Methodology
A deliberately conservative method: preserve what the source can support, suppress what it cannot, and separate descriptive transaction evidence from modelled house-price movement.
1. Source
The production input is HM Land Registry Price Paid Data for England and Wales. Current v2 feeds use standard Category A transactions from corrected yearly source files. We record source URLs, reference period, retrieval evidence and checksums before a release can pass validation.
2. Geography
Public exploration is by outward postcode district. A postcode district is not a town, local authority or housing-market boundary, so comparisons are labelled as geographic summaries rather than market definitions.
3. Headline periods
Headline medians and volumes use a 12-month cohort. The two latest transfer months are withheld from headline comparisons because registrations arrive with a lag. Older records can still be revised later.
4. Small samples
Median prices and percentage changes are withheld below the project threshold. This is a reporting safeguard, not a declaration of statistical significance or a guarantee of representativeness.
5. What a median means
A median sold price describes the middle observed transaction in the cohort. It is not a valuation and it is not quality adjusted. A changing mix of detached homes, flats, new builds or locations can move the median even when like-for-like prices do not.
6. Validation
Generated feeds are checked for schema, arithmetic, inventory, source period, SHA-256 integrity and publication boundaries. Five geographically diverse calibration districts are independently recalculated from raw source files before a staging release is considered valid.
Research boundary: the official UK House Price Index uses hedonic regression and property attributes to control for changing mix. SCC Land Registry will evaluate quality-adjusted and repeat-sales methods as separate research products; they will not be silently mixed into the descriptive postcode medians.
Planned methodological extensions
Historical backfill to 1995; inflation-adjusted illustrations with a declared deflator; repeat-sales cohorts; hedonic/mix-adjusted research; revision-impact tracking; uncertainty and thin-market diagnostics; and reproducible release-to-release change logs.