Why sample size matters in sports statistics

A hitter goes six for ten in a short stretch. The 60% jumps off the page. Another sample shows 60 hits in 100 at-bats, also 60%. The percentages match, but the amount of evidence doesn't.

Sample size means how many observations a statistic uses. Here the observations are at-bats. In other reports they might be shots, swings, carries or games. Read that count beside the result rather than assuming all percentages carry the same weight.

A short stretch can be worth exploring. It might show a new approach or a useful role change. It also leaves more room for a handful of results to dominate what you see. You need the count to tell those possibilities apart.

Add one hit to the six-for-ten sample. It becomes seven hits in eleven at-bats: about 63.64%. That's a jump of roughly 3.64 percentage points from one result.

Add one hit to the 60-for-100 sample and it becomes 61 in 101, about 60.40%. The same extra result moves the longer record by only about 0.40 percentage points.

This doesn't make the longer record a prediction. It simply shows why its percentage is less sensitive to the next observation. Be especially careful with a leaderboard where someone with a few chances sits beside someone with hundreds.

A full season gives you more observations, but it can include an old role, a former team or a different opponent mix. If a player just became a starter, a long average that combines bench and starting appearances may miss something useful.

Keep the longer record as context, then describe the recent change separately. You can say a rate rose after a role change without claiming the new rate is already the player's permanent level.

Splitting the data creates another tradeoff. A player's overall sample might be large, while their results against one pitch type are based on only a few balls. That slice answers a more specific question with less evidence.

Write down the result, the count and the dates. “Six hits in ten at-bats across three appearances” gives someone far more information than “batting .600 lately.” Then compare it with a relevant longer record.

Look for a reason beyond the result itself: more playing time, a different pitch mix, or a visible change in contact. Those details help you understand the stretch, but they don't turn ten at-bats into a hundred.

There isn't one magic sample cutoff that settles every metric. A rate built from rare events and a rate built from frequent opportunities can behave differently. The practical habit is simpler: whenever the number looks dramatic, ask how many observations made it.