<p>This paper investigates volatility regime switching for asset returns during extreme price movements. We propose a copula-based latent factor-driven asymmetric endogenous regime-switching model, where regime changes are endogenously influenced by the dependence between shocks to the asset returns and future shocks to the latent factor driving regime transitions. Unlike prior studies that assume symmetric relationships, our approach employs a flexible copula framework to capture asymmetric dependence between these shocks, introducing asymmetric endogeneity in regime switching. Our results demonstrate that, in the presence of extreme return shocks, our copula-based asymmetric model predicts state transition probabilities - and thus future volatilities - more accurately than the symmetric model. Specifically, we find that the symmetric model tends to underestimate the probability of transitions from a low-volatility (low-risk) to a high-volatility (high-risk) regimes following extreme returns. In an empirical application, our model delivers superior in-sample fit and out-of-sample forecasting performance compared to the symmetric model.</p>