually in previous research,residual error sequence are assumed to be the standard normal distribution,but in the actual process of statistical control residual sequence usually do not obey the assumed that generally follows skewed distribution,As a case study of the stock of the domestic banking sector in the empirical analysis,based on its daily return sequence modeling process,the use of skewed distribution residual autocorrelation process control charts to monitor individual stocks abnormal fluctuations,he results show that this method is more effective in early warning,to help market regulators and investors in the market.The theoretical part firstly introduces the general control chart and the basic theory of autocorrelation control chart,and focuses on the theory of residual control chart.Then,the random simulation method is used to analyze the residual sequence,and the results prove that the control effect of traditional residual control is weakened or even invalid.For this situation,introducing skewness coefficient method to redesign of control chart,during the process of residual control chart with partial autocorrelation theory system,through to the controlled and uncontrolled state,comparing the average run length.It proved the control effect of skewed autocorrelation process residual control chart system is superior to the traditional residual control chart.Finally,empirical analysis in order to verify the applicability of the skewed autocorrelation process residual control chart,this article choose the domestic banking sector 16 stocks in the past three years of data for research,test show that the data appear autocorrelation and fluctuation clustering,first build ARMA-GARCH model for such problem.Then,the residual error is tested,and the residual sequence is independent and skewed,so it is suitable to introduce the system of skewed autocorrelation process control chart.The results show that the monitoring and early warning of daily teturn can be realized,and the better detection result is obtained.