How prices are computed

The price of a model is the median asking price of its distinct active listings. Active means the listing was still live when the snapshot was taken. Distinct means the same car offered on more than one source is counted once.

Before the median is taken, listings outside the interquartile fence are removed. That fence is Q1 − 1.5 × IQR through Q3 + 1.5 × IQR, applied inside each model and, separately, inside each segment. The snapshot records how many listings that step dropped. On this snapshot the model-level filter removed 1,118 listings.

Listings are read from AutoScout24 and mobile.de. A car that appears on both sources contributes one listing to the median, the supply count, and the market cap. Seller names, listing URLs, photos, and the raw listing rows are not kept in the site data.

Tracked market cap for a model is its median asking price times its active listings.

market cap = median asking price × active listings

The number at the top of the market is the sum of those model market caps: €2,743,209,204 (€2.74B) across 28,261 active European listings and 424 listed models (models known only from auctions have no market cap). It is a way to compare supply at the current ask. It is not the value of cars that changed hands.

A segment or a model with fewer than 8 active listings is marked low liquidity. The median is still shown, with a badge, because a handful of asking prices moves too easily to read as a market. Low-liquidity models stay in the market-cap total. They are left out of top gainers, top losers, and trending once those cards have history to rank.

This snapshot has 472 segments. The badge is applied wherever the active count is under 8.

Every variant (for example 458 Italia, not “the 458”) and every region carries two separate prices, never blended or averaged together:

Asking price: the median of active listings (AutoScout24 and mobile.de today) after outlier removal and cross-source deduplication. The default view is Europe; country views use only that country’s listings. There are no tracked listings for the UK or the US, so asking is — there.

Realized price: auction results from Bring a Trailer, Collecting Cars and Bonhams (Bonhams: hammer price, its buyer’s premium removed and never mixed in), hammer price excluding buyer’s fees, converted to EUR at the ECB rate of the sale date. Regions are EU (EU-located sales), UK and US, never mixed. A region gets a 12-month median only with at least 30 sales in 5 years and 10 in the last 12 months; below that the page shows the count (n), the last sale with its date, and the individual sales as dots, but no median. Houses that report prices including the buyer’s premium would be shown as a separate series, never mixed with hammer prices.

Asking vs realized gap = (asking median − realized 12-month median) ÷ realized 12-month median, only for the same variant in the same region (European listings vs EU auctions), and only with at least 8 listings and 10 sales in 12 months; otherwise —. It is a signal, not a price: positive means sellers ask above what recently sold. UK or US auctions are never used for a European gap.

The default view is Europe everywhere. Realized columns on the home table use EU auctions, else UK auctions (tagged UK); US auctions appear only when US is selected. A choice of region is kept in the page address (?region=).

The period columns (30D, 3M, 6M, 1Y, All) and the change next to the asking price are changes of the asking price: today’s median asking price against the latest stored daily median on or before the start of the period (All = since the first daily snapshot, once 30 days exist). Until that earlier point exists, the cell is an em dash. Nothing is interpolated.

Realized changes are shown separately (Realized 1Y on the home table, the change row in the auction panel): the rolling 90-day median sale price in one region today against the same rolling median at the start of the period, with at least 8 sales in each window, otherwise —. A median covers every sale of the variant, so a shift in the mix of what sold moves it too.

The model page chart draws the asking median as a line and each auction sale of the same region as a dot (tap to open the lot). The two are on one time axis for comparison only; no figure is computed from both.

Top gainers and losers rank the 30D asking change among models that are not low liquidity. Trending is the largest absolute 1Y move of the realized rolling 90-day median in one region. Country views show — for realized prices because auction regions are not countries.

Shown: active listings (deduplicated) and the 30-day change of that count (— until 30 days of daily snapshots exist). Not shown: median days on market, because AutoScout24 does not expose a listing date and our own tracking started recently, so any figure would be biased; and the share of listings with a price cut, until we have tracked prices for long enough to measure it reliably. They will be added only when the data supports them.

A model’s segments split the same listings by generation, transmission, and body when the export distinguishes them. Each segment has its own median, after its own IQR filter and the same deduplication. Segment supply adds up to the model’s active listings. Days on market is shown only when the snapshot includes a median days-on-market field. This export does not, so that column is an em dash.

A brand index is the sum of market caps for the models in that group. Porsche, Ferrari, Lamborghini, Mercedes-Benz, Audi, and the smaller marques are their own groups. BMW M is every BMW model in the snapshot. JDM groups Nissan, Honda, Toyota, Mazda, Mitsubishi, and Subaru. Each level is a market-cap snapshot of today’s European listings. We do not publish index changes or history yet: while we keep adding variants and sources, the summed market cap mostly tracks our coverage, not prices. Index history starts once the constituent basket is fixed.

Auction volume and Vol / Supply on a model page count one auction region only: EU auctions in the default Europe view, or the US or UK when you choose it. Regions are never added together, and Vol / Supply is shown only for Europe because the listings behind it are European.

A deduplicated listing export is compiled into aggregates before the site is built. The browser never receives listing rows, URLs, photos, titles, or VINs. Histograms count cars inside wide price and mileage bins. The price-versus-mileage chart plots a dot only when a bin holds at least 3 listings, and the dot is that bin’s median, not a car.

A median for a year, generation, body, colour, country, or gearbox is published only when that slice has at least 3 listings. Smaller slices can still show a count. Colour names such as “Black Metallic” are grouped with the base colour. € per 1,000 km is the median of price ÷ (km / 1,000) among listings with at least 1,000 km, and only when five or more listings qualify. The mileage discount compares the median ask above a model’s median km with the median at or below it, and only when each side has at least five listings.

Generation Realized 1Y compares the last 12 months of auction prices with the previous 12 for one generation in one region (EU or UK, never pooled), mix-adjusted across its variants. It is published only with at least 3 variants and 15 sales on each side. A robustness check also applies: if either side has fewer than 25 sales and the other European region shows the opposite direction for the same generation, only the region with the deeper five-year sales history is shown (both are withheld on a tie).

“New on mobile.de” counts listings whose mobile.de created date falls on or after 2026-10-30 and on or before 6 Nov 2026: 49 listings. AutoScout24 does not provide a created date, so this is not all-source new supply. The home-table Volume column is that count, not euro turnover. Production volumes are not in the data. The supply bar compares a model’s active listings with the most-listed model, and “thinnest supply” means the fewest active listings, not rarity versus how many were built. 30D, 3M, 6M, 1Y and All changes use a stored median on or before that many days earlier (1Y uses the monthly auction index). There is no 24h change: daily moves in collector-car asks are noise from listings coming and going. Until a window has enough history it shows “—”. Gainers and losers rank by the 30d change. Per-model sentiment is not shown: the index components that score today are market-wide, not a reading for one model.

The index is 0–100. It is the equal-weight mean of the components that have a score, rounded to a whole number. Labels are Extreme Fear through 24, Fear through 44, Neutral through 55, Greed through 75, and Extreme Greed above that. Today three components score. Fresh-supply pace compares mobile.de listings created in the last 7 days, per day, with the per-day rate of the preceding 23 days. A matching pace scores 50 and double the pace scores 100. Listing age is the median age of those ads: 0 days scores 100 and 365 days scores 0. Mileage discount is the listing-weighted average of each model’s mileage discount: 0% scores 100 and 50% or more scores 0.

Asks versus a prior kilometre median, 30-day price momentum, 7-day supply change, the share of listings that dropped in price, and days on market stay unavailable until the snapshot or history actually contains those inputs. Manual share is context only. Each compiled day is stored on the fear-and-greed history series so a later day can draw a line. One day is a point, not a chart. The page is Fear & Greed.

Asking prices are published as aggregates only. Auction results are shown sale by sale (date, hammer price, year, model, km when stated, country, no-reserve flag) with a link to the original lot page. You will not find individual listings for sale, photos, descriptions, seller or buyer names, phone numbers, locations of a specific car, or portfolio and waitlist counts. There is no invented collector, no sample garage, and no illustrative price.