Optimization Of Function Coefficient Autoregressive Model (FAR) Based On LOWESS And Its Application In Finance

The study of financial series is very important for the analysis and prediction of economic situation,especially in the trading of stocks,futures,funds and so on.It can effectively analyze the characteristics of the series,explore the relationship between the series before and after,to achieve the purpose of early prediction and avoid risk.However,today's financial transactions are affected by too many uncertain factors,and the financial transaction series often has many characteristics,such as nonlinearity,memory,ARCH effect and so on.It is not suitable to study it by establishing a conventional parametric financial sequence model.Considering these complex characteristics of financial series,nonparametric model can deal with this kind of problem very well in practical analysis.In this paper,semi-parametric function coefficient autoregressive model(FAR)is introduced to analyze,and in view of the problems existing in the FAR model,a new improved method is proposed to improve the applicability and prediction ability of the model.In this paper,the local weighted scatter smoothing(LOWESS)optimization method is applied to the FAR model to solve the problem of FAR model.In the application of function-coefficient autoregressive models(FAR)to nonlinear time series data analysis,when there are few data or outliers at both ends of the sample values,the model has the problems of low accuracy and poor stability in estimating the regression coefficients.In this paper,the local weighted scattered smoothing(LOWESS)method is introduced to optimize the estimation of FAR model.Firstly,we calculate the Hurst exponent by using the method of rescaled range analysis(R/S),and distinguish the nonlinearity of the series.By the way,the ARFIMA model can be established according to the Hurst exponent as the comparative model of the target model.Then we establish the FAR model and estimate the regression coefficient.Finally,we apply the LOWESS smoothing method to optimize the FAR model,and the LOWESS-FAR model(LW-FAR)is established.The superiority of LW-FAR model is verified by simulation experiments,and the new model is applied to the financial series to analyze and predict.Through the empirical analysis of the London Financial Times 100 stock index(FTSE)and Cathay Pacific Fund return series,the fitting effect and prediction effect of LW-FAR model are compared with other models.According to the fitting effect and prediction results,the fitting effect and prediction effect of LW-FAR model are better than those of traditional parameter AR model and ARFIMA model.At the same time,it overcomes the estimation defect of FAR model when there are abnormal values.The prediction accuracy and stability of the model are improved.It broadens the applicability of the model.It shows that the new model achieves the optimization effect,can play a good advantage in dealing with complex financial series analysis,and can be better applied in meteorological research,biostatistics,air quality detection and other fields.