How the model reads him
Roberto's rating updates after every result and weights the quality of the opponent. Against a specific opponent it becomes a win probability, adjusted for the stylistic matchup — striking output against absorption, takedown rate against takedown defense — plus age and layoff. The weighting is ours; the output is published on the fight board and graded.
What we'd flag
The tape starts in 2015 and covers UFC and Contender Series only, so career records elsewhere aren't reflected. Beyond that: MMA has the highest single-event variance of anything we model, and no rating survives contact with a flush shot. For Roberto specifically, Roberto has only 2 fights on our tape, so every rate here is built on very little.