Bringing probability judgments into policy debates via forecasting tournaments

Abstract

Forecasting tournaments, which incentivize individuals or teams to make accurate predictions about specific events, can improve the accuracy of forecasts and provide better insights for policy decisions. These tournaments address the problem of vague verbiage predictions by requiring participants to produce numerical probability estimates, and they have several key findings. Firstly, methods like prediction polls and log-odds weighted-averaging can extract wisdom from crowds. Secondly, some forecasters are consistently more accurate than others, and certain personality traits are associated with these top performers. Thirdly, participants learn and improve their accuracy over time, even for unique and complex events. Additionally, forecasting tournaments can be used as a tool for depolarizing political debates and resolving policy disputes by introducing accountability and promoting cognitive complexity among participants. – AI-generated abstract.