Modeling the dynamics of energy commodity prices using neural networks

The deregulation of energy commodity markets has caused changes in the dynamics of commodity prices. One main aim of deregulation is to allow markets to respond to supply and demand conditions causing more competitive markets environments. Energy commodities have recently become an asset class used as an investment tool by various kind of market participants. Spot markets for oil, gas and electricity are at hand and are largely used also by financial institutions besides the traditional retailers or producers. In addition, a large set of energy derivatives are currently traded in most European and US Exchanges providing an useful tool to hedge risk. We use time series of daily prices of oil, gas and electricity for the period 2001-2010. Time series have been arranged in 3-years slots, since all the predictors have been trained on a 2-years time window and tested on the successive 1-year period.