A Player Can Shoot Better From Everywhere and Worse Overall
Published 11/17/2026 · 3 min read · Sport calculators
Split a hundred shots two ways and the totals can reverse the parts. Player A takes 10 shots close, hitting 9 for 90%, and 90 from distance, hitting 20 for 22.2%; the overall figure is 29 out of 100. Player B takes 60 close, hitting 45 for 75%, and 40 from distance, hitting 8 for 20%; the overall is 53 out of 100. A is the better shooter in both zones and the worse shooter overall, because A took nine tenths of the shots from the hard place and B took most from the easy one. Nothing is wrong with any of the six percentages.
A hits 90% up close and 22.2% from range; B hits 75% and 20%. A is better at both and finishes on 29% against B's 53%, because of where they shot from.
The reversal has a name and a long record
It is Simpson's paradox, and it turns up wherever a rate is averaged across groups of unequal size. A university can admit a higher share of women in every department and a lower share overall; a treatment can help in both age groups and appear to hurt in the combined figure; two batting averages can each be higher in both halves of a season and lower for the year. The pooled number is not lying — it answers a different question, namely what happened overall, and the answer genuinely depends on where the attempts were made.
Which number should the coach use
Both, for different decisions. To judge shooting ability, the split figures are the honest ones, because they compare like with like. To predict points from the next hundred attempts under the same shot selection, the pooled figure is the right one, since it already includes where the shots come from. The mistake is using one where the other belongs — praising A's accuracy and then projecting A's total, or vice versa. Whenever a rate is compared across people, the first question is whether they faced the same mix of situations.
| Player | Close / distance | Overall |
|---|---|---|
| A | 9/10 and 20/90 | 29% |
| B | 45/60 and 8/40 | 53% |
| Better? | A in both | B |
Worked with our own calculator
Basketball shooting percentage calculator
Given
- Shots made
- 8
- Shots attempted
- 20
Result
- Shooting percentage
- 40%
These figures are produced by the calculator below, not typed in by hand — they are recomputed whenever the tool changes.
Run it on your own figures →Frequently asked questions
- Is this common or a constructed curiosity?
- Common enough that medical trials are designed to prevent it. Randomisation exists precisely to make the group mixes comparable, and observational studies that cannot randomise report results stratified by the variables that might otherwise do this. The example above is extreme to make it visible, but mild versions appear whenever attempt counts differ between the people being compared, which in sport is almost always.
- How do I spot it in a table?
- Look at the denominators, not the percentages. If one row has ten attempts in a category where another has sixty, the pooled comparison is already suspect, and the direction of the distortion follows the direction of the imbalance. Any table of rates that hides its counts is unreadable for this purpose, which is a good argument for always printing both — a percentage without its denominator is half a statistic.
Articles you may find interesting
All guides →Related tools
Sources
Spotted a mistake in this article?