Version 2.6 · TradeBasis
This document describes, in full, how a TradeBasis vehicle market evidence report is produced. Re-running the same inputs against this methodology reproduces the same figures exactly. Reports state observed market data only; they contain no opinion of value.
Comparables are live retail listings for the same year, make, model and, where decodable from the VIN, the same trim. Search proceeds in widening tiers until comparables are found:
Local-first weighting. Listings from the same province as the subject vehicle are prioritized when the comparable set is selected and capped, and same-province listings are exempt from outlier trimming. Out-of-province comparables are included only where the local sample is insufficient; they are marked as such in the report and rank after local listings.
Every comparable used is printed in the report with its location, odometer, days on market, asking price, and adjusted figure. Nothing is withheld.
Per-seller cap. No single seller may contribute more than 3 comparables to the final set, so that one dealer's pricing policy cannot dominate the sample. The cap is applied after all other selection rules. Where a seller exceeds the cap, the listings retained are those closest in kilometres to the subject vehicle; ties are broken by listing URL, ascending, so the same inputs always retain the same listings. Listings whose seller cannot be attributed are not capped. Each report discloses the pre-cap listing count, the number of distinct sellers, and how many listings the cap excluded.
TCUI regional index. The TradeBasis Canadian Used-vehicle Index (TCUI) is a per-region index published against a 100.00 baseline. It is published ahead of, and independently of, any individual report request. Each comparable's asking price is normalized by the factor 100 / TCUI(region) for the requestor's region as it stood on the report's effective date, so that regional market temperature neither inflates nor deflates the printed figures.
The subject vehicle's mileage adjustment is fitted from the comparable set itself, not taken from a fixed curve. Asking price is regressed against the natural logarithm of kilometres by ordinary least squares over the final retained set — the same single fit that produces the §7 attribution figure and the report's scatter chart. The adjustment is:
adjustment = slope × (ln(subject km) − ln(comp-set average km)) × r²
The r² factor weights the adjustment continuously by how much of the price variation mileage actually explains in this set. There is no binary r² cutoff: two nearly identical comparable sets can never produce discontinuously different figures.
The adjustment is suppressed — and the report prints the market figure unadjusted, stating so — under any of these conditions, checked in order:
insufficient_sample).subject_outside_range).slope_not_negative).below_materiality).Two outer sanity rails are retained from earlier versions and applied before the materiality check: the adjustment is capped at $0.15 per kilometre of odometer difference, and at +20% / −50% of the market average asking price.
Days on market is how long each listing has been advertised unsold. A listing sitting past 60 days is priced above what its market will pay, so its asking price is not taken at face value: it is adjusted toward market at 0.03% per day beyond day 60, capped at 3.6% (120 days). Listings under 60 days are used as advertised.
The report prints the full comparable table, the median of the adjusted figures, and the range (lowest to highest adjusted figure). No weighting, trimming, or discretionary exclusion is applied beyond the rules above.
All figures are expressly conditional on the requestor's declared condition; no physical inspection is performed. A report is effective-dated to the stated date of loss and expires 30 days after issue. A market evidence report is not a signed FIN-320; British Columbia requires that form be signed by a licensed dealer or professional appraiser.
Each report states what share of the price variation across the comparable set is accounted for by mileage. Asking price is fitted against the natural logarithm of kilometres by ordinary least squares across the final retained comparable set, in canonical order (listing URL ascending), with integer kilometres and integer dollars. The published figure is the squared correlation of that fit (r²), rounded to four decimal places and displayed as an integer percentage. This measure describes the comparable set only; it never filters, reweights, or alters it, and it expresses no opinion of value.
The percentage is classified into one of three fixed bands, applied to the displayed integer percent so the band can never disagree with the number printed beside it:
The measure is suppressed — and the report states the refusal rather than omitting the section — under any of these conditions:
insufficient_sample).no_mileage_variance).no_price_variance).invalid_mileage).The accompanying scatter shows each comparable's asking price against kilometres, the fitted logarithmic curve, a band of ±1 residual standard deviation, and the subject vehicle's odometer position. When the attribution falls in the weak band (0–39%), the fitted curve and residual band are not drawn: the chart shows the comparables and the subject position only, and states that no fitted relationship is shown. Only the computed metric enters the report's content hash; explanatory prose is rendered from a versioned template at display time, so copy corrections do not invalidate verification of previously issued reports.