Home › The market › Curiosities

Things this data says that nobody set out to look for

Every figure on this page is computed from 31,324,023 recorded sales between 01/01/1995 and 30/06/2026, at build time, from the register itself. Each finding is shown with its control and with the thing that would make it wrong — because in 31 million rows, most patterns that look like a discovery turn out to be the calendar.

1. Almost half of England and Wales moves house on a Friday

What would make this wrong: nothing much — this is a completion date, and completions happen on working days when banks transfer money. The finding is the size of it: 48.8% on one weekday, and 0.3% across both weekend days combined.

2. Nobody wants to complete on the 13th

Corrected for the fact that months do not all have 29, 30 or 31 days, the 13th of the month receives 71.1% of an average day's completions. Its immediate neighbours are ordinary: the 12th gets 102% and the 14th 97.3%.

And on Friday the 13th specifically: 64.8% of an average Friday.

What would make this wrong: if the 13th were simply rarer, or if the dip were part of a slope across the month. Neither holds — the 13th occurs in every month, it is counted per occurrence, and both neighbours sit at roughly 100%. This is a real avoidance, and the only plausible explanation is superstition.

3. The month-end rush

What would make this wrong: counting raw totals instead of per occurrence. The 31st exists in only seven months a year, so a raw count would show it as one of the quietest days rather than the busiest. Correcting for that inverts the finding entirely.

4. Summer really does cost more — about 7%

What would make this wrong: comparing raw prices across years — thirty-one years of inflation would swamp any seasonal effect and produce whatever pattern the sample happened to hold. Every sale here is measured against its own year's median, so the only thing left is the season.

5. New-build sales have collapsed as a share of the market

What would make this wrong: the most recent year is always incomplete — sales reach the register months after completion, and new builds are registered more slowly than second-hand ones, so the final year understates itself. Read the trend, not the last row.