Research On SP 500 Stock Return Forecast Based On LSTM Neurual Network

In the development of the real economy,the performance of the stock market can well reflect its development status.Therefore,using certain methods to predict the stock price has a very important value to a certain extent.There are many factors that affect the stock price trend.In order to predict the trend of stock prices more accurately,reasonable and effective forecasting methods are very important.The research and forecast object of this article is the daily return of the S&P 500 index and its constituent stocks.The sample data comes from the Google website,which is a time series of about 30-40 characteristics from April 1,2019 to September 1,2019,a total of 153 days,including transaction information,financial information,external macro indicators,etc.data.In terms of prediction methods,this paper uses GARCH model,BP(Back Propagation)neural network,and Long Short Term Memory(Long Short Term Memory)to make predictions.In the choice of prediction deviation function,this paper adopts MSE(Mean Squared Error)mean square error to define the prediction ability of the model.Finally,by comparing the prediction capabilities of these three models,this article finds the optimal prediction model and draws the following main conclusions: The daily return of the S&P 500 index and its constituent stocks has a certain degree of predictability.The forecasting model uses the historical information of the stock to make a reasonable prediction of future returns.Compared with the GARCH model and the BP neural network model,the LSTM neural network has stronger prediction performance for the S&P 500 index and its constituent stocks.However,the prediction effect of the LSTM neural network,which is a mixture of multiple external indicators(mainly including financial indicators and macro indicators)as input features,is superior to the ordinary LSTM neural network.It can be seen that some external indicators can predict the model.However,after selecting the opening price indicator of the day as an additional input feature,the model's forecast deviation has risen.It can be seen that the forecast model needs to be carefully considered.When predicting the rise and fall of stock prices,data selection,processing,and feature extraction are very important links.In the stock investment decision,this article has certain practical significance to the judgment of stock price trend and the prediction of stock return rate.