Modeling Energy Markets Using Neural Networks and Spectral Analysis

Energy commodity markets have become crucial for the economic growth policies and investment strategies in the last decade. Crude oil prices have experienced unprecedented volatility due not only to structural features but also to geopolitical issues, to the recent discovery of shale gas and to the increased use of gas turbines. Moreover, the deregulation of electricity markets requires an efficient tool to describe the price features whilst standard econometric procedures cannot provide an accurate description of the actual price dynamics. In the present work we illustrate the application of filter banks and neural networks to predict prices of specific energy commodities, which play a crucial role in the international economic and financial context. The proposed approach is assessed by the numerical results obtained for US crude oil, natural gas and electricity daily prices.