A Combination Approach to Forecasting Precious Metals Correlations

Abstract

We propose a novel combination approach to forecast precious metals correlations at various horizons of relevance for investors and risk managers. The proposed combined predictor is based on the Model Confidence Set and allows to mitigate several sources of uncertainty associated with metal forecasts especially in periods of high market instability. Our results reveal the benefits of this approach for Value-at-Risk and portfolio optimization in terms of forecasting performance at medium and long horizons and in times of intense market instability, such as the recent pandemic crisis.

Keywords: Precious metals correlations, forecast combinations, Value at Risk, macroeconomic fundamentals, structural breaks