Limiting Extreme Behavior in Forecasting Competitions

Although the objectives of forecasting competitions vary, the primary goal of competitors is to win. As a result, competitors may take large risks in the hope of earning a top prize. But this is in sharp contrast to most business settings, where extreme behavior is not tolerated. To address this conflict, authors Schneider, Browell, and Rankin provide ideas on restructuring incentives to limit extreme behavior. They propose limiting downside risks and bad forecasts, rewarding long-term performance and statistical significance, and linking forecasts to value-add. In addition, they find that the recent HEFTcom and M6 competitions connected forecast skill to performance in decision making, but could be improved to limit extreme behavior.