Volatility Patterns of Islamic Equity Funds: Using Hybrid Machine Learning and GARCH Models

 This paper examined the volatility of Islamic equity funds using daily price data from 2009 to 2023. Volatility models for S-GARCH, GJR-GARCH, E-GARCH, SVM-GARCH hybrid, and neural network are implemented to measure the accuracy of volatility prediction. The results of this study show that past volatility, unconditional variance, and lagged conditional variance are revealed as strong predictors of Islamic funds volatility. In light of the findings, the squared residuals lagged conditional variance, and constant terms show a statistically significant positive effect on the ability to predict the volatility of the Islamic funds using the various models. Furthermore, employing historical information on volatility and the features of the specific market conditions vastly boosts the accuracy of the volatility forecast for the KMI30 index. This research demonstrates that SVM-GARCH hybrid models with linear kernel and neural network model offer high accuracy in Islamic funds volatility forecasting, as indicated by their corresponding root mean square and absolute error. Such implications benefit policymakers and practitioners in the Islamic financial market when policy making uses volatility models. These implications might be applied to risk management, economic stability, and market regulations. Additionally, regarding portfolio investment or financial market decisions, the SVM-GARCH hybrid and neural network model could be utilized in risk management, risk performance, and decision-making. Thus, this study will serve as a foundation for decision-making within the Islamic market