Forecasting the differences between various commercial oil prices in the Persian Gulf region by neural network

In this paper we have investigated the differences between the prices of different commercial oils of the Persian Gulf region. The prices of 7 different crude oils from Iran, Kuwait, Saudi Arabia, Oman, Abu Dhabi and Dubai were compared with the benchmark light oil of Saudi Arabia over the period January 2000–April 2010. A neural network is introduced to forecast the price of any commercial oil in these crude oils, provided that the price of the benchmark light oil of Saudi Arabia is already known or is predicted by another forecasting method. The designed neural network is able to predict the differences in the oil prices with an average error of 8.82% for testing and 7.24% for training data. It is claimed that the present method can promote the forecasting power of existing models to predict the price of any commercial oil instead of an average or benchmark value.

► This paper introduces a method to forecast the usually ignored commercial oil price differences. ► The method introduced in this paper may improve the ability of oil price forecasting methods. ► The method introduced in this paper may lead to a more realistic approach for prediction of commercial oil prices. ► This paper investigates the differences between the oil prices of Persian Gulf region.