“…Not surprisingly, large number of studies have looked into forecasting not only the daily conditional volatilities of gold and oil based on univariate and multivariate models from the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) family, but over the last decade, a burgeoning literature has focused on predicting realized volatility derived from intraday data, 1 using the Heterogeneous Autoregressive (HAR)-type model of Corsi (2009) (see for example, Pierdzioch, Risse, and Rohloff (2016), Degiannakis and Filis (2017), and Fang, Honghai, and Xiao (2018) for detailed reviews). Against this backdrop, given the evidence of significant volatility spillovers across the gold and oil markets, and the importance of geopolitical risks for asset markets (Car-ney, 2016), the objective of this paper is to forecast both the volatilities and co-volatility of the two most-traded commodities by incorporating the role of spillovers and geopolitical risks into the model specification.…”