Methodology

How VehiSales Research figures are produced.

Data source

VehiSales aggregates vehicle listings from dealer and private-seller sources across the United States — over one hundred million listing records observed, with millions live at any moment. Every listing carries an asking price, mileage, location (ZIP code), specifications and, in most cases, a VIN.

Markets

The Price Index tracks the 50 largest markets by live inventory. A market is defined as all ZIP codes within 50 miles of the city’s center — precisely the same geography (and the same active-with-photos filter) as the default listing view on the corresponding “Cars for sale in {city}” page, so every headline count in the index can be verified by clicking through. Neighboring cities whose 50-mile circles cover mostly the same inventory (over half of the smaller set) are treated as one agglomeration — only the largest view is listed (e.g. Phoenix absorbs Scottsdale; Dallas absorbs Fort Worth and Arlington), and the next largest distinct metro takes the freed slot. Residual partial overlaps may remain; “National” figures aggregate across the union of all tracked-market ZIP sets. Markets with fewer than 200 qualifying listings in a given month are excluded from that issue.

Metrics

• All headline figures are used cars only. New-vehicle inventory is excluded from every price, mileage and change metric (it is shown separately as the “all listings” count and median for transparency).
• Live listings (all / used) — “all” is the exact number shown on the linked market page (unique active listings with photos in the 50-mile ZIP set, no other filters); “used” is the subset the index is computed on.
• Used median price — the middle asking price of live used listings with photos and a plausible price ($500–$500,000), computed on the latest version of each listing. Medians, not averages: a handful of exotics cannot skew the figure.
• Month-over-month, like-for-like (constant mix) — for each make & model with enough newly-listed used cars in both windows (last 30 days and the 30 before), we compute that model’s change in median asking price, then aggregate the per-model changes as a sample-weighted median. Comparing the same models across windows means a shift in what arrives on the market (say, a wave of new trucks) cannot masquerade as a price move. When too few models qualify, we print “—” rather than a noisy number; the matched sample size is shown on hover.
• Middle 50% — the 25th–75th percentile band of used asking prices.
• Used median mileage — computed over used listings with odometer > 100 miles (guards against odometer glitches).

Model Price Trends (26-week report)

• Scope — the 200 highest-inventory make & model pairs by unique live VINs; models are matched by VIN prefix via the NHTSA vPIC decode.
• Window — the last 26 weeks of listing observations. Weekly average asking prices are weighted by that week’s listing count for the model.
• Change figure — the percent difference between the average of the first four weeks and the average of the last four weeks of the window.
• Quality gates — weeks with fewer than 20 listings for a model are excluded; a model must have at least 8 qualifying weeks to appear in the report.
• Refresh — recomputed nightly from live inventory; the report page states its exact window end date.

Known limitations

• Asking prices, not transaction prices. Negotiated discounts are not observed.
• Per-market figures are computed on the latest version of each listing; national and per-model aggregates are listing-level and may include a small share of superseded versions between refresh cycles.
• Month-over-month compares newly-listed cohorts, not repeated observations of the same physical car; within-model composition (trim, year, mileage) can still shift between windows.
• Every issue states its exact 30-day window bounds (shown on the issue page and in the JSON download) — figures are reproducible against a dated snapshot.
• This is the first monthly issue — year-over-year comparisons will appear as history accumulates.
• Market definitions follow seller-reported city; suburbs with their own city names are tracked separately.

Citation & press

All figures are free to publish under CC BY 4.0: attribute “VehiSales Used Car Price Index” and link to the issue page. Machine-readable exports (CSV and JSON) are linked from every issue. For custom data cuts (specific metro, model, or timeframe), embeddable charts, or interviews: [email protected] — media requests are typically answered within one business day.