Modeling the volatility of FTSE100 index returns using realized GARCH model with jumps

This study compares the fitting performance of the Realized GARCH(1,1) model with jumps with that of the standard GARCH(1,1) models. The realized volatility measure is based on the realized kernel. Each model is estimated by using the Generalized Reduced Gradient Non-Linear (GRGNL) method in Excel’s Solver and the Adaptive Random Walk Metropolis (ARWM) method in the Markov Chain Monte Carlo scheme. The ability of models and methods is evaluated on the FTSE100 index returns for the daily period from January 2000 to December 2017. The empirical results showed that the GRGNL’s Excel’s Solver method can estimate the model parameters well because the estimation results are not much different from that of the ARWM method. The best fitting model selected by Akaike Information Criterion is RealGARCH(1,1) model with jumps.