Store the retrieval date with every height and weight. Take the roster from one tournament’s official banzuke, retrieve profiles in one documented window, normalize centimetres and kilograms, and describe the figures as “official profile values” unless the Association identifies the measurement date. They are not measurements taken on the day you publish.
Weight can change during a career. Placing a current profile value beside a result from years ago silently assigns today’s body to the earlier wrestler. Unless a dated historical measurement is available, limit the analysis to a current cross-section instead of reconstructing a false timeline.
Keep an auditable row
Each row should contain a stable wrestler identifier, current shikona, rank, height, weight, source URL and retrieval timestamp. Preserve the precision published by the source and round only for display. A renamed wrestler remains one person; name history should not create duplicate observations.
Alongside the mean, report the median, interquartile range, minimum, maximum, sample size and missing count. State whether the sample is makuuchi, all sekitori or every ranked wrestler, and whether ranked but absent wrestlers remain included. Those choices often change the result more than a decimal place does.
BMI is especially easy to over-interpret. Do not add it merely because height and weight are available. If a research question genuinely needs a ratio, explain its limitation in an athletic population and avoid medical conclusions. Freeze the population by banzuke, division and extraction date; count a missing height or weight explicitly. A one-date profile snapshot is not itself right-censored data. Define right-censoring only in a separate longitudinal time-to-change study whose observation ends before an event, and never interpolate unrecorded past or future bodies.
Body size does not explain a bout by itself
A chart designed to prove that heavier wrestlers win more will miss rank, experience, opposition and absence. If you compare body data with results, use the same tournament and bout population, publish the denominator, and call an association an association rather than a cause. Even labels such as “small wrestler” or “very heavy wrestler” require a declared threshold that cannot move to fit the story.
Check graph axes as carefully as arithmetic. A truncated scale can make a modest difference look dramatic. Never estimate weight from a photograph or infer injury, fitness or motivation from a profile change.
Return to movement
After reading the table, watch one bout without letting the numbers write its ending. Notice a large wrestler making several small foot adjustments, or a lighter wrestler shifting an opponent away from the centre line. Body data can point you toward an interesting contrast; it should not reduce a wrestler to a silhouette or a predicted result.
Freeze the Question Before Reading the Result
The research question is whether the defined metric differs under the article’s stated conditions. “There should be a difference” is a hypothesis, not a conclusion entered before counting. Preserve a contrary result, and treat a small difference, heavy missingness or incomparable conditions as grounds to hold the conclusion. Recalculate mean and median, inclusion and exclusion of absent wrestlers, and complete-case sensitivity from the same profile rows as falsification checks. If direction changes, the alternative conclusion is snapshot- or missingness-sensitive.
Sources
- Japan Sumo Association, wrestler search and banzuke table (checked 3 September 2026; retain retrieval dates and tournament scope).
Source links used for verification
Each link was used for the corresponding claim in this article; inclusion is not a blanket endorsement of the whole page.
Article verified:
- Japan Sumo Association wrestler searchPrimary source (as checked for this article)
- Japan Sumo Association banzuke tablePrimary source (as checked for this article)