Research On Stock Index Forecast Based On Investor Sentiment And LSTM Neural Network

With the vigorous development of China’s economy and the gradual improvement of people’s income level,people begin to pay more and more attention to how to manage their savings funds.Stock investment has been favored by more and more investors because of its characteristics of quick return and strong liquidity.The trend of stock price index not only directly affects the stability of stock market,but also relates to the sound development of China’s economy and finance.Investors need to judge the short-term economic situation according to the rising and falling trend of the stock price index to judge the investment risk in advance.At the same time,the Chinese government also needs to monitor and guide the benign development of the stock market.However,due to the uncertainty and high noise of stock index time series data,it is very difficult to forecast effectively.In this paper,the most popular short and long-term memory neural network(LSTM)in the field of stock prediction is selected for model training and prediction of stock technical indicators.On the one hand,by selecting the most representative three stock indexes in China for predictive analysis,the LSTM neural network has excellent generalization ability.On the other hand,it shows that LSTM neural network has better forecasting ability of long and short term stock index by comparing with the prediction results of logistic regression model.In addition,in order to improve the accuracy of stock index prediction,the model input selects five technical indicators including opening price,closing price,highest price,lowest price and trading volume,and also adds investor sentiment indicators.Based on the data the crawler technology wealth from the east to climb on BBS took the investors in the related indexes BBS review data,the use of special financial dictionary for emotion classification to study the emotional tendencies of investor comments,and to quantify the investor sentiment to build investor sentiment indicators,to improve the effectiveness of the stock price forecast.Through empirical analysis this paper get the following conclusions,first LSTM neural network model in the stock index prediction has a good generalization ability,then through comparing with the logistic regression model,found that LSTM neural network model has better prediction accuracy and better stability,and finally after join investor sentiment indicators,It is found that the prediction accuracy of neural network model is higher than that of technical index only.