Measurement 6 of 9Research preview

Chance check

Any player has some kills that look unusual. The chance check asks whether a player has more than chance predicts.

Is this more unusual than chance?

One hundred kills, five highlighted
100 ordinary killsAbout 5 land in the most extreme 5% of any one measurement. Illustration.

Reading the count

Five in a hundred is ordinary. Ten in a hundred is twice what chance predicts. The check shows both numbers side by side and leaves the judgment to a person.

A ratio is only shown once enough kills are measured, because small samples swing wildly.

What it is

By definition, about 5 in every 100 ordinary kills fall in the most extreme 5% of any one measurement. So a player with 100 measured kills should have around 5 in that band for each measurement. The chance check counts how many of a player's kills are in the band and sets that next to the number expected.

Why it matters

A list of unusual kills sounds alarming even for a normal player. The expected count turns it into something a reader can judge: five of a hundred is ordinary, ten times that is not.

How Overshoot keeps it honest

A ratio is only shown once enough kills are measured, because small samples swing wildly. Kills from the same player and match are not independent, so the count is a guide for a person to read, not a statistical verdict.

Where it is used

This check is in the research fact sheet today. It is not in the app yet.

Keep in mind

What this cannot tell you

  • Kills from one player are not independent, so odds quoted from chance alone overstate certainty.
  • It says how much unusual there is, not what caused it.
  • It depends on the quality of the ordinary-player reference.