Brent Crude Oil Price Forecasting by Combining Grey Theory and Econometrics Techniques

Abstract

The characteristics of crude oil and the factors affecting the price of this energy carrier have made its price forecast always considered by researchers, oil market participants, governments, and policymakers. Because the price of crude oil is affected by many factors, ongoing studies should be done to make more accurate and reliable estimates over time. In this paper, a combination of GM (1,1) and ARIMA models and a hybrid model (GM-ARIMA) for crude oil price forecasting is proposed. The Brent crude oil price data for seasonal (2015Q1-2021Q2), monthly(2020m3-2020m12), and weekly(w12-2020: w16-2021) periods were used to examine this method. The results show that based on the evaluation criteria of mean absolute error percentage (MAPE) and square mean square error (RMSE), the evaluation criteria of MAPE and RMSE in the combined GM-ARIMA model are always lower than the GM and ARIMA models alone. Therefore, the GM-ARIMA hybrid model will be able to predict more accurately than the GM and ARIMA models. Therefore, for more accurate prediction, the GM-ARIMA hybrid model can be used instead of single models.

Keywords

• crude oil price
• crude oil price forecast
• gm grey model
• arima model
• gm-arima hybrid model