HYBRID GARCH Models and Intra-Daily Return Periodicity

We use the HYBRID GARCH model of Chen, Ghysels, and Wang (2009) to predict future volatility at daily horizons using intra-daily returns. The latter requires us to address intra-daily periodic patterns. We propose two approaches and compare their relative merits. The first approach uses raw intra-daily data—with the HYBRID process capturing the intra-daily periodic patterns—whereas the second approach involves pre-adjusted intra-daily returns. We find that the former approach dominates both in-sample and out-of-sample, although for different HYBRID GARCH model specifications.