A Comparative Study of Stock Forecasts by LSTM and RNN Neural Networks

Abstract

In the financial market, stocks have always played a very important role. The economic situation is closely related to the stock market. If people could make an effective prediction of the future trend of the stock market, it is of great significance to prevent the financial crisis and guide the investment direction. From this point of view, this paper uses artificial intelligence to obtain a feature representation through the analysis of massive stock price data, to predict the future stock price. Specifically, it uses recurrent neural network (RNN) and long short-term memory networks (LSTM) to predict the stock trend. Under the same experimental conditions, the experimental results predicted by the two methods are compared and analyzed. The experimental results show that RNN has an effective prediction for the trend of stock price, but LSTM has a better prediction accuracy, especially in the short-term prediction.