Model rankings

Loading ratings…

#	Fighter	Score	Record	Class	Form	Win%	Last 5	Last fight

Fight history

These rankings come from a predictive model, not Elo. Every rated fight since 2011 is fitted at once: each fighter gets a latent strength (their demonstrated class), and a pre-fight statistical profile — striking and takedown differentials, control, knockdowns, finishing, recent form, layoff, age past 33, reach — explains the rest, so losing to elite competition is priced by what those opponents are actually worth. Statistics are opponent-adjusted and weighted by cage time and recency. The score is class plus half of current form; strength itself never decays, so a championship run stays on a fighter’s record even through a losing streak. Round Robin instead matches everyone in a division against everyone else through the model and ranks by average predicted win chance. Tags mark the top or bottom quarter of a division on a statistical dimension. Doctor stoppages from cuts count as wins; freak-injury and foul endings are on the record but not rated, and fighters inactive for over two years are not ranked. In walk-forward tests (both fighters with at least three UFC fights), this model called 66% of 2025–26 fights correctly; the old Elo called 58%.