Stock Prediction Using Evolutionary Attention-based LSTM

Stock prediction is an important research issue that has attracted extensive attention from both academia and the industry. In previous studies, advanced deep learning models have shown better performance in predicting stock trends using time series data compared to traditional statistical time series models such as ARIMA and machine learning methods such as SVM. This study proposes a new deep learning method called Evolutionary Attention-based LSTM (EA-LSTM), which has not yet been used to predict stocks. Simulations were conducted on the famous stock index Dow Jones Index (DJI) to compare the prediction efficiency of EA-LSTM with other deep learning models including RNN, GRU, and LSTM as baselines, which are popular time series prediction models. The experimental results using RMSE and MAE as evaluation indicators indicate that the performance of EA-LSTM is superior to other methods.