A valuation you can defend in front of a credit committee starts with a transparent method. Here is how comparable-sales models produce one.
A comparable-sales valuation begins with actual wholesale results: vehicles of the same year, make, model and trim family that sold at auction within a recent window. The window matters — wholesale prices move, and a comparable from nine months ago describes a different market. Most working models weight the last 60–90 days heaviest and discard anything too old to be relevant.
No two vehicles are identical, so raw comparables are adjusted before they are used. The standard adjustments are mileage (the single largest driver in most segments), condition grade, region, and seasonality. A truck with 40,000 fewer miles than the comparable set is not worth the set's average; it is worth the average plus the mileage adjustment that the market itself demonstrates.
Honest models report a value with a range around it. The width of that range carries information: tight ranges mean dense, consistent comparables; wide ranges mean thin or volatile data. Treating a point estimate from a thin comparable set as precise is how portfolios get mispriced.
A valuation built on 30 comparable sales and one built on 3 can produce the same number with completely different reliability. That is why every valuation on the AB Imperial platform states the number of comparables used. When you review a figure — yours or a vendor's — the first question is not "what is the value?" but "what is it standing on?"
A good valuation system says when it is unsure. Treat any figure without an explicit range or comparable count as an opinion, not a measurement.