A study of financial time series volatility forecasting method based on GARCH modeling

This paper focuses on the problem of financial time series volatility modeling, constructs and compares various heteroskedasticity models such as GARCH(1,1), EGARCH, GJR-GARCH and APARCH, and conducts empirical research by combining with CSI 300 index return data. Rolling forecasting and multi-step evaluation methods are used to systematically compare the model performance using RMSE, MAPE, R² and other indicators. The results show that EGARCH and APARCH models perform better in terms of fitting accuracy and forecasting stability, and are more effective in capturing volatility aggregation and asymmetric effects.