Reading local price trends without being misled

Local price statistics are irresistible and treacherous. The smaller the area, the more meaningful the number feels and the less meaningful it actually is. A postcode median can rise twelve per cent in a quarter because two large detached houses happened to complete, while every home in the area is worth exactly what it was worth three months earlier.

Why a local average moves when nothing happened

The culprit is the mix effect. A price average describes the homes that sold, not the homes that exist. Change the composition of what sold and the average moves on its own, with no change whatsoever in underlying value.

Picture a road with thirty terraced houses and six large semis. In one quarter, four terraces sell. In the next, two semis sell. The median leaps, a local news headline writes itself, and the correct interpretation is that a different kind of house sold. The same mechanism operates on a much larger scale: a new development of flats completing in one month can drag a whole ward median downwards while pushing the transaction count up.

Medians, means and the four-sale month

• Prefer the median to the mean at local level. One unusual sale moves a mean substantially and a median barely at all.
• Always publish the transaction count next to the price. A median over four sales is not a statistic, it is four anecdotes wearing a hat.
• Set a minimum sample before you draw any conclusion. Fewer than about ten comparable transactions in the window and you should be describing rather than concluding.
• Look at the spread, not only the middle. A stable median hiding a widening range usually means the top and bottom of the market are separating.
• Never quote a percentage change on a small sample without saying how many sales sit behind it.

The provisional tail at the right hand edge

Sold price records enter the file only after registration, which typically trails completion by two weeks to two months and considerably longer for new builds and first registrations. Every chart built from this data therefore falls away at its most recent end, because those months are still filling up. Read that decline as a market signal and you will be wrong roughly every month, forever.

The remedy is mechanical: treat the two most recent months as incomplete, and either exclude them or shade them differently. The same applies to comparisons. A year to date figure compared against a full previous year is not a comparison at all. The detail sits in how long a sale takes to appear on the register.

Seasonality and the transaction count

Housing activity is seasonal. Completions cluster and thin at predictable points in the year, and small areas feel that far more sharply than national figures do, because a quiet month locally can mean two sales instead of six. Comparing this quarter with the last quarter mixes seasonality into whatever else is happening.

Compare the same period year on year instead, and read the transaction count as a series in its own right. Volume often turns before price does: a market where sales are falling while prices hold is telling you something that the price line alone conceals.

Index or raw records: pick the right tool

The UK House Price Index reports are the standard reference for direction, published monthly and adjusted so that changing composition does not masquerade as price movement. The Office for National Statistics housing pages add context on tenure, affordability and rents. Use those for the trend, and use the sold price records for the level of an individual street.

A checklist before you believe a trend

1. How many transactions are behind the number? Below ten in the window, be descriptive rather than conclusive.
2. Has the mix changed? Check property type, tenure and the new-build flag across both periods.
3. Are the most recent two months included? If so, exclude them and recalculate.
4. Is the comparison like for like in season, meaning the same months a year apart?
5. Does a mix-adjusted index for the wider area point the same way? If it flatly contradicts your local figure, the local figure is probably composition.
6. Would the conclusion survive removing the single highest and single lowest sale? If not, it was never a trend.

Applied honestly, that checklist kills most local trend claims, including some you would rather keep. What survives it is worth acting on, and it feeds straight into the comparable method described in valuing a home from comparable sales.

Sources: transaction level data is HM Land Registry Price Paid Data for England and Wales, available from the downloads page and updated monthly around the 20th. Contains HM Land Registry data © Crown copyright and database right 2026.