Forecasting the Price Development of Crude Oil with Artificial Neural Networks

Abstract

The objective of the presented project was to develop and implement a forecasting instrument to predict the oil price in short-, mid- and long-term. Because there are a lot of different and complex factors influencing the oil price, the neural net method was chosen. Many data that could be relevant for the prediction was integrated in the net and several architecture models were tested. The data base consisted of about 2000 data records reflecting the period between 1999 until 2006. As result of the project it can be summarized that the implemented neural nets could not achieve sufficient results in the short-term forecasting but achieved very good results in the mid- and long-term predictions. Therefore it should be a valuable instrument for supporting management decisions in this field. © 2009 Springer Berlin Heidelberg.