SSE50 Prediction Based On Adaptive Lasso And Bayesian Neural Network Combination Model

The stock market reflects the current economic situation of a country and has a predictive effect on the future development of the economic situation of a country or region.The main research object of this paper is the SSE 50 comprehensive stock index.The SSE 50 Index is a comprehensive stock index compiled from 50 stocks with good liquidity,large scale and most representative in the Shanghai stock market.Therefore,the SSE 50 Index comprehensively reflects the overall situation of a group of leading enterprises with the most market influence in the Shanghai securities market.It can be said that the changes in the Shanghai Stock Exchange 50 directly reflect the operation of the entire stock market in Shanghai and the economic situation in Shanghai.Therefore,it is especially important to predict the SSE 50.As a comprehensive indicator reflecting the stock market,SSE 50 has a very important role.If it can be more accurately predicted,it will create immeasurable value.This paper mainly studies the changes in the closing price of SSE 50.Based on the closing price data of the SSE 50 Index and its constituent stocks from September 3,2018 to December 12,2018.First,the model is simplified by adaptive Lasso method to simplify the model.Therefore,the important explanatory variables of SSE 50 are screened out.Secondly,the combined Bayesian dynamic model and BP neural network are used to construct the combined model: the series model and the variable weight combination model(parallel model).Finally,the effect of the two models on the closing price of SSE 50 is compared.The variable weight combination model with better forecasting effect is selected to predict the closing price of SSE 50.In the variable selection,the important explanatory variables of the model are selected by the adaptive Lasso method.These explanatory variables mainly include stocks in industries such as banking,insurance,real estate,and building materials.These industries are the industrial pillars of a region and can represent a regional economic situation,and the actual situation is consistent.When constructing the tandem model,the Bayesian dynamic model is used to predict the selected explanatory variables.The BP neural network model is used to predict the closing price of SSE 50 by using the predicted values of the explanatory variables and historical data.It is found that the prediction effect of the tandem model performs better in the training set,but the performance on the test set is relatively poor.When constructing the parallel model,the Bayesiandynamic prediction model is first used to predict the closing price of the SSE 50.Secondly,based on the screening explanatory variables,the BP neural network model is used to predict the closing price of SSE 50.Finally,according to the theory of variable weight,the weights of different predicted values are given respectively,and the variable weight combination model of the two is constructed to predict the closing price of SSE 50.It is found that the prediction effect of the model is better in both the training set and the test set.And comparing the two models,the change weight combination model with better selection effect is used to predict the closing price of SSE 50.