Individual foresight: Concept, operationalization, and correlates

Judgmental forecasting research on superforecasters has demonstrated that individuals differ in their foresight. However, the concept underlying this work focuses on accuracy and does not fully incorporate the time dimension of foresight. We reconceptualize foresight as the ability to predict future states of the world accurately, where accuracy becomes continuously more important over time. To operationalize foresight in forecasting tournaments, we propose various strictly proper scoring rules and compare them with existing scoring rules using a simulation study and real-world forecasting data consisting of 414,168 scores for 9694 forecasters on 498 questions from a four-year geopolitical forecasting tournament. The results suggest that the linear time-weighted Brier score should be the default operationalization of foresight and that probability training and teaming interventions as proposed by prior research may not improve foresight as we conceptualize it. We contribute to judgmental forecasting research by clarifying the concept, operationalization, and correlates of foresight.