UFC Fight Night: Usman vs Du Plessis

This Saturday night brings us a main event featuring two former champions looking to get back in the saddle. In this post we're going to break down the main card. We'll take a look at some interesting datapoints, make some predictions, and let you know what the AI is thinking.

As always, if you're interested in getting this kind of data (and more) for every fight, consider a Pro subscription to PredictedSports.

Usman vs Du Plessis

Start with the datapoint that frames the whole thing: age. Dig into our fight database and the younger fighter wins 61% of middleweight bouts, and when the gap stretches to five years or more that jumps to 70.9%, with a 58.1% finish rate on top of it. Du Plessis is 32. Usman is 39. That is a seven-year gap, and it is the number that frames this fight.

Get more specific and it gets louder. Filter for a fighter who is the favorite, younger, and at middleweight, and we have just 67 examples in the database, but that fighter has won 83.6% of the time. Du Plessis checks all three boxes.

The model has him at 74%, a touch ahead of the market's 67.8%, and all nine AIs on our board are on Du Plessis, from Gemini's 68% up to GPT-5.5 and GLM at 74%. The case against Usman is not really about Usman. It is about a 39-year-old walking in off a 399-day layoff, and fighters returning from a year or more off win just 43.8% of the time. He said himself he did not start preparing until the fight was announced. That is a rough way to meet a 32-year-old who has spent the last two years in five-round wars with Chimaev, Strickland, and Adesanya. Our number leans decision, with the fight going past the 2.5-round mark about two times in three.

Cannonier vs Duncan

Another age game, and this one is even more lopsided on paper. Jared Cannonier is 42 and has fought once in the last 11 months. Christian Leroy Duncan is 30, riding a four-fight win streak, and was in the cage 119 days ago. The reach edge is Duncan's too, 79 inches to 77. Every AI on the board sides with Duncan, and the model has him at 69%.

Here is where we get a little contrarian, though. The market has Duncan at 75%, and our number on Cannonier (31%) is actually higher than the market's (27%). So while the pick is Duncan, we think the line is a touch wide. Cannonier still cracks, but he is 0-2 against switch-stance opponents, which is exactly what Duncan is, and the defensive gap is the real tell: Duncan absorbs 2.59 significant strikes a minute to Cannonier's 4.35. Over three rounds that leakage compounds, and at 42 off a long layoff, grinding for 15 minutes is a big ask. We lean Duncan on the cards, but if you like a live dog, our numbers say Cannonier is the spot.

Hooper vs Mitch Ramirez

And to close, the easiest pick on the whole card. Chase Hooper is the one fight where the model, the market, and all nine AIs are holding hands: 74% on our number, 75.5% on the market, and unanimous across the board. This is also our highest-conviction call of the night, and the Oracle likes Hooper to do it on the ground, with a submission the single most likely finish at 49%. Mitch Ramirez has a puncher's chance, everyone does, but if there is a chalk pick on this card, this is it.

That is the main card, run through the model and the AI board. We grade every one of these in public, win or lose, so check back after Saturday to see how it aged. And if you want this same breakdown for every fight on every card, not just the ones we write up, that is what Pro is for.