Forecasting crude oil prices volatility by reconstructing EEMD components using ARIMA and FFNN models

The energy sector which includes gas and oil is concerned to explore and develop refined oil and it's a multitrillion business. As crude oil is a very important source of energy, and it has a very valuable impact on a country's economic growth, national security, and social stability. Therefore, accurately predicting the crude oil price volatility is a very important topic of research and still, it is a challenge for researchers to accurately forecast crude oil prices. Therefore, this study is conducted to address the said problem significantly. This research presents a novel hybrid method for reconstructing EEMD IMFs that involves two steps. Visual analysis of Average Mutual Information (AMI) graphs were used to rebuild IMFs. EEMD IMFs were split into two components called stochastic and deterministic. In the proposed method, reconstruction of IMFs of EEMD was done at two stages to see if the stochastic components have more variation. Later, ARIMA and FFNN models were used to test the suggested method's performance. For this purpose, Brent crude oil prices data was used, and the hybrid model EEMD-S2D1D2-ARIMA/FFNN outperformed the other existing hybrid model with minimum MAE = 0.2323, RMSE = 0.3058 and MAPE = 0.5273. A simulation study was also conducted to check the robustness of the proposed method for N = 50, 500, 1,000, 2000, 5,000, and 7,500. The simulation results also confirm that the unpredictability present in the reconstructed IMFs of the hybrid models EEMD-ARIMA/FFNN and EEMD-SD-ARIMA/FFNN has been reduced by the proposed hybrid models.

Automated summary

This study proposes a novel hybrid method for accurately predicting the volatility of crude oil prices. By splitting EEMD IMFs into stochastic and deterministic components and testing with ARIMA and FFNN models, the proposed hybrid model performs well in terms of performance. Simulation results demonstrate the robustness of the method under different quantities.