Methodology.

MSRP

Every trim in the catalog stores the source its MSRP was read from and the date it was last checked against that source. Both are printed beside the figure on the page, so a reader never has to assume a number is current.

A figure checked within the last 120 days is labeled “Current MSRP”. An older one is labeled with the model year it belongs to. This distinction matters more than it first appears: every projection on a price-history page is the MSRP multiplied by a residual percentage, so a stale MSRP would quietly shift every number on the page while presenting itself as current.

Prices are rounded to the nearest thousand dollars. For a trim no longer in production, the figure is the last MSRP that model was sold at — a historical price, labeled as one. Those pages exist because the buyer researching a discontinued car is shopping the used market, which is precisely where a depreciation curve is useful.

Depreciation curves

A curve is a six-point series — years zero through five — expressed as residual percent of MSRP. It is built at the brand level, not the trim level, and anchored on that marque’s volume model. The anchor model and the full source citation are printed directly under every chart, along with any caveat specific to that marque.

An anchor can outlive the car it is named after. Five model years of residual history is the minimum a curve can be built from, so when a marque retires its volume model the baseline stays on the car that has the history — the replacement has none, and drawing a line for it would be the invention this page exists to rule out. Three of the eleven baselines are anchored on a car you can no longer order: the Ferrari Roma (through 2024, replaced by the Amalfi), Lamborghini Huracán Tecnica (through 2024, replaced by the Temerario), and Maserati Ghibli (through 2024, not replaced). Every chart drawn from one of those says so beneath the curve rather than leaving the reader to notice.

The underlying figures are composites of published industry residual data — Hagerty, Kelley Blue Book, Edmunds, CarEdge and iSeeCars, depending on what each marque has coverage in. Where those sources disagree materially, the citation says so and gives both ends of the range rather than presenting a single figure with more confidence than the evidence supports.

A trim page then applies its brand curve to that trim’s MSRP and discusses, in writing, where the trim is expected to diverge from the baseline and why. A GT-class 911 does not trace the 911 Carrera’s curve; the page says that in prose rather than drawing a second line that no source supports. No per-trim depreciation percentage is invented. This is the single firmest rule in how this site handles numbers.

A curve, worked through

Describing a method is not the same as showing one. Below is a single published curve traced end to end — the figures that went in and where they came from, the figure that was available and deliberately left out, and how the rest were reconciled into the series this site publishes.

It is shown on the marque where the sources disagree most, rather than one where the answer was obvious.

Spec tiers

A brand curve describes a typical example. Configuration moves a specific car off that line — a documented paint-to-sample color, a manual gearbox, a Mulliner or Manufaktur commission. The calculator handles this with 4 tiers applied as a multiplier to the projected resale value, not to the depreciation itself:

• Standard spec — ×1.00
• Sought-after spec — ×1.05
• Highly sought-after — ×1.15
• Collector-grade — ×1.30

These are editorial estimates tracking observed dealer ask premiums for confirmed-rare configurations, anchored to 2026-Q2. Configurator codes shift constantly and the right tier for a given car is a judgment rather than a formula, which is why the tier is a selector the reader sets rather than a value assigned on their behalf.

Live market data

Where a trim has enough current inventory, the page shows what that car is actually listed at. Listings come from the Auto.dev API, refreshed every 7 days and covering the last 6 model years. The page reports the median and the tenth and ninetieth percentiles, plus a per-model-year breakdown.

Two floors govern whether anything appears at all. A trim needs at least 10 current listings before any live figure is shown, and a single model year needs at least 3 before it gets its own row. Below those thresholds the page shows the projection and says that is what it is showing.

Two deliberate choices sit behind this. First, observed listing prices are never merged into the projected curve — they are rendered as separate figures, because blending a measurement with an estimate produces a number that is neither. Second, the query that finds listings is a plain make-and-model match with no per-brand tuning. Some marques are consequently not reachable through it — a Mercedes-AMG or a Range Rover is listed under its parent make — and those pages show the projection only. Widening the query until the count cleared the threshold would mean labeling a model-wide median as a specific trim’s price, which would be a more precise number and a less true one.

Auction results

Some marques cannot carry a residual curve honestly. Where annual production runs to tens of cars, allocation closes at announcement and the secondary market trades at or above delivery price, no published residual series exists — and none is fabricated to fill the space. Those brand pages show completed public auction results instead.

Each result is cited to the selling house’s own lot or results record rather than to an aggregator, and is shown in the currency that house published, never converted. A conversion would introduce an exchange rate and a date that were not in the source. Auction results are evidence of what specific cars made on specific days; they are not a valuation of any other example, and the page says so.

What is measured and what is judgment

Figures carrying a source line — MSRP, residual percentages, listing medians and percentiles, auction results — are sourced claims, and the source is named on the page that makes them.

Other things on these pages are editorial judgment by The Marque Editors: that year three is a marque’s buying sweet spot, that a configuration belongs in one spec tier rather than another, how a manufacturer allocates its cars, how a model is positioned against its rivals. These are informed by the data above, but they are readings of it rather than measurements. They are worth what the reasoning behind them is worth, and a reader is entitled to disagree with them while still relying on the figures.

Limits

Stated plainly, because a methodology that only lists strengths is a sales document:

• Mileage and condition are not modeled. Two cars of the same year and trim can sit far apart on mileage, service history, accident record and cosmetic condition. None of that is in these figures.
• Coverage is the United States. Pricing, allocation and dealer structure differ materially in other markets.
• Listings are asking prices. What a dealer advertises is not what a car transacts at, and the gap varies by marque and by market conditions.
• Options are handled broadly. 4 tiers cannot capture a full configurator, and a genuinely unusual commission may sit outside all of them.
• Curves are brand-level. A trim that diverges sharply from its marque’s baseline is described in prose rather than drawn as its own line.
• Thin trims show projections only. Below the listing floor there is no live evidence behind the figure, and the page says so rather than filling the gap.

None of these figures should be treated as a valuation of a specific car. They are a starting point for a conversation with a dealer, a specialist or an inspector — which is what research is for.

Corrections

If a figure here is wrong, it gets corrected and the correction is dated on the page. The policy behind that — along with independence, dealer referral disclosure, AI review and bylines — is set out in editorial standards, and who makes these decisions is set out in about.