Detecting market transitions and energy futures risk management using principal components

Abstract

An empirical approach to analysing the forward curve dynamics of energy futures is presented. For non-seasonal commodities—such as crude oil—the forward curve is well described by the first three principal components: the level, slope and curvature. A principal component indicator is described that detects transitions between the two fundamental market states remarkably well. For seasonal commodities—such as electricity and natural gas—it is shown how to extract the seasonal component from the forward curve. The principal component indicator can then be applied to the de-seasoned forward curve to detect significant price deviations that may support profitable trading strategies. A principal component approach to forward curve modelling is applied to computing portfolio value-at-risk. This approach is combined with a new two-step resampling procedure to improve value-at-risk estimates.