This prediction rests on one fighter only. We have no UFC record for José Henrique Souza — no bouts, no style profile, no age and no reach — so the model is comparing Jonny Parsons against the welterweight prior, and Parsons is the pick at 67.9%, against 32.1% for Souza. The largest factor is the rating difference, which is a comparison against a default rather than against a measured opponent. The second and third factors are the two layoff terms, and they contradict each other: Parsons has not fought since July 2023, 39.1 months, which the model penalises through one term and credits through the other. Parsons's own profile is close to the welterweight average and built on few fights, with an uncertainty of 283 rating points. Treat the 67.0% as the division prior plus a long layoff, not as a read on this fight.

Parsons is an average welterweight on the rates we can measure, and the uncertainties on most of them are large. He lands 49.5 significant strikes per 15 minutes against the 48.3 division average — with an uncertainty of 37.1, which is three quarters of the value — and absorbs 55.0 against 54.6. His 52% striking defence and 43% accuracy are both at the norm. On the mat he attempts 4.1 takedowns per 15 minutes against a 4.2 average, holds 187.2 seconds of control against 189.1, and concedes 239.2 against 247.0. The two numbers that lean his way are a 59% takedown defence just above the 58% average and 12.6 clinch strikes per 15 minutes near the division's 13.0. His last fight was a second-round knockout of Danny Roberts in July 2023. For José Henrique Souza there is nothing to report. The database holds no completed UFC bout for him, so there is no style profile, no rating and no age, and nothing about his striking, grappling or pace can be stated. This analysis will not guess at it. What remains is a fight the model should not be read as having an opinion about. The 67.9% is built on Parsons having a measurable record at all, and the two layoff terms pointing in opposite directions is a sign the model is uncomfortable with 39.1 months away — the longest layoff on this card by some distance, and long enough that his profile describes a fighter from three years ago. The one concrete thing in his numbers is that they are unremarkable: no rate in his profile is more than a strike or two from the welterweight average, and the only finish on his recent record is the knockout that preceded the layoff. The model gives 39.9% for a knockout and 43.5% for a decision, which with one side of the fight unmeasured is the division's shape rather than this bout's. A reader should take the number as near-uninformative and wait for the first exchange.