PM Atlas PM Atlas home

Concept · information-theory

Calibration

When stated probabilities match empirical frequencies · events given 70% odds happen 70% of the time. The first-order quality metric for any probabilistic forecaster (human, model, or market), and the basis for Brier-score evaluation.

Key insights

In their words

Polymarket's headline Brier score of 0.047 masks category-specific failures like sports markets scoring 0.325 (worse than a coin flip).· Mandloi
Persistent underconfidence in political markets, where prices are chronically compressed toward 50%.· Nam Anh Le
Some models game the benchmark by copying prediction market prices rather than reasoning independently.· Forecasting Research Institute

Where it matters

Calibration is what every "Brier" headline reduces to, but the 2026 work is decomposing the headline number into more useful diagnostics: domain-specific bias, horizon effects, trade-size effects, platform microstructure. Builders should publish decomposed calibration metrics, not aggregate Brier scores, or they'll be accused of laundering bad sports calibration through good political calibration. Underconfidence in political markets (prices compress toward 50%) is a specific actionable mispricing · a curve-drawing market like Dekant could potentially fix this by giving traders a richer way to express bimodal or skewed beliefs than "buy YES at 52¢."

Connections

Platforms linked to this concept

Related concepts

Sources

Open in the interactive atlas →