fightcurve

Does the forecaster know what it is talking about

Every contest the forecaster priced before it happened, against what happened. Where a betting market closed a line on the same contest, the market is beside it.

When it said 70%, did he win 70% of the time?

Each contest is filed under the probability the forecaster gave its favourite, in steps of five points, and the dot shows how often that favourite won. A dot on the dashed line is a forecast that means what it says; above it, the favourites were better than the number; below it, worse. The size of a dot is how many contests sit in it. The forecaster's dots are the contests the market also priced, so the two are on the same fights; the faint line is the forecaster on every contest it priced. the forecaster   the market's closing line.

Year by year, against the market

The Brier score is the average squared miss of a probability: a forecast of 70% on a fight the favourite wins misses by 0.3, and squares to 0.09. Zero would be perfect; a coin scores 0.25. Lower is better. Only contests the market also priced are counted here, so the two lines are on the same fights. the forecaster   the market.

The market is the better forecaster on the fights it prices, and has been every year. It sees things the forecaster does not: injuries, camps, weigh-ins, and the forecaster's own ratings among everything else. The forecaster prices every professional contest in the corpus, most of which no market ever touched; the market priced about one contest in twelve.