Gamrot vs Salkilld: where our model splits from the board
Gamrot vs Salkilld predictions: our model disagrees with 25 sportsbooks on four fights tonight, and marks down every favorite priced above 70 percent.
UFC Fight Night: Gamrot vs Salkilld runs tonight from the Meta Apex in Las Vegas, prelims at 5pm ET and the main card at 8pm ET, on Paramount+ in the US with CBS carrying it free, TNT Sports in the UK, Polsat Sport Fight in Poland. Twelve fights survived the week. I will get to which ones died in a minute, because that is part of tonight's story too.
Here is the part I actually want to talk about. We price every UFC bout with a model, and this morning I lined its numbers up against the closing consensus from 25 sportsbooks with the vig stripped out. On four of the twelve fights the two disagree by more than nine percentage points. That is a lot for one card. What makes it worth writing about is that the disagreements are not random.
The model does not believe in big favorites tonight
Four fights on this card have a market favorite priced at 70 percent or better. The model marks down all four of them.
| Fight | Market | Our model | Gap |
|---|---|---|---|
| Ty Miller vs Billy Ray Goff | Miller 77.2% | Miller 68.0% | -9.2 |
| Juliana Miller vs Ravena Oliveira | Miller 72.7% | Miller 61.7% | -11.0 |
| Steven Asplund vs Guilherme Pat | Asplund 72.4% | Asplund 61.8% | -10.6 |
| Manoel Sousa vs Richie Miranda | Sousa 72.6% | Sousa 68.2% | -4.4 |
Four for four, all in the same direction. Go the other way and the pattern holds: of the two fighters the market has priced below a third, the model lifts both. Darren Elkins goes from 14.4 percent to 20.8. GiGi Canuto goes from 32.3 percent to 50.7, which is the single biggest split on the card.
I want to be honest about what that means, because it would be easy to dress it up as the machine seeing something 25 books missed. It mostly is not. A model trained on thousands of fights learns something the betting market keeps having to relearn: heavy favorites in the UFC lose more often than their price implies, and short-notice hype dogs are live. Shrinking extreme prices toward the middle is a sensible habit, not a prophecy. When the model says a 77 percent favorite is really 68 percent, it is not naming an upset. It is refusing to be as sure as the board is.
Where that habit becomes interesting is when it collides with a fight nobody has any tape on.
The biggest gap is the one I trust least
Canuto vs Foro is a women's strawweight prelim that was bolted onto this card on Wednesday, three days out. The market has Carol Foro (9-1) as a better than two to one favorite at -237, with Canuto out at +198. Our model calls it 50.7 percent Canuto, a coin flip.
An eighteen point disagreement is the sort of number that makes you want to write "the model loves the underdog." I am not going to, and the reason is in what the model is looking at. Both women are making their UFC debuts tonight. Neither has a single fight in our native database. The model is pricing this off pre-UFC records, a reach edge, and an age gap: Canuto is 23, Foro is 29, and age is the strongest single input in that bout after the base rating. Those are real signals. They are also all it has.
Foro walks in 9-1 with a two thirds finish rate before the UFC. Canuto is 7-1 and finished 14 percent of hers. If you think the professional record and the finishing habit matter more than six years of youth, you agree with the sportsbooks, and I think you have the better of it. This is the fight on the card where I would take the model's number as a question rather than an answer.
Prediction: Foro by decision, and I reserve the right to be embarrassed by this in three hours.
Gamrot was never the favorite
A small correction to something you may have read this week. Mateusz Gamrot did not open as the favorite and get bet down. He opened an underdog and stayed one. Of the 25 books we track, 21 opened him at plus money and the rest no better than a pick'em. He sits at +116 now against Quillan Salkilld at -136. Strip the vig and the opening consensus had Gamrot at 45.8 percent; he is at 44.5 today. The line has moved about a point since it opened. Nothing happened. The market decided this one in advance and never wavered.
The model is a little harsher than the board: 39.0 percent for Gamrot against the market's 44.5. And the driver that does the damage is not the tape, it is the birth certificate. Age is the single largest input pushing that number down, ahead of the ratings gap and ahead of both men's UFC records. Gamrot is 35. Salkilld is 26.
You can see why the market got here. Salkilld is 5-0 in the UFC with four first round finishes, and in May he walked through Beneil Dariush in a round at UFC Fight Night: Della Maddalena vs Prates. That is the kind of win that reprices a man. Gamrot, ranked eighth at lightweight to Salkilld's twelfth, is 9-4 in thirteen UFC fights and just submitted Esteban Ribovics in the second round at UFC 327. He is not fading. He is being priced as though the calendar decides this.
So here was my argument, and then here is what happened to it. The one man on Salkilld's UFC ledger who made him work, Yanal Ashmouz, dragged him into the only decision of his run. Gamrot lands 5.22 takedowns per fifteen minutes and stuffs 83 percent of what comes back. Wrestle the boy, I wrote in my notes, and this is a live dog.
Then I pulled Salkilld's numbers. He shoots more than Gamrot does, 7.25 takedowns per fifteen minutes, and defends at 80 percent. This is not a striker against a wrestler. It is two wrestlers, and the grappling more or less cancels. What does not cancel is that Salkilld gets hit: 45 percent striking defense against Gamrot's 60, while landing 5.01 significant strikes a minute to Gamrot's 3.21. He is the more dangerous man and the more available target, and nobody has yet asked him a question in round three.
Prediction: Gamrot by decision. I am taking the plus money and the older man's defense, and I know both the market and our own model think I am wrong.
Elkins at 42, and 16 months away
Photo: RapidSpin33 (talk) (Zach Able) / CC BY-SA 3.0
Elkins (29-12) is +574 against Yadier del Valle (10-1) at -825, the widest price on the card. He is 42 years old, he has 30 UFC fights, and he has not fought since April 2025, when Julian Erosa knocked him out in the first round at UFC 314. Sixteen months and a first round knockout is not a résumé the numbers can love.
The model gives him 20.8 percent anyway, six points more than the board, and its two biggest inputs against him are exactly the two you would guess. It is not a contrarian read. It is the same shrinkage as everywhere else on this card, applied to a man the market has written off completely.
I have watched Darren Elkins get written off since roughly 2017. Every time, the honest thing to say is that the record is the record and the odds are approximately right. That is still the honest thing to say.
Prediction: del Valle by TKO.
The card that kept moving
One more thing worth logging, because it is the sort of detail that vanishes once the results are in. This card churned hard in its final 72 hours. Amanda Lemos vs Alexia Thainara, a matchup between the eighth and twelfth ranked strawweights, came off the card on Thursday and went back on inside half an hour. Miles Johns lost Jessie Rosas and got Gianni Vazquez instead on Friday, roughly a day out, which forced the model to reprice that bout from 57.3 percent to 62.5. The whole event moved three hours earlier on Tuesday. Canuto vs Foro was bolted on with three days to spare.
None of that will be in the record book tomorrow. All of it is in the prices.
So: four fights where our model and 25 sportsbooks cannot agree, one of them a debut matchup where the model is guessing with more confidence than it has earned, and a main event that the market settled two weeks ago and never reopened. The pattern says fade the heavy favorites. The pattern has also been wrong before, and unlike the model, I have to watch it happen live.
Full card, odds, model numbers and start times are on the event page, and the country by country broadcast list is here. Join us tonight.