A regime-switching dynamic conditional angular correlation GARCH (RSAC) model is proposed for optimal portfolio diversification. RSAC specifies a regime-switching angular correlation dynamic for estimating the instantaneous state-dependent correlation matrix with a single multivariate realization. RSAC is applied to investigate the diversification benefit of precious metal and energy futures for stock sector indices traded on Shenzhen stock market in China. The empirical results reveal that fuel oil futures is an effective diversifier for stock sector holdings and angular correlation GARCH is superior to conventional varying-correlation GARCH under both state-dependent and state-independent specifications in terms of risk adjusted return and reward to semivariance ratio.