Stock Price Prediction based on Recurrent Neural Network with Long Short-Term Memory Units

Stock price prediction has been playing a very comprehensive impact on the financial industry. However, it has been considered as one of the most challenging tasks due to the financial time series data has some unpredictable and volatile characteristics. Conventional statistical models can only give a reasonable prediction for the next following one step. In this paper, we propose two novel prediction methods embedded with a deep learning framework using long short-term memory units, providing predictions for both short and long-term horizons. Comparison tests has been carried out between the two proposed methods, and the experiment results have shown that our methods have a significant performance in capturing both the trend and the exact value of the stock data.