UFC 330: our model's two tickets and the champ it won't back

The sharpest disagreement between our fight model and the sportsbooks tonight is not the main event. It is the last fight most people will scroll past: Rafael Tobias against Lucas Fernando, two Brazilians with one UFC fight between them. The books make Fernando a -281 favorite, about 71%. Our model reads the same fight 58-42 for Tobias, who is +232. That is a 29-point gap, the widest on the card, and it is either the best price of the night or a flashing sign that nobody actually knows anything. I will take both possibilities seriously below.

First, the standing disclaimer, because it belongs at the top and not buried in a footer: everything here is our model's calibrated opinion measured against 24 to 25 sportsbooks, odds as of 14:57 UTC today. It is not betting advice, and if you are under 18 none of this is for you.

Photo: Head of the Respublic of Dagestan / CC BY 3.0

UFC 330: Makhachev vs Machado Garry goes down tonight, Saturday, August 15, at Xfinity Mobile Arena in Philadelphia, streaming on Paramount+ in the US (full country list here). Every fight on the card carries our model's live win probability on the event page. What follows is how those numbers assemble into the two tickets now sitting there, and, just as important, what we refused to put on them.

Why the champ is not on either ticket

Start with the name you expected to see first. The books price Islam Makhachev at -347 over Ian Machado Garry, an implied 75%. Our model gives the champion 61%. That 14-point spread is the widest favorite overvaluation on the card, and it is exactly the kind of leg that quietly kills parlays: you pay 75-cent prices for what the model believes is a 61-cent probability.

To be clear about what the model is and is not saying. It is not predicting a Garry upset; 61% still makes Makhachev (28-1, on a 16-fight win streak) the clear favorite over a man who has won 10 of 11 in the UFC. It is saying the price is wrong. Garry at +283 is the mathematically interesting side of that fight, and we still left it off the upset ticket, because pairing it with our other underdogs blew past the price ceiling we cap tickets at. Discipline is boring. That is rather the point.

The favorites ticket: 9.89x, four model picks and one house call

Five legs, all market favorites or coin flips, combined price 9.89x (+889 American).

Photo: Jprietf / CC BY-SA 4.0

Kaue Fernandes over Jalin Turner, -106. The books call this a coin flip. The model makes it 58-42, its biggest favorite-side edge of the night. Turner (8-6 UFC) has lost three of his last five, two of them by first-round finish, while Fernandes rides a three-fight streak with first-round knockouts in two of his last three. The stat pair that decides it for me: Fernandes absorbs 1.4 significant strikes a minute, Turner absorbs 4.3, and Fernandes has won round one on the scorecards in every UFC appearance. Someone starts fast here, and it is probably not the man who was submitted inside a round by Ignacio Bahamondes last year.

Joel Álvarez over Chidi Njokuani, -314. Model 71%, books 73%. When the model and the market agree this closely, a leg is doing exactly what a favorites ticket needs it to do. Álvarez (8-3 UFC) has finished seven of his eight UFC wins and holds a large grappling edge over the 37-year-old Njokuani, who has dropped two straight. The caveat I owe you: Álvarez was submitted by Yaroslav Amosov in May. Amosov is a different class of grappler than anyone Njokuani has ever pretended to be.

Charles Johnson over Eduardo Chapolin, -117. This leg carries a thin-data warning, and honesty requires explaining it: the warning is about Chapolin, a debutant with zero UFC fights, not about Johnson, who has fifteen of them against far stronger opposition. Chapolin was submitted on The Ultimate Fighter as recently as February 2025. When one corner is a mystery, we side with the corner that is not, and the model leans the same way at 53%.

Tresean Gore over Vicente Luque, -110. The house call, and the only leg on either ticket where we overrule our own model, which leans Luque at 56%. I want that stated plainly rather than laundered. The case for Gore: three of his UFC wins have come by submission, including Azamat Bekoev this April, and he lands his takedowns at a 73% clip against a 34-year-old Luque who defends fewer than 60% of takedowns and has been finished twice since 2024, knocked out by Joaquin Buckley and submitted by Kevin Holland. Luque remains the more proven fighter, which is why the model likes him. I think the grappling is where this fight lives, and it is not his half of the mat.

Myktybek Orolbai over Jeremiah Wells, -1104. The cheapest leg and the surest. Orolbai averages nearly nine takedowns per 15 minutes, has won three straight including a first-round knockout of Jack Hermansson, and faces a 39-year-old Wells who has fought exactly once since February 2024. Even our model's 79% is more cautious than the books' 88%.

The underdogs ticket: 19.85x, two dogs the model refuses to price like the books

Dustin Stoltzfus over Mansur Abdul-Malik, +498. The model sees this 60-40. The books see 84-16. A 24-point gap is normally our cue for humility, so here is the concrete case instead of hand-waving: Abdul-Malik is coming off a third-round knockout loss in March, and when his fights leave the first round his numbers collapse. Stoltzfus (3-7 UFC, and yes, I can hear you) has gone the full 15 minutes in back-to-back fights against Kelvin Gastelum and Nursulton Ruziboev, and the experience gap is stark: 104 career minutes in the cage to Abdul-Malik's 26. Full disclosure, because we show our work even when it cuts against us: the one stylistic edge our Fight DNA data flags on this card actually favors Abdul-Malik's wrestling against Stoltzfus's ground defense. Even granting that, a fighter the model prices at 40% is being sold at 16%.

Rafael Tobias over Lucas Fernando, +232. The 29-point gap from the top of this piece. Both corners are nearly data-free at UFC level, which is why this leg carries every warning our system can attach, but thin data cuts both ways: the books are laying -281 on Fernando's LFA tape (he has won four straight there since a 2023 decision loss), while Tobias, 23 years old to Fernando's 29 and 14-1 as a pro before arriving, is the only man in the cage who has actually fought in the UFC. He lost that debut on the cards to Diyar Nurgozhay in March, and apparently that was enough for the market to write him off. The model disagrees by 29 points. When the gap is that wide we treat it as our ignorance rather than the books' error, and at +232, those are terms I accept.

What we refused, and what the rules caught

Photo: MM-eh / CC BY 3.0

The tickets are built under standing trust rules, and tonight they are worth spelling out because the discipline is the actual product. Probabilities get clipped to a 5-to-90 percent band. Any corner with fewer than three UFC fights gets flagged. Any fight priced by fewer than three books gets flagged. And any model-market gap over 15 points is treated as our ignorance, not the books' mistake, flagged on every surface where the number appears. Tonight those rules excluded nothing outright, all twelve fights were priceable, but three of our seven legs carry flags in the open, and you have seen each one above.

The refusals were choices, not rules. Mackenzie Dern at -195 never made the favorites ticket because the model does not back her at all: it makes Gillian Robertson the 53% side of that fight at +167 odds. Robertson, riding a five-fight win streak into a title shot, is the one that got away, left off the underdogs ticket only because Stoltzfus plus Tobias filled the price band with a higher combined probability, and our tickets never share or split a fight. If you want one model-approved underdog outside our tickets, she is the pick, and I say that knowing Dern just won the belt.

The honest math

The favorites ticket combines to a 7.6% model probability against the books' implied 8.3%. Fair price for our probability would be 13.20x; the offered stack pays 9.89x, an expected value of 0.75. Read that plainly: the Gore override costs real money in expectation (the model's own version of this ticket, with Luque, priced at 0.97), and we publish that cost instead of hiding it. The underdogs ticket combines to a 23.1% model probability against the market's 4.7%, fair value 4.32x against an offered 19.85x, expected value 4.59 if, and only if, the model is right.

One more sentence you should act on: legs on the same card are not independent, injuries and judging and a shared fight-night environment correlate outcomes, so treat every combined probability above as an upper bound, and size anything you do accordingly.

Both tickets, with per-leg reasoning and every warning attached, are live on the UFC 330 event page. Our model's opinion, not betting advice, 18 and up, and if the 3-7 guy taps the phenom tonight, you saw the number here first. Make your own calls in pick'em and argue with me when the model is wrong. It will be, somewhere, before the main event walks.