Used car price data from public listings, region by region

Make, model, trim and derivative

Mapped to a reference vehicle list. The unit every price comparison is made in.

First registration date or model year

Age drives depreciation. Store the date as printed and derive age at the observation date.

Mileage

The second depreciation driver. Normalise units (kilometres or miles) at collection.

Powertrain and fuel type

Petrol, diesel, hybrid and battery electric depreciate differently, and their residuals are where lenders' risk sits.

Asking price and price changes

Every change with the date first seen. Price cuts on ageing stock are an early signal of softening demand.

Seller type and VAT status

Dealer or private; in some markets whether VAT is reclaimable. Both shift the price for an otherwise identical car.

Location (postcode area or city)

The region the car is offered in, for regional series. Store the area, not the seller's address.

First seen and last seen dates

Your own observation window, from which days on market is derived.

1. 01

   Partition searches by region, make and age band

   Split each marketplace's search into region × make × registration-year bands small enough that no search hits the result cap. Re-split any partition that does.

2. 02

   Sweep result pages daily from inside each region

   Fetch result pages through a city-targeted exit, sorted newest first, and extract listing ID, price, mileage and year from the summary. That pass alone tells you what is new, repriced or gone.

   Region-targeted result sweep17 linescopy

   import httpxGATEWAY, PASSWORD = "proxy.arcproxies.io:8000", "PASSWORD"REGIONS = {  # city tokens from the published targeting list    "de": ["berlin", "munich", "hamburg", "cologne", "frankfurt"],    "gb": ["london", "manchester", "birmingham", "leeds"],}def sweep(search_url: str, country: str, city: str, params: dict, pages: int):    user = f"USERNAME_pool-arc_country-{country}_city-{city}"    with httpx.Client(proxy=f"http://{user}:{PASSWORD}@{GATEWAY}", timeout=30) as c:        for page in range(1, pages + 1):            r = c.get(search_url, params={**params, "page": page, "sort": "newest"})            if r.status_code != 200:                break  # retry this partition later rather than leave a gap            yield r.text  # extract id, price, mileage, year, trim

3. 03

   Open detail pages only for new or repriced cars

   Fetch the detail page when a car first appears or its price changes, with images blocked, to read trim, equipment, powertrain and seller type.

4. 04

   Normalise and deduplicate

   Map trim text to a reference vehicle, then match adverts across marketplaces on make, model, year, mileage within a tolerance and region. Treat a match that reappears within a set window as one spell.

5. 05

   Build mix-adjusted series

   Fit price on age, mileage, trim and powertrain per segment, and publish the adjusted price, stock and median days on market per region and week. Keep the raw medians alongside, labelled as such.

Prices are list prices in EUR, excl. VAT. Volume discounts apply automatically at checkout — full ladder on the pricing page.