A New Hybrid Deep Learning Model for Oil Prices Forecasting

The forecast of crude oil prices has always been important for investors and scholars, and has drawn more attention to applying deep learning techniques in recent years. Under this circumstance, firstly, this paper proposes a novel hybrid deep learning forecasting model named Mod-EMD-LSTM based on the empirical mode decomposition (EMD) and long short-term memory (LSTM) algorithms. Next, several empirical studies and statistical evaluations are carried out to evaluate its forecasting performance. The results show that the proposed model has excellent forecasting accuracy. Compared with the LSTM/EMD-LSTM model, the R2 and directional accuracy (DA) for Mod-EMD-LSTM increase by 19.67%/1632.61% and 48.649 pct/16.216 pct, respectively; Meanwhile, mean squared error (MSE) and mean absolute error (MAE) reduce by 39.26% /78.72% and 28.70%/55.84%, respectively. And all the above evaluation values can pass the corresponding statistical tests. Finally, through the robustness tests, we confirm that our model is robust for prediction and generalization.