MODELLING CONDITIONAL VOLATILITY AND ASYMMETRY IN GOLD FUTURES RETURNS

This study investigates the time-varying volatility dynamics of Gold Futures (GCZ5) daily returns over a ten-year period (October 28, 2015, to October 28, 2025). Given gold's critical role as a safe-haven asset, accurate volatility modeling is essential for contemporary risk management and derivative pricing. We employ the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) family of models, including symmetric and asymmetric specifications (GJR-GARCH and EGARCH), using heavy-tailed distributions (Student's t and GED) to capture the observed leptokurtosis. Preliminary diagnostics confirmed the presence of volatility clustering and non-normality in the returns series. Model selection, based on the Akaike Information Criterion (AIC) and Schwarz Criterion (SIC), identified the EGARCH(1,1) model with a Student's t error distribution as the superior specification. The key findings reveal extremely high volatility persistence (beta approx 0.984), indicating that volatility shocks have a long-lasting impact on the GCZ5 risk profile. Furthermore, we detect a statistically significant contravariant asymmetry, where positive returns (gains) increase future volatility more than negative returns (losses) of the same magnitude. These quantitative insights are vital for institutional investors seeking to optimize dynamic hedging ratios, accurately price options, and set appropriate risk limits in the highly volatile gold market.