A Hybrid B.Spline-Garch (BSGARCH) Model For Stock Market Volatility

Abstract

This study aimed to propose the BSGARCH (1,1) model, a hybrid Basis Spline GARCH-type model suitable for modelling the volatility of nancial time series. The proposed model was compared with other classical GARCH-type models such as GARCH, EGARCH, GJR-GARCH, and APARCH models, and it was found to outperform them in terms of predictive accuracy, as measured by RMSE, MAPE, TIC, and QLIKE. The results showed that the BSGARCH (1,1) model had a superior predictive ability compared to the other models. The study also compared the performance of the BSGARCH (1,1) model with the Spline-GARCH model of Engle and Rangel, which used the exponential quadratic spline to model the non-stationary part of volatility. In this comparison, the Spline- GARCH model slightly outperformed the proposed BSGARCH (1,1) model. However, the di erence in performance was negligible, and thus, the proposed BSGARCH (1,1) model can be considered a good alternative to the Spline-GARCH model. The study demonstrates that the proposed BSGARCH (1,1) model is a useful tool for modelling non-linear and non-stationary nancial time series data, with superior predictive ability compared to other classical GARCH-type models.