Brier score calculator

How the calculator works

For each forecast, the calculator squares the difference between your probability and what happened (1 if it happened, 0 if not), then averages over all your forecasts:

Brier score = average of (forecast − outcome)²

Lower is better: 0 is perfect, 0.25 is what you get by always saying 50%, and 1 is the worst possible. The full explanation, with worked examples, is in the Brier score, explained.

The Brier skill score

A raw Brier score depends on how hard the questions were, so the calculator also compares it with simple baselines on the same questions:

Brier skill score = 1 − (your Brier score ÷ baseline Brier score)

• Against the base rate: the baseline always forecasts how often the events in your list actually happened. If 30% of them happened, it says 30% every time, and scores 0.3 × 0.7 = 0.21. This is the usual reference, called climatology in weather forecasting.
• Against a coin flip: the baseline always says 50%, and scores 0.25.
• Against the market: add the market's probability as a third column, and the calculator scores the market too. A positive skill score means your forecasts beat the market's prices on these questions.

A skill score of 1 is perfect, above 0 beats the baseline, 0 ties it, and below 0 means the forecasts did worse than the baseline.

Input format

One forecast per line: the probability, then the outcome, and optionally the market's probability.

• Probabilities: 0.7, 70% or 70 all mean 70%.
• Outcomes: 1 or 0, yes or no, true or false.
• Separators: commas, semicolons, tabs or spaces. You can paste two or three columns straight from a spreadsheet.

Everything runs in your browser. Nothing you paste is sent anywhere.

The calibration check

The table under your score groups your forecasts by probability (0–10%, 10–20% and so on) and shows how often the events in each group happened. If you are well calibrated, your 70% forecasts come true about 70% of the time. With only a few forecasts per group, expect noise: calibration needs dozens of forecasts to say much.

How Sikt uses the Brier score

Every Sikt forecast is graded when the answer arrives, misses included. The Sikt Midterm Bench puts Sikt's forecasts for the 2026 midterms next to Kalshi and Polymarket, and after election day they will be scored exactly like this, against the markets.

FAQ

How do you calculate a Brier score?

For each forecast, subtract the outcome (1 if it happened, 0 if not) from your probability and square the result. Then take the average over all your forecasts. A 70% forecast on an event that happened scores (0.7 − 1)² = 0.09.

What is a good Brier score?

It depends on the questions. Always saying 50% scores 0.25, so anything below that beats a coin flip. On hard real-world questions, the best human and AI forecasters score well below 0.2. Compare scores only on the same questions, ideally with a skill score.

How do you calculate the Brier skill score?

Divide your Brier score by the Brier score of a baseline on the same questions, and subtract the result from 1. With the base rate as the baseline, the baseline score is the base rate times one minus the base rate.

Can I compare my forecasts with a prediction market?

Yes. Add the market's probability for each question as a third column. The calculator scores the market's forecasts the same way and shows your skill score against them.

Is my data stored?

No. The calculation runs entirely in your browser, and nothing you paste leaves your device.