Why baseball is a lesson in ratios
It’s pennant-race season, and box scores are everywhere. Fans trade numbers like .300, .377 and 3.00 without a second thought. Behind each one sits a middle-school idea, the ratio, plus a statistical trap that fools even seasoned analysts. All players and numbers below are hypothetical, made up for the example.
The concept: a rate is outcomes over opportunities
A ratio compares two quantities by dividing them. When the denominator counts opportunities, the ratio becomes a rate: the share of chances that worked out. Batting average is the classic one, with for hits and for official at-bats.
Baseball drops the leading zero and shows three decimals, so 0.300 is written .300 and read “three hundred.” Mathematically it’s just 30%: three hits every ten chances. That sounds low, but in baseball it’s excellent, which is a reminder that a rate only means something next to other rates from the same context.
On-base percentage (OBP) widens both parts of the fraction. It also counts walks () and hit-by-pitches (), and adds sacrifice flies () to the denominator.
ERA is a proportion in disguise
For pitchers, the classic stat is ERA: earned runs () allowed per 9 innings, the length of a full game. If a pitcher threw innings, set up the proportion: is to as ERA is to 9.
One notation catch: partial innings are written 6.1 or 6.2, meaning 6 innings plus 1 or 2 outs. With three outs per inning, 6.1 is , not 6.1. Typing 6.1 into a calculator gives the wrong ERA.
Worked example, step by step
- A hitter has 150 hits in 500 at-bats. Divide.
- Same hitter: 60 walks, 5 hit-by-pitches, 5 sacrifice flies. Add them to the right places.
- A pitcher allows 20 earned runs in 60 innings. Apply the proportion.
- Another pitcher allows 5 earned runs in 6.1 innings. Convert 6.1 to 6 1/3 first.
- Check: .300 × 500 = 150 hits; 3.00 × 60 ÷ 9 = 20 runs; 7.11 × 19/3 ÷ 9 ≈ 5 runs. Everything maps back to the original data.
Simpson's paradox: when averaging averages lies
Now the trap. Take two hypothetical hitters, Ana and Ben, across two halves of a season. First half: Ana goes 4 for 10 (.400), Ben 35 for 100 (.350). Second half: Ana goes 25 for 100 (.250), Ben 2 for 10 (.200). Ana wins both halves.
Average the two averages and Ana gets .325 to Ben’s .275, which seems to settle it. But add up hits and at-bats: Ana is 29 for 110, about .264, and Ben is 37 for 110, about .336. Over the full season, Ben is better. That’s Simpson’s paradox: a trend inside every group reverses when the groups are combined.
The culprit is weighting. Most of Ana’s at-bats came in her weak half; most of Ben’s came in his strong half. A plain average of averages treats 10 chances and 100 chances as equal, which they aren’t. The correct weighted average is simply the ratio of the totals.
Common mistakes
- Averaging averages with different denominators. Go back to the totals.
- Reading 6.1 innings as 6.1. It’s 6 and one-third.
- Trusting tiny samples. A .500 average in 4 at-bats tells you almost nothing.
- Forgetting .300 means 0.300, which is 30%, not 300 or 3%.
- Comparing across denominators: OBP divides by more than AVG does, so the two aren’t directly comparable.
Frequently asked questions
Is batting average a percentage?
It's a ratio between 0 and 1. Multiply by 100 to get a percentage: .300 means a hit in 30% of at-bats.
Why does ERA use 9 innings?
A regulation game is 9 innings. Scaling to 9 lets you compare pitchers with very different workloads, exactly like a proportion problem.
Is Simpson's paradox a math error?
No. Every calculation is correct; the mistake is treating group averages of different sizes as if they carried equal weight.
Does this show up outside sports?
Yes. Pass rates by class, treatment success by hospital and sales conversion by channel all fall into the same trap when group sizes differ.