Title: Crude oil prices and kernel-based models

Authors: Massimo Panella; Rita L. D'Ecclesia; David G. Stack; Francesco Barcellona

Addresses: Department of Information Engineering, Electronics and Telecommunications, University of Rome 'La Sapienza', Via Eudossiana 18, 00184 Rome, Italy ' Department of Quantitative Finance, University of Rome 'La Sapienza', Via Del Castro Laurenziano 9, 00161 Rome, Italy ' ESCP Europe Business School, 527 Finchley Road, London NW3 7BG, UK ' InfoSapienza - Informative Systems for Network Communications, University of Rome 'La Sapienza', Piazzale Aldo Moro 5, 00185 Rome, Italy

Abstract: In this paper, we use a kernel-based approach to crude oil price prediction which would allow us to set up efficient risk management strategies. Practitioners find strong evidence that investor flows follow prices so commodity investments are likely to continue to grow, and we believe this will drive an increasing importance for methodologies like neural networks for risk quantification, measurement and management. Crude oil prices for both Brent and WTI in the last 12 year period are used to provide an accurate analysis for both time series. Four different neural network models are used. The superior model is the neuro-fuzzy network based on Sugeno first-order type rules, also known as the adaptive neuro-fuzzy inference system method, which provides both an accurate prediction of prices and their probability distribution.

Keywords: crude oil prices; oil price dynamics; time series prediction; neural networks; modelling; kernel-based models; risk management; neuro-fuzzy inference system; fuzzy logic.

DOI: 10.1504/IJFERM.2014.058761

International Journal of Financial Engineering and Risk Management, 2014 Vol.1 No.3, pp.214 - 238

Received: 21 Nov 2012
Accepted: 29 Apr 2013

Published online: 21 Oct 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article