How we can say "pay less"

What the Overpricing Index measures

Every night, Nestraq aggregates the listings its radar has already scored over the last 90 days — per postcode district (outcode) and nationally — into three statistics:

• Median asking-vs-estimate premium — how far the typical asking price sits above (or below) our free sold-price estimate for that home.
• Share of listings reduced — the percentage of tracked listings with at least one observed asking-price cut.
• Median reduction size — how deep the typical cut goes when a listing is reduced.

How each number is computed

The estimate underneath is the same one the app shows: a free, deterministic valuation built from HM Land Registry sold prices for comparable homes near the listing, time-adjusted with the UK House Price Index. The premium for one listing is simply (asking − estimate) ÷ estimate, using only that listing's most recent score in the window, so a re-scored home never counts twice. Reductions come from price changes the radar itself observed — a listing counts as reduced once, however many times it's cut, while every cut feeds the median reduction size.

We publish medians, not averages, so a single mispriced mansion can't drag an area's number. The whole pipeline is deterministic: observed rows in, medians out. No LLM touches the numbers path, nothing is extrapolated, and no individual listing, address or price ever appears in the published index — aggregates only.

The n ≥ 30 rule

No statistic is published for any area — or nationally — until it's backed by at least 30 scored listings, and the reduction statistics carry the same threshold on their own sample. Below that, we publish nothing rather than a shaky number. The gate is baked into the serving code, not left to editorial judgement: a sub-threshold aggregate cannot leave the API.

Coverage, honestly

Nestraqvalues UK listings — not every UK listing. Exact-address matching (UPRN resolution) currently succeeds for roughly half of the listings the radar sees, and only listings we actually scored enter the index. Coverage differs by area and we show it per area in the app; where an area hasn't reached the publication threshold, the index shows nothing for it. As coverage grows, the index simply gets denser — the method doesn't change.

What the wider market shows

Overpricing isn't our invention — the portals' own research documents the gap between asking and agreed prices. Zoopla's House Price Index commentary put it like this:

"This month, the average discount is 5.5%, or £18,000 off the asking price."

— Zoopla, "Seller discounts reach 5-year high of £18,000", 28 November 2023.

That figure is Zoopla's, from Zoopla's data, quoted with attribution — it is not the Overpricing Index and we don't restate it as our own. Once an area's index passes the n ≥ 30 gate, the live Nestraqnumbers stand on their own. Zoopla and Rightmove are trademarks of their owners and don't endorse Nestraq.

Sources and licensing

The estimate layer is built on open Crown data, used under the Open Government Licence v3.0:

• Contains HM Land Registry data © Crown copyright and database right. Licensed under the Open Government Licence v3.0.
• Contains HM Land Registry UK House Price Index data © Crown copyright. Licensed under the Open Government Licence v3.0.

These sources do not endorse Nestraq.

Check it yourself

The published index is a public JSON endpoint — the same gated aggregates, with a provenance block (sources, licence, method, and when it was computed):

GET https://api.nestraq.com/stats/overpricing

If an area you care about isn't there yet, that's the n ≥ 30 gate doing its job — not missing data we're hiding.

The same numbers drive the per-area negotiation-headroom pages; how our estimates settle against what actually sold is published on the receipts page, and the failure modes are stated plainly on the methodology page.