Crew ratings
How the crew rating on a sailor's page is built and how to read it.
What it measures
How much faster or slower boats finish with this sailor crewing than their skipper's rating predicts. Skipper ratings come from the main model; crew ratings are fitted on what is left over.
How it is computed
- In every boat-race the skipper's rating gives an expected finish. The gap between that and the actual finish is shared between the crew aboard.
- Each crew has one effect,
e, estimated together with a per-skipper offseta, so a crew is not credited for a skipper who simply beats their rating. - A prior (spread 0.12) pulls every effect toward zero. A crew with few races stays near the middle: the page marks these as limited evidence.
The number on the page
rating = team skipper average + 18.7 × (crew effect − roster mean)
Crews on one team differ far less than skippers do. The 18.7 stretches the spread so a team's crews vary like its skippers (it is the ratio of the two spreads inside teams). Use the rating to line crews up across teams, not to rank them within a team.
Reading the columns
- Range. Dot is the rating, band the 90% range, dashed line the team's skipper average. A wide band means little evidence.
- League percentile. Rank among all rostered crews in that league and season. It also reflects how strong the team's skippers are.
- Team percentile. Share of teammates the crew beats that season, ties counting half.
- Career line. The effect over all races with rated skippers, not scaled to a team, as a percentile of crews with 30+ races.
When there is no season rating
A season rating needs 8 or more crewed races in one season on a team with at least 3 such crews. Seasons from a linked high-school or college profile are included, so the rating follows a sailor between schools.