Research On Stock Index Price Forecasting Based On LSTM Neural Network

Accurate market prediction to obtain arbitrage opportunities is the goal of many investors.With the rapid development of financial technology,the prediction results of the stock market are getting better.From the perspective of the effect of stock market prediction,traditional statistical methods,econometrics methods and time series analysis have lag problems,while neural network machine learning can deal with non-stationary complex and nonlinear problems,and achieve better results in the prediction of financial time series.This paper builds an LSTM neural network model to predict the closing price of SSE50 Index the next day.Among them,technical indicators and prices of related assets are added to the selection of features.This paper uses LASSO regression to screen the features and optimize the model to obtain the best prediction effect.Finally,some evaluation indicators and methods of formulating trading strategies are used to evaluate the effect and actual feasibility of the forecast.Firstly,this paper theoretically introduces the artificial neural network,cyclic neural network,and improved neural network,including LSTM network and GRU,through the perspective of time series analysis of their advantages and disadvantages,so as to choose LSTM neural network as the basis of the model.Secondly,data is an important part of the research.This article analyzes and processes the SSE50 index and characteristic data of the research object respectively.For the SSE50 index,this article uses the return rate and the time series data analyzed by wavelet transform as the research object,and the price of next day is set as the label.And for the feature data,use LASSO regression to select specific features and process the features.Then build and train the LSTM model to obtain the prediction results,evaluate the prediction effect through the evaluation indicators,find that the effect of the model is effective,and design a trading strategy based on the results,and compare the return rate of the strategy with the return of the SSE50 index rates are compared to reflect the pros and cons of the prediction effect.Finally,the conclusion is drawn that the LSTM model is effective and feasible in the research of stock price prediction,which has an important reference value for the application of artificial intelligence in the financial field.