How PropDetect calculates GDV and end value

Every comparable is an HM Land Registry transaction record, supplied through PropertyData: a price at which a property actually changed hands. We do not build valuations from asking prices, from listings marked sold subject to contract, or from portal estimates, because none of those are evidence of what a buyer actually paid. An SSTC sale can fall through or complete at a different figure; an asking price is a hope.

For each analysis the system pulls up to 100 sold transactions around the subject postcode. That breadth matters: a valuation built on three hand-picked sales is an argument, one built on a ranked pool of a hundred with the rejects shown is evidence.

Each sale in the pool is scored on how well it matches the subject property across four dimensions: property type, bedroom count, distance, and floor area where it is known. A three-bed terrace is valued against three-bed terraces, not against the detached house two streets over that happens to be recent.

Sales that fail the match are excluded, and the report says so in plain English, with the counts: "based on 22 recently sold properties near this 3-bedroom end-terrace house (829 sqft). 49 sales were excluded from the pool." Every comparable in the table shows its sold date, price, distance and floor area, and is marked as used or excluded, so you can disagree with any individual choice and flip it, at which point the valuation recalculates in front of you.

A second, independent route to the value runs alongside: the price per square foot implied by the matched sales, applied to the subject floor area taken from the EPC or floor plan. Size is the biggest systematic difference between otherwise similar properties, and the cross-check catches the case where the subject is meaningfully bigger or smaller than its comparables.

When the two routes disagree, the report shows both rather than quietly picking the flattering one. A disagreement between them is information: it usually means the subject is unusual for its street, and that is exactly the property to be careful with.

The concluded figure is a trimmed, conservative reading of the matched set rather than its most optimistic end. Index data is used only as a fallback sanity bound, never blended into the comparable figure, because an index describes a region and a comparable describes a street.

The same discipline runs the other way: when your plan justifies a premium, the report says so and shows the reasoning, rather than baking optimism silently into the number. A valuation you can defend to a lender is worth more than one that flatters the deal.

Some postcodes simply have not sold enough matching stock recently. When the matched pool is too small, the system widens the sale-date window in stages rather than fabricating confidence from three data points, and the report states that widening happened and how far. The confidence score drops with the evidence, and the methodology text says why in a sentence.

This is deliberate product behaviour, not a failure state. A tool that produces the same crisp number from five comparables as from fifty is telling you about its formatting, not about the property.

Most software in this market does one of two things: hands you raw sold-price data and leaves the judgement to you, or produces an automated estimate without showing which sales drove it. We think the valuable position is the third one: a concluded figure, with every input visible and reversible, and the reasoning written down. That is the difference between a calculator and an appraisal, and it is why the comparable table, the exclusions and the written justification ship on every report rather than behind a support request.