Application Of A New Adaptive Lasso Method In Stock Market

Combined with the characteristics of the time series model,a new punishment method based on Adaptive Lasso is proposed,that is,the heavier the order is,the heavier the punishment is.In this paper,the parameter estimation and model selection of the simulated data are carried out by using the Lasso-Adaptive Lasso method and the new Adaptive Lasso method proposed in this paper.Considering the accuracy of parameter estimation and prediction accuracy,the simulation results show that the new Adaptive Lasso method performs better than Adaptive Lasso and Lasso methods.The empirical part selects the data of Shanghai Composite Index and Yunnan Baiyao,in which the Shanghai Composite Index takes the data of the daily closing price after the first order difference as the explanatory variable,and takes the data with the lag of 1 to 5 order as the explanatory variable.In Yunnan Baiyao,the data of 1to10 order lag was used as the explanatory variable,and the ARP)time series model was established.The Adaptive Lasso method and the new Adaptive Lasso method were used to select the variables and estimate the parameters,respectively.The results show that the new Adaptive Lasso method is slightly better than the traditional Adaptive Lasso method in modeling the stock price closing price,and it has a good performance in forecasting.This provides new ideas and methods for future research on stock price series and other time series.