To investigate the advantages of the composite hedge over the single hedge, this paper uses the ordinary least squares (OLS), dynamic conditional correlation (DCC), asymmetric dynamic conditional correlation (ADCC), copula-DCC, copula-ADCC, generalized orthogonal generalized autoregressive conditional heteroskedastic (GO-GARCH) and generalized orthogonal Glosten-Jagannathan-Runkle (GO-GJR) models to estimate the hedge ratio in the minimum variance and expected utility maximization objective under single and composite cross hedge settings. Under all the hedging performance criteria, the composite hedge is better than the single hedge in most circumstances. In addition, the copula-ADCC, GO-GARCH and GO-GJR models perform better than other relatively simple models. In this cross composite hedging context, the naïve strategy can be beaten. The findings are robust to model specifications and contexts.