Automated Box-Jenkins methodology to forecast the prices of crude oil and its derivatives

Abstract

Developing forecasting models that incorporate that external parameters in addition to past data for crude oil and derivatives are a challenging task since it is highly dependent on economic, geographical, and political issues. However, forecasting the prices is very important for strategic planning and oil refineries’ operational decisions. This paper presents an automated tool to predict crude oil prices and their main products by applying Box-Jenkins methodology for the next two months at the beginning of each month in a rolling horizon manner. The resulting forecast is shared with related departments to develop their production plans accordingly. We show that improved accuracy with this forecasting approach is beneficial in any planning and decision-making process and increases profit.