Method

Regression to the mean: why runs don't last

Why teams "in form" tend to slow down and how to spot it early in the numbers.

Scout Specialist editorial team · · 5 min read

One team wins six in a row and looks unbeatable. Another loses four times and looks in crisis. Often, in the following weeks, both drift back to their normal level. It isn't bad luck or magic: it's regression to the mean.

What it is

Every result is a mix of quality and chance. In extreme runs, good or bad, chance has usually helped in one direction. When chance goes back to neutral, results move back towards real quality.

The small-sample problem

Ten matches sound like a lot, but statistically they're few. With a small sample percentages are unstable: an 80% Over across 5 home matches fits an "Over team" but also a normal team that's had a couple of crazy games.

How to spot it early

  • Goals − xG: if a team scores far more than its expected goals, the winning run is probably inflated.
  • Extreme percentages: values close to 0% or 100% over few matches almost always come back down.
  • xG difference: it's more stable than results and tells you early who is really playing well.
In practiceWhen a number looks too good to be true, check how many matches it's based on and what the xG says.
This data for today's matchesStatistics, probabilities and the news Radar are in Scout Specialist Pro.
Subscribe to Pro