Modeling, Risk Assessment and Portfolio Optimization of Energy Futures∗

This paper examines the portfolio optimization of energy futures by using the STARR ratio that can evaluate the risk and return relationship for skewed distributed returns. We model the price returns for energy futures by using the ARMA(1,1)-GARCH(1,1)-PCA model with stable distributed innovations that reflects the charac-teristics of energy: mean reversion, heteroskedasticity, seasonality, and spikes. Then, we propose the method for selecting the portfolio of energy futures by maximizing the STARR ratio, what we call “Winner portfolio”. The empirical studies by using energy futures of WTI crude oil, heating oil, and natural gas traded on the NYMEX compare the price return models with stable distributed innovations to those with normal ones. We show that the models with stable ones are more appropriate for energy futures than those with normal ones. In addition, we discuss what characteristics of energy futures cause the stable distributed innovations in the returns. Then, we generate the price returns of energy futures using the ARMA(1,1)-GARCH(1,1)-PCA model with stable ones and choose the portfolio of energy futures employing the generated pric