Core Insights - The report innovatively combines the GARCH-EVT-VaR model with a defensive timing mechanism to construct a dynamic risk management framework, effectively capturing market tail features and volatility clustering while controlling downside risk and retaining upside elasticity, significantly enhancing portfolio performance [2][10][12] - The framework provides an important risk management tool for asset allocation in the current high-volatility environment, particularly during extreme market events such as the COVID-19 pandemic and geopolitical conflicts [10][12] Model Characteristics - The GARCH model parameters for the four indices indicate significant volatility clustering, with the ChiNext index and CSI 300 showing greater resilience to volatility, while the CSI 500 exhibits a notable leverage effect [3][4][49] - Different indices have varying degrees of fit for the GARCH family models, with ChiNext and CSI 300 best suited for GARCH(1,1), CSI 500 for GARCH(1,1,1), and CSI 1000 for EGARCH(1,1,0) [3][4][49] Backtesting Results - The timing models for each index show positive excess returns, with annualized excess returns of 4.42% for ChiNext and 4.81% for CSI 1000, alongside reduced annualized volatility compared to benchmarks [4][71] - Each timing model has demonstrated a decrease in maximum drawdown, with ChiNext's maximum drawdown reduced from 49.4% to 36.8% and CSI 1000 from 42.0% to 30.0% [4][71] Strategy Advantages - The GARCH-EVT-VaR model provides more precise risk measurement, effectively capturing extreme tail risks, and demonstrates significant defensive effects, especially during extreme market events [4][10] - The strategy shows clear excess returns and higher Sharpe ratios compared to benchmarks, indicating improved risk-adjusted performance [4][71]