A dual-criteria method for forecast combination: statistical and economic perspectives

This paper proposes a novel forecast combination approach that integrates both statistical and economic evaluation criteria. Applying a rich set of climate-related predictors to forecast energy stock volatility, we show that the proposed method consistently outperforms a range of competing combination techniques. Our results provide strong evidence that climate indicators possess substantial predictive power for energy stock volatility. Moreover, further analysis reveals that information related to atmospheric pressure, drought, and solar energy exerts a particularly pronounced influence relative to other climate-related factors.