Brier score
A way to score probability forecasts: square the probability you gave to what didn't happen, and average over many forecasts. 0 is perfect; always saying 50% scores 0.25.
In plain English
A forecast of 70% isn't proven wrong by one miss, but it shouldn't get away with every miss either. The Brier score charges more for confident misses: a 70% forecast that fails scores 0.49, a 99% forecast that fails scores 0.98. Averaged over hundreds of forecasts, it rewards forecasters who are both well calibrated and willing to commit.
Uncle Gary explains
Mia scored Uncle Gary's family forecasts for a year. He said 100% that Linda's lasagne would be ready by six, that Kevin would be on time and that his team would win the derby. His Brier score is the worst in the family.
Why it matters for your money
Anyone who predicts markets with probabilities can be scored. A forecaster who is never less than 95% sure will have a terrible Brier score long before you notice from the headlines.
Where we tested it
Related terms
All glossary terms. Educational content, not financial advice.