Introducing Streaks: patterns that shouldn't exist

Football results are noisy. On any given Saturday, the better team loses, the favourite draws, the goalkeeper has a nightmare. The randomness is half the point. But look at enough data and something strange emerges: patterns that are far too consistent to be accidental.

We call these streaks.

Not simple winning runs — those are easy to find and largely uninteresting. We're looking for conditional chains: "whenever X happened, Y followed — and it held for N matches in a row". The longer N is and the more improbable Y is given X, the stranger the finding.

What counts as a streak

A streak has two parts: a when condition and a then outcome. The when defines which matches are even eligible — it might be "when Manchester City played at home" or "when a specific player started". The then is what happened in every one of those eligible matches: "they didn't lose", "they scored two or more", "the game had over three goals".

The streak count is how many consecutive eligible matches the outcome held. Gaps — matches where the when condition didn't apply — don't break it. A home streak only counts home matches.

Why this is interesting

A team winning 10 home games in a row is unremarkable. But a team that hasn't lost a home match in December in 18 years? That's the kind of thing that earns a raised eyebrow over a half-time pint. These streaks surface quirks in how teams play in specific conditions — patterns that a simple table never reveals.

How we score them

Raw length isn't enough. A streak of 50 matches in a trivial condition ("scored at least 0 goals") is useless. We score each streak using an estimate of how improbable it is: the rarer the outcome and the longer the run, the higher the score. Only the highest-scoring streaks make it to the page — and they're filtered to only show ones relevant to upcoming fixtures.

Head to the Streaks page to see what's live right now.