A birthday must not rewrite a wrestler’s age in every historical tournament. Preserve person ID, official birth date, tournament, fixed reference date, calculated age and ranked population. Treat a new banzuke, an official profile correction and a methodological change as three different events.
The current age on a profile is useful today but wrong for a basho years earlier. Historical comparisons recalculate from the birth date and the reference date chosen for each tournament. The calculation can update without destroying the source value or the prior release.
Fields in the history
Store person ID, historical shikona, official birth date, tournament ID, banzuke announcement or opening date, unrounded day difference, completed years, retrieval time, official URL and dataset version. A name change modifies the display history, not the identity count. Freeze the population by banzuke, division and reference date; count an officially unverified birth date as missing. Future ages in an active career are right-censored, and an unpublished next-tournament rank is never filled in by prediction.
Mark an unverified birth date as missing. If an official source later supplies it, preserve the former missing state and add the retrieval date rather than pretending the value was always available.
Routine release, source correction and definition change
A new banzuke changes the measured roster and is a routine release. A revised birth date on an official profile is a source correction. Switching the study’s reference date from announcement day to opening day is a definition change. The last case requires a new version and full-period recalculation while the previous table remains identifiable.
When an average changes, decompose the cause: birthdays, promotions and demotions, retirements, cohort changes, newly resolved missing values or corrections. Compare the median, complete cases and a bounded missing-data sensitivity run as alternative specifications, and test whether the interpretation reverses. A display-rounding change is recorded separately if the full-precision values did not change.
Current public log baseline
- 3 September 2026, v1.0: fixed one reference date per tournament and calculate age from the official birth date at day precision.
- 3 September 2026, v1.0: publish mean, median, cohort size and missing count together.
- Next checkpoint: after the next banzuke announcement, version the roster and publish the difference from the former release.
A numerical release must include its source banzuke, extraction timestamp, code version and file hash. Copy edits do not create a data version. A correction to one birth date does if it changes a table; the affected figures and conclusions are listed explicitly. Stop release when deterministic recalculation fails or an alternative specification reverses the conclusion; ‘no established age difference’ is a valid result.
A correction brings one person back into view
A one-year difference may barely move a mean, but it matters to the history of the wrestler concerned. After fixing the table, return to the next bout and look past the “young” or “veteran” label. Better versioning makes the person outside the category easier to see.
Sources
- Japan Sumo Association wrestler search, banzuke table, and annual schedule (checked 3 September 2026; refresh after each banzuke announcement).
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)
- Japan Sumo Association annual schedulePrimary source (as checked for this article)