The whole formula is on this page. If you disagree with a grade, you should be able to check it yourself in about a minute.
When a car sells at public auction and the same VIN then appears in a dealer’s public listing, both numbers are on the record. The spread is:
spread = (dealer asking price − auction result) ÷ auction result
A dealer’s grade is the median of its spreads — median rather than average, so one unusual car cannot define a business. The market median across all 51 measured cars is +24.1%, and the bands below are set around it.
A dealer’s listing is current. The auction behind it might be four years old. Over four years a collector car moves with its market, so the gap between those two numbers is mostly appreciation, not the dealer’s margin — and grading it as margin would be plainly wrong.
Measured here: cars whose auction is more than a year old carry a median spread of +36.7%, against +24.1% for recent ones. One dealer moved two full grades once the stale rows were removed.
So only auctions within 365 days of the inventory snapshot (2026-07-31) are measured. 15 cars were excluded on this rule, and 1 more had no usable auction date.
A dealer needs 3 or more measurable cars to receive a grade. Below that it is listed as not rated. Thin data never produces a poor grade here, because an absence of evidence is not evidence.
We see: public auction results, and public dealer asking prices for the same VIN.
We do not see: what a car finally sold for after negotiation, what the dealer spent on reconditioning, transport, warranty or floorplan, or whether an auction result ever settled. Every one of those sits inside the spread.
So the grade answers one question: how far above a recent auction result is this dealer’s opening number, compared with other dealers? That is useful to a buyer and it is all it is.
Any dealer can dispute any row. Send the VIN and what is wrong; if the record is mistaken it is corrected and the grade recomputed, and if it stands the dispute is published next to it either way.