Volatility Clustering and Short-Term Forecasting of the Hang Seng Tech Index Based on the ARIMA-GARCH Model

Abstract

This study investigates the asymmetric risk characteristics of technology stocks by examining the Hang Seng Technology Index (HSTECH). Daily data from 2021 to 2026 are analyzed using a hybrid ARIMA-EGARCH model with a Student-t distribution, which combines linear mean modeling with non-linear volatility modeling to examine volatility clustering, leverage effects, and short-term forecasting performance. The results show that the HSTECH return series has clear heavy-tailed and leptokurtic characteristics, together with strong volatility clustering, indicating that the normality assumption is not suitable for this market. The analysis also finds a significant negative leverage effect, which shows that negative market shocks increase future volatility more than positive shocks of a similar size. Dynamic out-of-sample rolling forecasts further show that the hybrid ARIMA-EGARCH model produces lower prediction errors than the standard linear ARIMA model, demonstrating its ability to capture the volatility dynamics of the HSTECH more accurately. These findings provide useful evidence for dynamic hedging, Value-at-Risk (VaR) estimation, and downside risk management in technology-focused equity markets.