StatLab
← Level 2: Luck, context, and the 100 scale

Sample size: when does hot become real?

After how many at-bats does a hot streak stop being noise?

Every stat has a stabilization point — roughly, the sample where a player's number starts reflecting more skill than noise. They differ wildly, and knowing the rough order is a superpower for reading April box scores.

The rough ladder

Fast: swing rates and K% (a few weeks — ~60 PA). Medium: BB%, ISO (a couple months — 120–160 PA). Slow: AVG and BABIP (a full season plus — 800+ balls in play). Exit velocity (Level 4) stabilizes almost immediately, which is a big reason Statcast changed scouting.

Practical rule: in small samples, trust the fast-stabilizing stats and the underlying process (how hard is he hitting it? is he chasing?), not the results. A .350 April AVG with a 30% K rate is a warning, not a breakout.

How this stat lies to you

  • 'Stabilizes' doesn't mean 'is now true talent' — it means signal finally outweighs noise. Regression still applies.
  • Stabilization points are averages across players, not laws for one player.
  • Even a full season of some stats (pitcher BABIP, clutch numbers) remains mostly noise.

Check yourself

1. Which April stat should you trust most, 80 PA in?

2. A hitter is at .350 after three weeks, but with career-worst chase and whiff rates. Verdict?