UFC 333 predictions: model backs Reyes and Keita against the odds

The two title fights at UFC 333 have had their week on this blog already. Volkanovski vs Evloev is a coin flip at the sportsbooks, and Yan vs Dvalishvili 3 has the champion as the favorite. This piece is about three fights below them where our AI model and the betting market look at the same two people and see different fights.

The model's numbers for the card were generated on October 19, and both corners match the bouts as they stand today. The line is the UFCalendar consensus across 18 sportsbooks as of Wednesday morning. All of it is information, not betting advice.

Let me get one thing out of the way first, because it matters more than any single number below. Since mid-July, 174 UFC bouts have ended with a winner, a pre-fight model number and a consensus line on our site. The model picked the winner in 108 of them. The betting favorite won 123. In the 35 fights where the two disagreed, the model was right 10 times. So when the model breaks with the books on Saturday, treat it as a question, not a verdict.

The market column is the line converted into a probability with the bookmakers' margin stripped out, so it can be read side by side with the model.

Murzakanov vs Reyes: the widest gap

Photo: Chamatkar Sandhu / CC BY 3.0

No. 7 Azamat Murzakanov (16-1) against No. 8 Dominick Reyes (16-5) tops the prelims at light heavyweight, and it is the biggest disagreement on the card between two fighters with a real UFC sample. The line opened at -205 on Murzakanov on October 14, when four books had it up, stretched to -239 on October 18, and sits at -223 with Reyes at +180. That is roughly two in three for Murzakanov. Our model gives Reyes 58.3 percent.

The market's case is easy to write. Murzakanov has knocked out five of the six men he has beaten in the UFC, and he dropped each of Alonzo Menifield, Brendson Ribeiro and Aleksandar Rakic before stopping them. Reyes has been knocked out four times: by Jan Blachowicz and Jiri Prochazka in the second round, and by Ryan Spann and Carlos Ulberg in the first. The Ulberg loss, at UFC Perth last September, took 4:27.

The model is reading other inputs. The ones it weighs most heavily are size (Reyes is 6-foot-4 with a 77-inch reach, Murzakanov 5-foot-10 with 71), striking output (Reyes lands 4.99 significant strikes a minute in the UFC, Murzakanov 4.47) and age, with both men past 36. It also knows both are coming off the same night, UFC 327 in April, 196 days before this one. Murzakanov lost his unbeaten record there, knocked out by Paulo Costa at 1:23 of the third round after being outlanded 55 to 34 and dropped once. Reyes beat Johnny Walker on a split, 29-28 from Junichiro Kamijo and Sal D'Amato against 29-28 for Walker from Solimar Miranda, in a fight where Walker landed more significant strikes, 42 to 34.

One more number, labeled for what it is: our Power Index rating has Murzakanov slightly ahead, 53 to 47. So the model's lean toward Reyes is not coming from who has the better résumé. It is coming from the body types and the striking rates. The pick'em crowd is nearly split, 22 of 41 votes on Murzakanov.

Volkov vs Kuniev: same pick, different certainty

Photo: S.Troshchenko / CC BY-SA 4.0

Here the model and the market agree on the name and disagree on the confidence. No. 2 Alexander Volkov (40-11) opened at -130 against No. 7 Rizvan Kuniev (14-3-1, 1 NC), drifted out to -117 on October 17, and is -123 against +101 now. That is 52.6 percent, a hair over a coin flip. Our model gives Volkov 63.8 percent. The fans are further out than either: 48 of 60 pick'em votes are on Volkov.

Volkov turns 38 on fight night, and the model counts that against him. What it counts for him is the sample: 51 pro fights to Kuniev's 18 in its inputs, 19 of Volkov's in the UFC against three for Kuniev outside the Contender Series. Volkov's UFC striking numbers are the cleaner set, 5.08 significant strikes landed a minute at 58 percent accuracy to Kuniev's 3.74 at 48.

The market's case has a name, and it is Jailton Almeida. Volkov beat him at UFC 321 last October by split decision, 29-28 from Clemens Werner and Tony Weeks, 29-28 the other way from Mike Bell. Almeida landed 7 of 9 takedowns and held Volkov for 10:47. Volkov won it on efficiency, 24 significant strikes from 27 attempts. Kuniev met the same Almeida in February and won a unanimous decision. Almeida went 0 for 8 on takedowns, and it was Kuniev who did the controlling, 6:35 of it.

That is one fight each, and styles make fights, but it is the kind of comparison a line moves on. Kuniev's only UFC loss was a split to Curtis Blaydes in June 2025 (29-28 twice for Blaydes, once for Kuniev), and he stopped Tyrell Fortune at 1:12 of the third round in July.

Pico vs Keita: the line that flipped

Photo: WrestlingSavesLives / CC BY-SA 4.0

Losene Keita (17-2) took this fight on October 7, after Arnold Allen came out against No. 11 Aaron Pico (14-5). When the line first went up on October 14 it was -110 on both men. By the end of that day Keita was the -129 favorite with Pico at +103, and Keita stayed the favorite through October 17. Then it turned. Pico went favorite on October 19 and is -125 now, with Keita at +103. Our model has not moved with it: Keita, 59.6 percent. The pick'em crowd leans the model's way, 15 of 28.

Pico's case is the wrestling. In his last fight, at UFC 327, he beat Patricio Pitbull by unanimous decision, landing 100 of 181 significant strikes and 6 of 13 takedowns with 3:25 of control. Across his two UFC fights he averages 6.54 takedowns per 15 minutes. Keita has never attempted a takedown in the UFC, and in just under 18 minutes of UFC cage time nobody has put him down either: Nathaniel Wood tried once and missed.

The model's case is what happened in Pico's other UFC fight. Lerone Murphy knocked him out at 3:21 of the first round at UFC 319, even after Pico had landed 2 of 3 takedowns. Keita, meanwhile, knocks people down. He dropped Muhammad Naimov twice before finishing him at 2:54 of the first round in September, which works out to 1.7 knockdowns per 15 minutes in the UFC, while Pico has been dropped 0.8 times per 15. The model also likes Keita's record: a 90 percent career win rate to Pico's 74.

Keita's one UFC loss is the honest caveat. He dropped a split to Nathaniel Wood in March, on the night Movsar Evloev beat Lerone Murphy, with Derek Cleary and Ben Cartlidge giving it 29-28 to Wood and Vito Paolillo 29-28 to Keita. Wood outlanded him 67 to 52. That was a loss on volume. Pico is a featherweight who wants to take him down, which is a different problem, and a week of money has decided it is the harder one.

The number that decides how much any of this is worth

When the model has backed a betting underdog at 58 percent or more since mid-July, it has been right three times in nine. On Saturday it does it twice, on Reyes and on Keita, and it puts more faith in a 38-year-old heavyweight than the market does.

If both underdogs win, two of the model's longest calls of the autumn land on one card. If neither does, the books will have done what they did in 25 of the 35 fights where the two have disagreed. The prelims start at 12:00 p.m. ET, the numbers update on the event page, and the pick'em is open until the first bell. Who are you siding with, the machine or the money?