Method
The sources, the formula, and what does not get published.
Where the numbers come from
Two sources, both open. Wikidata (CC0) supplies the roster — who is a boxer, who is an MMA fighter, and how many language editions of Wikipedia carry an article about them, which is used as a rough notability signal. English Wikipedia (CC BY-SA 4.0) supplies every measurement and record, parsed from the article's own infobox.
Wikidata is not used for the measurements themselves, because it cannot be: it has no arm-span property at all, and competition class is filled in for a few dozen fighters out of twenty-odd thousand. That is why this site reads infoboxes.
Deliberately not used: BoxRec, UFCStats, Sherdog and Tapology. Those are proprietary databases whose terms do not permit republication, and routing them through a third-party API does not change that.
Reach advantage
The one number this site is built on:
reach advantage = (reach − height) ÷ height × 100
A fighter whose reach equals their height scores 0%. Positive means the arms are long for the frame. Because it is a ratio it is comparable between a strawweight and a heavyweight, which a raw reach figure in inches is not.
What gets excluded, and why
Of 8,656 candidates, 2,515 were published and 6,141 were not. The reasons:
| Reason | Fighters |
|---|---|
| no weight class | 2,424 |
| no reach — primary metric not computable | 2,170 |
| no usable record | 784 |
| no height | 501 |
| career too short (<6 bouts) | 222 |
| cohort too small for percentiles (n<30) | 40 |
The largest single reason is a missing reach figure. A page without one cannot carry the metric this site exists to show, so it is not built — an empty template with ads on it is not a page, it is a liability.
A division needs 30 fighters before any percentile is quoted from it. A percentile against a dozen people is a number with no meaning, and publishing it would undermine every honest figure on the site.
Known limitations
- Reach is not measured to any standard. Promotions self-report it, fighters round it, and Wikipedia inherits whatever was published. Treat differences under an inch as noise.
- The dataset skews toward fighters notable enough for an English Wikipedia article, which means it under-represents regional and early-career competitors.
- Percentiles describe this dataset, not the sport. They are a fair comparison between the fighters here and nothing more.
- No page predicts the outcome of a fight. The comparison pages compare measurements.