Abstract
This study examines key characteristics of stock market time series, such as regime shifts and non-linearity, which necessitate specialized methods for capturing market volatility. To improve volatility forecasting for the Muscat Securities Market Index (MSMI), the paper proposes a Back-Propagation Neural Network (BPNN) model. The neural network takes as input the volatility estimated by the Markov-Switching GARCH (MS-GARCH) model and uses the Close-Close volatility estimator as the output. The findings indicate that incorporating the neural network enhances the forecasting accuracy of the MS-GARCH model, as measured by Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). The main contribution of this study lies in the integration of machine learning with traditional econometric models to improve volatility forecasting. Specifically, it demonstrates that augmenting the MS-GARCH model with a BPNN significantly enhances predictive accuracy, offering a hybrid approach well-suited to capturing the complex dynamics of financial time series.