RESEARCH NOTE

Bias Traps in Comparing East and West on the Sumo Banzuke

Last verified: 3 Sep 2026Linked sources support only the corresponding claims in this article. Times, prices and venue operations can change, so check the current linked guidance before travelling.
A miniature arrangement of reference materials with matching conditions.

Compare east and west only after matching the same division, rank and numbered position. Do not pool East Maegashira 1 with the entire west side, and do not mistake a wrestler’s left or right position on a results screen for official banzuke side.

The official banzuke is primary evidence for rank, number and side. East is placed above west at the same listed rank and number, but that ordering rule does not prove that the word “east” itself causes a wrestler to win more bouts.

Rank and number are major confounders

A single east-versus-west average from yokozuna through lower maegashira can reflect unequal rank composition or a vacancy. Build matched pairs within the same division, rank and number. If one side has no partner, keep the row as unmatched rather than borrowing a wrestler from another rank.

Extra positions, unusual rank counts and vacancies need explicit notes. Join daily results to the tournament’s banzuke by stable wrestler identity, not name alone. Name changes, spelling variants and repeated names can otherwise create false switches.

The visual left side of a website or broadcast is not a reliable east-west field. The side used in the analysis must come from the official banzuke for that basho.

Opponents and tournament selection still matter

East wrestlers do not always face west wrestlers in a neat paired experiment. Retain every opponent, opponent rank and pre-bout record. Publish the raw side-level rate, the within-pair difference and any opponent-stratified result together rather than choosing whichever looks most dramatic.

One tournament can produce a large random gap. Set the interval before looking at results and retain withdrawals, defaults, promotions, demotions and unmatched slots. If adding older tournaments removes an apparent difference, that disappearance is a useful result, not a reason to hide the update. Freeze the population by division, rank, number, interval and extraction date; count unmatched or unjoinable rows as missing. Right-censor tournament follow-up at the latest officially completed day, while treating later individual bouts as not yet contested rather than missing outcomes; never impute them as wins or losses.

Fields readers need beside the chart

  • tournament, division, rank, number and official banzuke side;
  • pair status, vacancy, extra position, withdrawal and default flags;
  • opponent and opponent’s banzuke position;
  • counts, rates, differences, interval and extraction version.

Never turn banzuke east and west into claims about personality, birthplace or fighting stance. These are different concepts that happen to share directional words.

Start with a hypothesis that permits no difference

Ask whether a difference remains in official outcomes after building exact rank-and-number pairs within a declared interval. “East is stronger” is not the starting conclusion; no difference must be an equally reportable result. Test sensitivity by excluding unmatched slots, separating withdrawal and defaults, stratifying opponent rank and shifting the interval one basho at a time.

If the gap disappears or reverses, report a composition-sensitive finding rather than a directional effect. The official banzuke establishes position, but it cannot by itself prove that position caused a result. When a new banzuke weakens an earlier pattern, preserve the old chart and let the update show how the hypothesis changed. Preserve raw-table hashes, join rows, exclusion reasons and code version. Recalculate alternative pair rules, default handling and shifted intervals as falsification checks; if the gap is unstable, the alternative conclusion is that no east-west label effect has been established.

A small way to make the banzuke come alive

Find one east-west pair at the same number and follow their opponents across fifteen days. They may finish with identical records after taking very different routes. The most interesting part of the comparison is often not which side “won,” but how two neighbouring names travelled through the basho.

Sources

  • Japan Sumo Association banzuke, daily bouts/results and tournament scorecard (checked 3 September 2026).

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: