After November 2024 the settled story was that the betting markets called the election and the polls did not. Below is the one contract that story was built on: who would win the Electoral College. You will see how the market price scored against three forecasting models built on polls, and then be asked for a verdict, exactly as the coverage was.
The headline contract · 1 race · $3,686,335,059 staked
The test: a bettor believes one model, bets against the market price every day for the last 90 days, and holds to the result. What would they have made?
Every model was on the wrong side of the price. On average daily closeness to the outcome too (Brier score, where lower is better and a coin flip scores 0.25), Polymarket's 0.229 beat all three models.
1. On this evidence, who forecast the 2024 election better?
2. How widely does your verdict hold?
All figures from Sethi, R. et al. (2025) ‘Political prediction and the wisdom of crowds’, Proceedings of the ACM Collective Intelligence Conference, pp. 214–225. One election cycle, the United States, 2024: these are that cycle's numbers, not a standing result. Silver Bulletin published no congressional forecasts.