How we value a property
Every figure on Stede is produced automatically from public records. No one visits the property and no private information is used. This page sets out exactly which data goes in, how the calculation works, and where it is most likely to be wrong.
The data behind an estimate
- HM Land Registry Price Paid Data
- Every residential sale in England and Wales registered since 1995, published monthly under the Open Government Licence. It gives us this property's own sale history, its type (detached, semi-detached, terraced or flat) and its location. Where the same sale is reported twice by the Land Registry's two collection processes we keep one copy; a standalone additional entry — a repossession, or a corporate or buy-to-let transfer — is kept as the only record of that sale. It also gives us the recent sales of similar homes nearby.
- UK House Price Index
- Produced jointly by HM Land Registry, the Office for National Statistics, Registers of Scotland and Land & Property Services Northern Ireland. It reports an average price and an index for each region, broken down by property type, for every month back to the 1990s. It is published around two months in arrears, also under the Open Government Licence.
- Energy Performance Certificates
- The register of domestic energy certificates, published by the Ministry of Housing, Communities & Local Government. It gives us a property's floor area — and the floor area of the homes sold around it — where a certificate has been lodged.
- Addresses
- The address list you search is built from the Price Paid Data itself. Stede holds nothing else about a property: no condition, no photographs, no listing history and no personal data.
The data is refreshed twice a month, a few days after the government republishes it.
Matching a property to the index
A property's type is taken from its most recent sale record. Its region is its town, when the index has an unbroken monthly series for it — otherwise its district, otherwise its local authority as the ONS identifies it from the postcode, otherwise its county. The valuation and the value-over-time chart always use the same regional series, so the two never disagree about where the property sits.
The evidence
An estimate draws on up to three pieces of evidence. Each property page lists the ones used for it, with what each suggests on its own and how much it counted for.
Its own last sale, re-indexed
Takes the most recent sale of this exact property and moves it forward to today using the House Price Index for its type and region.
last sale price × ( today's average price for this type and region ÷ the average price for this type and region in the month of that sale )
The index is an average, and homes like this one drift from it in predictable ways: those resold within a couple of years have often been improved and sell above it; flats last sold decades ago tend to sell below it. So the figure is corrected by how far, on average, homes of the same type and time since their last sale actually landed from the index.
- Strong when
- the last sale is recent and was a normal open-market sale.
- Weaker when
- that sale was a long time ago, was not at arm's length (a transfer between family, or a probate sale), or the property has since been extended, refurbished or left to decline.
Similar homes sold nearby
The median price of homes of the same type sold in the last two years, each brought up to today with its own local index. We use the closest area that has enough of them: the same postcode (at least 3 sales), else the same postcode sector (5), else the same postcode district (10).
- Strong when
- the street is made of similar homes and plenty have sold recently — above all when this home's own last sale is old.
- Weaker when
- the home is unusually large or small for its type, or the nearest comparables are a postcode district away.
Price per square metre
The same nearby sales, measured by price per square metre of floor area from their energy certificates, multiplied by this home's own floor area.
- Strong when
- the home is bigger or smaller than its neighbours — size is exactly what the comparables' median misses.
- Only available when
- this home and enough of the nearby sales have an energy certificate on record.
Weighing the evidence
The estimate is a weighted average of whichever evidence a home has. The weights aren't chosen by hand: for every group of homes — by how long ago they last sold, and which evidence they have — we try every mix in steps of a quarter against real sales from an earlier two-year period, and keep the one that came closest. A home last sold three years ago leans mostly on its own sale; one last sold in the 1990s leans mostly on its neighbours.
The weights are fitted on that earlier period and then judged on the later one, below, so the accuracy figures measure the method on sales it has never seen.
Sales treated with caution
Some recorded prices say little about what a home is worth on the open market. Three kinds are marked in a property's sale history:
- additional price paid entries — HM Land Registry's category for repossessions, buy-to-let and corporate transfers
- bulk transfers — several homes sold together on the same day at the same price
- recorded prices under £10,000
The own-sale evidence uses the most recent sale that isn't one of these. If a home has only those, a cautious sale still counts, with its own (wider) likely range, except a price under £10,000, where the estimate rests on the nearby evidence alone. The marks state what the record says; they aren't a judgement on anyone involved in the sale.
Bringing it up to this month
The House Price Index is published around two months in arrears, so its latest month is always a little behind today. Every estimate is dated with the month it was published, at the prices of the index's latest month.
We test bringing it forward to the current month: predicting each sale with the index as it stood a few months before, then projecting it to the month of the sale on the area's trend over the past year, that trend with the usual seasonal pattern, or the national trend. So far none has come closer than leaving the figure where it was, so the estimate isn't projected; we repeat the test each time the data is refreshed. On the value-over-time chart, the months since the index's latest are drawn dashed.
How accurate the estimates are
We test the method the way it is used: for about a million homes sold in England and Wales over the last two years that had sold before, we worked out what Stede would have estimated just before the sale, using only the sales and index figures available then, and compared it with the price it actually sold for.
8.6%
median difference
56%
within 10% of the sale price
82%
within 20% of the sale price
Re-indexing the last sale on its own, as Stede did before, was typically 10.2% out, with half of homes within 10% and three in four within 20%; weighing in the nearby evidence did better for every age and type of home we tested. Accuracy still depends most on how long ago the property last sold. Even two near-identical homes on the same street sell around 8% apart, so no estimate built from public records can be exact. Figures from Stede's November 2026 test; we re-run it each time the data is refreshed.
The likely range
The range is measured, not guessed. From the test above we know how far estimates like each one tend to be from the real sale price: for homes with the same evidence behind them, and within that for each combination of property type and time since the last sale, or separately for homes whose last sale is treated with caution or out of line with the property's own history. The range covers the middle 80% of those differences: four in five similar homes sold inside theirs. Homes with less evidence — no recent sales nearby, or only a sale treated with caution — get wider ranges, because estimates for them are further out. We check this on sales the range wasn't built from: a range measured on one period should still hold four in five sales in the next.
That is why ranges are not always symmetrical: some kinds of home are more likely to sell well above their estimate than well below it, or the other way round.
The confidence score
The confidence score comes from the same test: it reflects how close estimates like this one typically are to the sale price. An estimate typically within 5% scores 80; within 10%, 60; within 15%, 40. Scores of 65 and above are shown as High, 45 and above as Medium, and anything lower as Low.
It tells you how much weight to put on the estimate — not which way prices are heading next.
The value-over-time chart
The line on the chart passes through every recorded sale of the property, apart from those treated with caution, and ends at today's estimate. Between two sales it follows the regional index forward from the earlier one and back from the later one, leaning on whichever is closer in time; before the first sale it follows the index back (never before 2000). The shaded band is the line's own likely range, measured the same way against real sales: it is zero at a recorded sale, widens with every year away from one, and today matches the estimate's likely range.
When we don't publish an estimate
Some properties get no figure. Their page still shows the property's details and recorded sales, and says why:
- estimates for homes like it were typically more than 25% out when tested — too far to be useful
- its latest recorded sale is under £10,000 — a price that says little about its value — and the other evidence we have isn't enough to value it reliably instead
- the House Price Index doesn't cover its area
Refining an estimate
On a property's page you can answer questions about the home to refine its estimate for yourself. Your answers never change the estimate anyone else sees.
A few answers can change the figure: the type of property, its floor area, its number of rooms when the floor area isn't known (we then use the typical floor area of homes like it nearby), and an extension or conversion since it last sold. Each is tested first, like everything else here: we take real sales, work out what the answer would have done to the estimate, and use it only if estimates with it were more accurate than without. An owner's figure won't be as exact as an assessor's, so the floor-area test assumes answers up to a fifth out either way, and a refined figure's range comes from those same tests.
The other questions, about condition, parking, outdoor space and more, can't change the figure yet: public records don't show how they affect sale prices. If you let us keep your answers, we compare them with what homes go on to sell for, to learn what each is worth. How we handle your answers
What the estimate cannot see
Because it works only from recorded sale prices, energy certificates and a regional index, the estimate is blind to a great deal:
- the property's condition, layout and interior
- its plot size and aspect, and its floor area where no energy certificate records it
- extensions, conversions and planning permission granted since the last sale
- leasehold terms — remaining lease length, ground rent, service charges, cladding
- differences between one street or block and the regional average
- what buyers are paying right now, ahead of the published data
- incentives bundled into a new-build's first sale
- errors in the underlying public records, which pass straight through
When you refine an estimate, you can tell us some of these yourself, such as the floor area and any extensions.
It is not a RICS valuation, a Home Report or a mortgage valuation, and should not be relied on for lending or legal purposes. For a figure you can act on, instruct a chartered surveyor.
Every number on a property page can be traced back to public sale records of that property and its neighbours, energy certificates and a named, published index. Search for a property →
