Forecasting Crude Oil Prices: A Comparison Between Artificial Neural Networks and Vector Autoregressive Models

Given the importance of crude oil prices for businesses, governments and policy makers, this paper investigates predictability of oil prices using artificial neural networks taking into account the exhaustible nature of crude oil and impact of monetary policy along with other major drivers of crude oil prices. A multilayer perceptron neural network is developed and trained with historical data from 1980 to 2014 and using mean square error for testing data, optimal number of hidden layer neurons is determined and the designed MLP neural network is used for estimation of the forecasting model. Meanwhile, an economic model for crude oil prices is developed and estimated using a vector autoregressive model. Results from the proposed ANN are then compared to those of the vector autoregressive model and based on the corresponding R-squared for each model, it is concluded that the MLP neural network can more accurately predict crude oil prices than a VAR model. It is shown, via empirical analysis, that with a combination of appropriate neural network design, feature engineering, and incorporation of crude oil market realities in the model, an accurate prediction of crude oil prices can be attained.