Sample Size
Sample size is the count of underlying events behind a statistic, and it determines the width of the uncertainty band around the figure.
What it means
Football's event rates are low, so most metrics need many matches to stabilise, and rate measures built from rare events need longer still.
Why it matters
It is the quiet variable behind almost every disagreement in analytics, and quoting a figure without it makes the claim unfalsifiable.
Where it is used
Reading any rate, conversion figure or defensive metric, and deciding whether a difference between two values is meaningful.
Where it misleads
Stabilisation is not correctness: a metric can be stable and still biased. A stable mean can also conceal a heavy tail of very poor performances.
Key points
- Volumes stabilise first, value measures next, rates last.
- Conversion needs multiple seasons to say anything.
- Name the sample in the sentence or the claim cannot be checked.
- A stable mean can hide a heavy tail; check the spread.
- Marking a figure as unstable is a valid conclusion.
- Differences of a few per cent are usually inside the noise.
Nothing here is a forecast; the number describes play that has already happened and the limits above describe what it cannot reach.