Published March 1, 2026
| Version v1
Journal article
Open
A Graph-Theoretic Alternative to Forced Ranking
Description
Introduces sparse cross-team comparison using Bradley-Terry-Luce / Elo / TrueSkill pairwise models. Adding 1–2 role-matched guest comparisons per team per cycle halves misclassification to ~15.5%, approaching the information-theoretic floor.