Wisdom of the Crowd

Individual forecasters each hold partial, imperfect information and are subject to their own cognitive biases. When their estimates are aggregated, random individual errors tend to cancel out while shared accurate signals reinforce each other. This statistical cancellation is why simple averaging of independent prior estimates can dramatically outperform any single prediction. The effect is strongest when participants have genuinely diverse information sources and form their views independently. Prediction markets implement this principle mechanically: each trader risks real money on their private information, and the Market Price that emerges aggregates these distributed signals into a single probability estimate. The result is a form of distributed Information Aggregation with strong financial incentives for accuracy.

The wisdom of the crowd requires diversity, independence, and a good aggregation method. When these conditions fail, crowds can be systematically wrong. Herding behavior, where participants copy visible prices or opinions instead of thinking independently, reduces the effective number of independent signals and can produce Mispricing. Forecast Aggregation research from the Good Judgment Project found that trimmed means, extremized averages, and performance-weighted combinations all improve on raw crowd averages. Prediction markets on Kalshi and Polymarket demonstrate the effect in practice: contested political and economic markets tend to be well-calibrated over time, with prices outperforming polling averages and expert panels on a range of event categories.