FORECASTING THE PRICE OF CRUDE OIL USING ARTIFICIAL NEURAL NETWORKS

Hassan Khazem

Nova SouthEastern University

Abdelkader Mazouz

Hamdan Bin Mohamed University

ABSTRACT

Crude oil is the commodity de jour and its pricing is of paramount importance to the

layperson as well as to any responsible government. However, one of the main challenges facing

econometric pricing models is the forecasting accuracy. Historically, linear and non-linear time

series models were used. Although, a great success was achieved in that regard, yet there were no

definite and universal conclusions drawn. The crude oil forecasting field is still wide open for

improvement, especially when applying different forecasting models and alternative techniques.

Toward this end, the proposed research implemented Artificial Neural Network models (ANN).

The models will forecast the daily crude oil futures prices from 1996 to 2006, listed in New York

Mercantile Exchange Group (NYMEXG). Due to the nonlinearity presented by the test results of

the crude oil pricing, it is expected that the ANN models will improve forecasting accuracy. An

evaluation of the outcomes of the forecasts among different models was done to authenticate that

this is undeniably the situation.