This ranks keelboats and match racing, sailed by a skipper with a crew of two or more. Skippers are rated, and each boat's crew and school add to its strength.

There is far less keelboat racing than doublehanded fleet racing, so a boat's strength is built in layers, each pulled toward zero when there is little evidence:

• the skipper's main rating (doublehanded fleet racing), of which about 41% carries over
• the mean main rating of the crew aboard
• the school's program effect: how much better or worse its boats do than its sailors' ratings predict

A sailor or school with few races therefore falls back on the main rating, and the other layers grow as results come in. The list leaves out sailors with fewer than 10 races unless you untick that box. A race from 2 years ago counts half as much as one sailed today.

Boats that did not finish a race (DNF, DSQ, OCS and similar) are left out of that race rather than counted as last. The numbers are on their own scale and cannot be compared with the main rankings.

How well it forecasts

Each version was fitted on races before a season and then asked to call that season's races: for every pair of boats in a race, which finished ahead (14,666 pairs over the last four seasons). A coin flip scores 50% and 0.693; lower log-loss is better. Layers that did not help in this test are not used.