The LSTM Model with Error Rectification in Stock Price Prediction

Unpredictable stock prices cause investors to have unreasonable investments with heavy economic losses, causing them to lose confidence in the stock market. Therefore, this article proposed the long-short term memory (LSTM) network based on error correction to predict stock prices accurately. Here, we first developed the LSTM module to predict stocks based on historical stock price information. However, the fluctuation of stock still could not be accurately captured, leading to inferior predicted performance. Based on this limitation, we designed the error rectification model to rectify the error, where we added the average difference of the past three days to adjust the prediction. The experimental results show that by using the error rectification model, the Mean Absolute Error (MAE) in the LSTM model decreased from 68.28 to 0.79, and the Root Mean Square Error (RMSE) decreased from 70.37 to 1.71.