A novel hybrid strategy for crude oil future hedging based on the combination of three minimum-CVaR models

Abstract

Effective and robust crude oil hedging strategies are becoming increasingly important for investors. However, due to the differences between data characteristics, the choice of model usually has a significant impact on hedging performance, and an incorrect model can make the hedging performance less efficient and robust. In this paper, a novel hybrid hedging model is proposed to reduce the model uncertainty. In detail, different hedging models are combined to construct a hybrid. Then, the maximum percentage reduction of the hybrid is selected as the optimization target to derive optimal hedge ratios and combination weights. The essence of the hybrid hedging model is to choose a better one among different strategies. In the empirical analysis, we construct the hybrid model by combining three single models: parametric, nonparametric, and semiparametric minimum-CVaR hedging models. The empirical results show that the novel hybrid model significantly outperforms other four competitive models in return, Sharpe ratio, maximum drawdown and downside volatility, while slightly outperforms other competitors in CVaR reduction. The robustness tests by changing the window width, confidence level, data frequency, empirical data, and objective function to be optimized validate the above conclusions.