Risk management in modern financial markets requires models that are both adaptive and robust to regime changes and extreme events. In this paper, we propose a hybrid framework that integrates a Gaussian Hidden Markov Model (HMM) for regime detection, a regime-specific GARCH(1,1) model with Student’s t innovations for conditional volatility estimation, and Extreme Value Theory (EVT) for tail risk calibration. The resulting dynamic Value-at-Risk(VaR) measure adapts to market conditions. Fallback mechanisms are implemented to improve the robustness of the model in the face of convergence failures and data limitations. The model’s performance is evaluated through extensive backtesting and stress-testing, demonstrating that our model captures extreme losses with improved accuracy and maintains statistical consistency across regimes.