Which boats in a host's fleet are fast and which are slow, and by how much. High school regattas are often hosted at a club or sailing center, so the host here is the venue that owns the boats, as Techscore lists it. Every result a team has had in a boat is compared with what its sailors' ratings predicted, with the team's own form that weekend removed, and the boats of the fleet are fitted together inside the rating model. The number shown for each sail is what a team should expect to gain or lose per race, in the next regatta: the fitted effect scaled to the 40% that held up when the model was tested on regattas it had never seen. A fleet is split where it was replaced (a confirmed new fleet, or a multi-year gap in hosting); the current fleet is shown first. How this is estimated, and how much to trust it.
Places per race a team gains (green) or loses (red) in that sail, compared with what its sailors' ratings predict. The bar is the estimate; the whisker is one standard error. A sail whose whisker crosses zero is not clearly different from an average boat.
The same sails (rows, in the order above), with each season's realised effect in places per race: green fast, red slow, grey not sailed with a rotation. The number in a cell is the regattas it rests on. Dashed lines mark where the fleet was replaced: a sail number before the line is a different boat from the same number after it. Within one fleet, a row that stays the same colour is a boat whose speed is real; a row that flickers is noise.
