How our sold-price figures work

The data and what it covers

Figures use HM Land Registry Price Paid Data, standard sales only - HM Land Registry's "category A". That excludes repossessions and other sales under a power of sale, buy-to-lets identifiable by a mortgage, sales to companies and other non-private buyers, and properties classed as "Other". We also ignore entries under £1,000. Auctions are not excluded: a normal auction sale between a private buyer and seller is a standard sale and is counted here. HM Land Registry only began flagging these categories in October 2013, so almost every sale before then is counted as standard. Price Paid Data also leaves out sales never lodged for registration and transfers not for value - right-to-buy discounts, gifts, divorce transfers and compulsory purchases. Land Registry records some towns under two spellings that differ only in punctuation, such as ST. ALBANS and ST ALBANS; we treat those as one town. England and Wales only; HM Land Registry publishes with a 1-2 month lag.

Each street page shows a "Last updated" date: the date we last added a sale on that street, or, where we have no such date, when our data was last refreshed. It also says which month our HM Land Registry data runs to and when we last refreshed it.

Contains HM Land Registry data © Crown copyright and database right 2021. This data is licensed under the Open Government Licence v3.0.

The figures on a street page

Average price

The arithmetic mean of every standard full-value sale on the street, all property types and all years since 1995 combined. Because it spans decades it typically sits below current values - see the year-by-year trend for the district on the street page for the recent level.

Typical sold price (median)

The middle recorded sale price on the street - half of recorded sales were below this and half above. Less swayed by a single unusual sale than the mean, which matters on a street where a handful of sales is normal - and why the headline and stats lead with it. With fewer than 3 recorded sales no median is stated; the actual prices are shown instead, and with 3 or 4 sales the median is flagged with its sale count.

Total sales

A straight count of the standard sales recorded on the street, with the dates of the first and most recent.

Lowest / highest sale

The single lowest and highest individual sale prices recorded on the street in any year - two transactions, not a typical range.

Every-sale chart

One mark per standard sale recorded on the street: its position is the completion date and the price actually paid. Nothing is averaged, smoothed or projected, and no estimate of today's value is implied. On a street with more than 600 recorded sales the chart, and the sold-again list, cover the 600 most recent.

Sold again (same home)

Addresses with two or more recorded sales, matched on the exact address. The annualised % is the steady yearly rate that would turn the earlier price into the later one over the time between the two sales (compound growth), and is only stated when the sales are at least two years apart (closer pairs show the time between them instead). A rise or fall of more than 15% a year, or of 15% or more on a home resold within a year, is marked as one that may reflect works or an unusual earlier sale. It takes no account of extensions, renovation or condition, and no value today is estimated from it.

Filtered stats bar

The count, median, mean, lowest and highest above the sales table recompute over exactly the sales matching your current filters - the same set the table and the CSV download cover. With fewer than 3 matching sales no median is shown; the lowest and highest are then the actual individual prices.

District trend (% change & chart)

Calculated across the wider postcode district the street is in, not the street alone (one street rarely has enough sales each year for a reliable trend). We compare the district's median sold price in its earliest year with at least 10 recorded sales against its most recent complete such year - a median-on-median comparison. The shorter figure beside it compares that same latest year with the most recent such year at least two years before it. We count only years that have had time to settle: never the current calendar year, and not the year before it until our data reaches March, because registrations are still arriving.

Property-type mix

Share of standard sales by property type across the street's postcode district. Any new-build figure compares new-build and existing homes type by type, each type weighted by how many sales it has, so the mix of types does not drive the gap.

Instant estimate

An automated estimate from recorded HM Land Registry sales nearby, or this home's own earlier sale, brought to today's prices with the UK House Price Index, or the House Price Index average where there are too few sales - not a formal valuation. A local agent's valuation will be more accurate. The full method is below, under "The instant estimate".

Change since previous sale

Where a sale row shows "up/down X% since {year}", that compares the sale price with the most recent earlier standard sale (category A, £1,000 or more) of the same property - not an area trend. Rows in buildings split into flats with no flat number recorded are not paired, because the earlier sale may be a different flat.

The charts

Price trend chart

On a street page this chart covers the wider postcode district, not the street alone. The median is the middle sale price of all standard (HM Land Registry category-A) sales in the area in each calendar year - half of recorded sales were below it, half above. The mean is shown as a secondary line for comparison. Where no median series is available the line is the mean price instead, and a few expensive sales can pull it upward.

Years with no sales are left as gaps; years with fewer than 3 sales get no plotted point and no typical price (hovering shows their sale count); years with fewer than 5 sales are drawn hollow and flagged. The year-on-year panel compares each year's median (or, where there is no median series, average) with the year before, only where both years had at least 10 sales, and skips the incomplete final year. These are the prices of what happened to sell each year, not a like-for-like index - a shift in the mix of homes sold moves the line as surely as a shift in prices.

Indexed view

Each point is the year's median (or mean) sold price divided by the base year's figure, so 200 means double the base-year value in cash terms. The base year is the first year with at least 10 recorded sales - re-basing on a thinner year would make every later reading a multiple of noise. Indexing keeps a long series readable along its whole length - a pound axis hides the early years behind late-year scale. Values are nominal: not adjusted for general inflation.

The instant estimate

This is the method behind the instant estimate on street pages, where you can add a house number. The valuation page states its own method.

An automated estimate, not a formal valuation.

Sales of the same type nearby

We take the 20 most recent HM Land Registry standard sales of your property type in your postcode - or in its postcode sector, then its district, if the postcode has none of that type. We use those from the last 5 years if there are at least three, otherwise the last 10, then 20 years, then all of them; restate each in today's money with the UK House Price Index; and take the median. We then scale that by a fixed bedroom factor of our own (x0.65 for 1 bedroom up to x1.5 for 5 or more). Land Registry does not record bedrooms, so this factor has not been tested.

Too few sales: the UK House Price Index

If fewer than three sales of your type are found, we use the House Price Index average price for your type in your local authority instead - that figure takes no account of bedrooms. We show no estimate at all where there is no sale of your type on record in your postcode, its sector or its district, or where that average would have to be the UK-wide figure because the index cannot place your postcode in a local authority.

Corrections by property type

The figure from either route is then corrected for a bias we measured by property type against real sales: reduced by 2% for detached homes and reduced by 10% for flats. Other types carry no correction.

The home's own earlier sale

If you give a house number or name and HM Land Registry's most recent standard sale of that home was between 1 and 5 years ago, and the house number picks out one home in that postcode, we start from that sale, restated in today's money, instead - with no bedroom factor or type correction - unless it is less than 0.5 times or more than 2 times the median of the same-type sales we found. Where fewer than 3 such sales were found to check it against, we use it unchecked and widen the range. A sale between 1 and 2 years ago is used with a wider range than an older one: anchors that young measured less reliable in our back-test. A most recent sale under 1 year old is not used at all: estimates anchored on a sale that recent measured less reliable than the comparable-sales method, so the estimate is made without it (we do not look back for an older sale of the home).

Energy rating

We make no adjustment for energy rating: when we tested a fixed one against real sales it made no clear difference to accuracy.

The range

We show a range rather than a single figure, rounded outward to the nearest £5,000. A local agent's valuation will be more accurate.

What the estimate is based on

Under each estimate, a line says which sales, or which price index, it rests on. The range is never narrower than the typical error we measured when we tested the way this estimate was made against real sales: for sales of your type in your postcode, the error for that property type; where the postcode has only a few or only old sales of your type, or none (so nearby postcodes were used), the wider error we measured for those cases across all types; for a home's own earlier sale, the error measured for that method; for the House Price Index fallback, its own figure, per property type where that is wider. It can be wider still when the estimate rests on few or old sales, on sales we could not restate in today's money, or without a local price index. We do not show a confidence rating: when we tested one, estimates it rated more confident were not reliably more accurate.