A Feedforward Neural Network Model for Forecasting Crude Oil Prices Using Macroeconomic and Geopolitical Factors

Crude oil is one of the most important determinants of both global and national economies and a crucial factor for industrial activities. For that reason, numerous mathematical and machine learning approaches have been conducted to predict the future trend of oil market. Yet, predict oil price is one of the most challenging issues because high volatile nature of the oil market. The most common approaches to predicting oil prices using machine learning algorithms consider only the opening and closing prices. By reviewing the historical data of oil market, we observed that the crude oil market is heavily affected by certain geopolitical and macroeconomic factors. In consequence, we propose a model considering the macroeconomic and geopolitical factors to predict oil prices using feedforward neural network. After considering these factors, we achieved 86% R-squared value which proved that our argument is correct.