Different pages answer different questions, so they count different cohorts. A country page counts boats by certificate provenance — the national flag on the paperwork. A fleet or race page counts a competitive cohort — the boats that sail together, which can include entries from other countries. Two totals that differ are not a contradiction; they are two well-posed questions with two correct answers.

That only works if every figure declares its scope. Across FleetEdge, each figure carries three things: the cohort it counts, the denominator behind any percentage, and the date of the published data view it comes from. This page defines the terms those labels use.

The terms behind the counts.

ORC-rated

A boat carrying an ORC certificate in the analyzed corpus. ORC-rated boats bring measured hull, rig, and stability data — the physics foundation of the analysis.

Mapped-IRC

An IRC-rated boat matched to its ORC design family using measured hull dimensions — length, beam, draft, and displacement ratio — and carrying a projected physics profile from that family. FleetEdge never reads IRC rating values; the match uses public dimensions only.

Classified / unclassified

A classified boat has been assigned to one of the eight performance archetypes (one of the fleet's eight recurring design types). A boat remains unclassified when its available data is insufficient for a confident assignment. Archetype percentages state their denominator — the classified pool or the whole fleet — wherever they appear.

Race data vs. authority counts

An authority count reflects certificates issued under a national authority — the registry view. A race-data count reflects boats linked to actual event results — the competition view. A boat can hold a certificate and never race; the two counts answer different questions.

Banked race evidence

Race results accumulate. When a reading cites race results, it rests on every scored result banked for that boat — not on a single day’s outcome — so the reading firms up as the boat races more. “Banked” names that accumulation; elsewhere the site says simply race results.

Scope labels

National fleet view counts by certificate provenance (flag). Global fleet view spans the full analyzed corpus. Domestic competitive fleet is a competition cohort anchored to a country’s racing circuit — and can include foreign-flagged entries. Race and event sections count only the boats in that event.

As-of date

Every macro figure cites the date of the published data view it comes from. Race and event sections cite their own publication vintage separately — see below.

Why two totals can both be right.

Consider one boat: flagged in one country, racing a neighboring country’s offshore circuit, entered in a championship abroad. She appears in her flag country’s national fleet view, in the neighboring circuit’s domestic competitive fleet, and in the championship’s race cohort — three pages, three cohorts, one boat, no double counting within any one view.

The rule of thumb: before comparing two FleetEdge figures, compare their scope labels. If the cohorts differ, the numbers are answering different questions — and should differ.

The canonical count ladder.

Four macro boat counts appear across this site, each at its own scope and date — defined here once:

Published records — 14,135 boats in the current publication (as of 2026-11-19). Every boat record carried by the current publication of the FleetEdge data service, whatever its analysis depth.

Governed corpus. Boats admitted through FleetEdge’s identity and eligibility governance — deduplicated, provenance-checked, and eligible for analysis.

Analyzed boats. The governed boats whose certificates support the full 52-dimension analysis; boats whose certificate data cannot support it are held at the governed tier, not padded forward.

ORC boats scored. The ORC-rated subset of analyzed boats scored by PPI (the Performance Potential Index); boats outside that subset stay at the analyzed tier. Scoring at this rung is the certificate-derived Design PPI — race data is not required, so a boat that has never raced still carries a score. The race-dependent readings — Race PPI and the crew residual (a model-adjusted residual that helps investigate execution: what remains after the boat's evaluated pace is accounted for — negative is faster, in s/nm) — are narrower cohorts and state their own coverage where they appear.

Fingerprinted boats — 11,976 in the current publication. The boats whose certificate physics the classification model reads into a fingerprint: the population the eight performance types are counted over (the 2026 census).

Every count on this ladder is the current publication’s and moves with it; the rungs between published records and fingerprinted boats are defined above without a count, because the current publication does not carry one. Validation cohorts are narrower still and say so where they appear: the current fairness-validation cohort (the October 2026 edition) is 17,757 boat-race results across 1,525 class-level fields, and empirical crew-residual coverage spans 1,317 boats in the current publication.

One race-record count appears beside the boat ladder, at its own grain — also defined here once:

Race events. 1,002 race events in the current publication carry at least one of its boats, counted at event grain: a regatta or a series is one event, however many races it holds. The event directory lists events by name, date and venue, and states how its own count reconciles with this one.

Races. This publication states no race total. On some events the published record counts a class’s result table as a race, so the total waits until those records are corrected.

Archetype assignments are dated.

A boat’s archetype comes from its certificate, and certificates are re-measured over time — so the same boat can sit in one archetype for one season and a neighboring archetype after a re-measurement. Race and event sections render the archetype assignments read at that event’s certificate year and say so in their provenance line. When the FleetEdge analysis is dated after the event ended — the case for every past event — the line names that analysis date and calls the grouping a retrospective view: the current analysis reading the boats’ certificates of the event’s year, not a record made at the time. Current fleet surfaces always reflect the latest analysis.

The archetype names themselves are one vocabulary everywhere: an older race section groups its boats as they were grouped then, under the same archetype names used across FleetEdge today. Compare a dated section with other sections of the same date; compare current fleet surfaces with each other.

Reading a race section: the provenance line beneath a Race Archetypes section tells you which dated analysis its groupings come from. Shares and boat counts belong to that dated view — not to today’s fleet view.

What kind of claim am I reading?

Alongside its scope, every analytical card on FleetEdge carries one of five public evidence badges — a closed, stable lexicon that tells you whether a figure is a live fleet-wide measurement, a result under on-site validation, a structural readout without race attribution, a low-sample structural view, or a forward-looking pre-race scenario. The badge lexicon is defined on the How It Works page.

In beta as a product. These plain-language definitions travel with every Media Desk story pack — the press packs the desk composes for journalists and broadcasters. See Media Desk →

Read the fleet with confidence.

Every figure names its cohort, its denominator, and its date. That is the standard.