UFC Fight night: Umar Nurmagomedov vs Song Yadong

The primary advantage remains Umar’s ability to change phases repeatedly. Song is dangerous enough on the feet that Umar has little incentive to conduct a prolonged kickboxing fight. Umar can kick from range, threaten level changes, attack from the clinch and force Song to continually defend transitions. Importantly, this passes our Offensive Wrestling Validation requirement: Umar’s ability to impose wrestling has been demonstrated against high-level UFC bantamweights rather than inferred from pedigree.

Song nevertheless deserves considerably more equity than a typical opponent facing Umar. He is powerful, fast and difficult to overwhelm physically. Umar’s title fight against Merab also showed that elite pressure, physicality and sustained scrambling can disrupt his preferred rhythm. Song cannot reproduce Merab’s wrestling volume, but he can make failed entries expensive with punches and force Umar to work harder than expected.

Our newer TDGD + SCV rules matter here. Even if Umar consistently gets Song down, takedowns should not automatically be converted into submission probability. Song has extensive high-level experience and just demonstrated offensive submission ability himself against Figueiredo. Umar may obtain control positions without finding a clean finishing sequence. That pushes a substantial portion of Umar’s method probability toward decision rather than submission.

Song’s best path is damage. He carries the harder conventional boxing threat and should have opportunities when Umar exits exchanges or enters from too far away. Because this is five rounds, however, Song needs to repeatedly prevent Umar from banking control-heavy rounds. Even moderately successful wrestling can force Song to hesitate on combinations and reduce the striking volume necessary to win a decision.

I also ran the Market Sanity Check because the current UFC-listed line is approximately Umar -500 / Song +375. That market implies a substantially larger gap than my model. I would not follow the market to ~83% Umar. Song’s power, durability, improved grappling evidence, home-country setting and five-round opportunity create too many legitimate upset mechanisms. At the same time, I don’t see enough to pull Umar below ~70% because his phase versatility and proven wrestling remain the most reliable weapons in the matchup.

Model confidence: High

Nurmagomedov 72% vs Song 28%

At distance, I favor Yan fairly clearly. Her UFC distance striking differential is approximately +2.72 significant strikes per minute, compared with +0.89 for Gomes, while Yan’s distance striking defense is substantially better. Gomes’ pressure and power make individual exchanges dangerous, but over three rounds Yan should be better at controlling where those exchanges happen, exiting afterward and accumulating cleaner scoring sequences.

However, this is exactly the type of matchup where our VADU + EWAU rules prevent me from giving the established veteran too much credit. Yan is 37; Gomes is 26. That’s an enormous athletic-age differential for strawweight. Gomes is also still developing, whereas Yan is much more likely to be at or beyond her physical peak. Yan’s technical superiority therefore gets discounted rather than projected forward unchanged.

Gomes also isn’t merely a knockout hunter. Against Pennington she landed 3 of 10 takedowns and accumulated 3:51 of control, including substantial third-round control. More broadly, she has been much more willing than Yan to wrestle. Under USN/OWV, I wouldn’t call Gomes the superior grappler simply from those numbers, but the wrestling gives her another mechanism for disrupting Yan’s preferred clean kickboxing fight.

The major argument for Gomes is therefore pressure + youth + physicality + damage. She owns seven KO/TKO victories, while Yan has historically been much more decision-oriented at UFC level. If Gomes consistently closes distance rather than following Yan around the perimeter, her harder individual connections could outweigh Yan’s cleaner volume. This activates our Damage Priority Override, and it’s why I think Gomes has a very real upset path rather than treating this as a veteran-versus-prospect mismatch.

The Market Sanity Check agrees with keeping the probabilities relatively close. Current prices are around Yan -150 to -170 and Gomes roughly +120 to +135. My 61/39 split is therefore close to the market. I considered moving Gomes into the 42–44% range because of the age differential, but Yan’s proven distance defense, substantially stronger strength of schedule and technical consistency keep her above 60%.

Model confidence: Moderate

Yan 61% vs Gomes 39%

The central matchup advantage is speed plus damaging offense. Aori generally needs to occupy the pocket and push opponents backward to produce his best work. That’s dangerous against Asakura because Kai is substantially more explosive in those transitional exchanges. Aori’s willingness to absorb strikes while advancing gives Asakura repeated opportunities for straight counters, knees and combinations before Aori can establish his preferred pressure.

The model nevertheless needs to respect DLP with Asakura. His offensive explosiveness has historically been accompanied by defensive vulnerability, and Alexandre Pantoja submitted him in his UFC debut before Tim Elliott submitted him again. Those losses don’t make Aori a likely submission winner—SAR/SCV prevents that extrapolation—but they do demonstrate that Asakura shouldn’t receive an enormous probability merely because his offensive ceiling is much higher. Aori’s toughness, pressure and experience give him meaningful upset equity if Kai fails to get him out early.

Aori’s wrestling is the most interesting alternative path, but OWV limits how heavily I weight it. He has used takedowns in UFC competition, yet I don’t see sufficiently strong evidence that he can consistently impose a control-heavy wrestling game on someone of Asakura’s athletic level. More importantly, access to occasional takedowns isn’t equivalent to sustained control or submission conversion. I therefore wouldn’t build the upset prediction around grappling.

Our PDM/CDI rules also reduce Kai’s advantage somewhat. Aori can turn technically cleaner fights into ugly pocket exchanges, and Asakura isn’t the kind of defensively airtight favorite we should price at 80%+ simply because he’s faster and more dangerous. If Aori survives the opening exchanges and makes this a high-volume attritional fight, Kai’s probability decreases noticeably.

Still, DPO ultimately favors Asakura. Aori’s style provides too many opportunities for the faster and considerably more dangerous offensive fighter. I expect Aori to survive some early damage, which is why I’m choosing Round 2 rather than Round 1. Asakura’s finishing threat remains stronger than his ability to reliably dominate three rounds, making KO/TKO the most coherent primary method.

Model confidence: High

Aori 27% vs Asakura 73%

The biggest piece of evidence is unusually strong: they already fought on May 30, with the fight ending as a no contest from an accidental low blow at 1:45 of Round 2. Before the stoppage, Mudaerji was clearly having the better fight statistically. He had outlanded Perez 26-9 in significant strikes, scored the fight’s knockdown, landed 1 of 2 takedowns and accumulated more control time. That doesn’t guarantee the rematch follows the same trajectory, but direct matchup evidence deserves considerably more weight than theoretical stylistic projections.

The striking matchup explains why. Mudaerji has approximately a 6.5-inch reach advantage and absorbs only 2.29 significant strikes per minute while maintaining 63% significant-strike defense. Perez lands 4.22 but absorbs 3.24 with 59% defense. In their first meeting, Perez struggled badly to bridge that distance, connecting on only 9 of 45 significant attempts. That is exactly the kind of matchup-specific evidence our model shouldn’t ignore.

Perez’s strongest adjustment is wrestling. He averages 2.07 takedowns per 15 minutes, and Mudaerji historically has been most vulnerable when opponents force him into grappling exchanges. But OWV + Control Conversion Requirement keep me from automatically converting Perez’s wrestling advantage into round-winning control. In the first fight he went 0-for-2 on takedowns while, surprisingly, Mudaerji was the fighter who actually completed one. Perez can certainly improve his entries in the rematch, but the first fight weakened rather than strengthened his projected wrestling advantage.

There is also a major submission-defense asymmetry, although MVC prevents me from overreacting to it. Six of Perez’s ten professional losses have come by submission, including recent losses to Alexandre Pantoja and Asu Almabayev. Conversely, six of Mudaerji’s seven losses have also come by submission. Neither fighter, however, has sufficiently reliable submission access in this particular matchup for me to make submission the primary prediction. SAR + SCVtherefore push both submission paths below the decision outcome.

The age/trajectory component also favors Mudaerji. He’s 30 versus Perez’s 34, has won three consecutive completed fights, and Perez has accumulated substantially more high-level mileage. I wouldn’t activate an aggressive VADU penalty at only 34, especially because Perez knocked out Charles Johnson this January, demonstrating that his offensive explosiveness remains intact. But it provides another small reason not to assume Perez will dramatically reverse the first fight.

What keeps this at 59/41 rather than 65/35 is rematch adjustment. Perez now knows exactly what failed. More committed cage pressure, low kicks and layered wrestling could materially change the fight, and Mudaerji’s historical submission vulnerability means one bad scramble can alter everything. Perez also has significantly more UFC experience against elite flyweights. This is therefore not a matchup where I want the model extrapolating 6½ minutes of dominance into excessive certainty.

Model confidence: Moderate

Perez 41% vs Mudaerji 59%

This is a matchup where USN needs to be applied aggressively. Neither fighter has enough high-level evidence for me to assign strong negatives simply because certain skills haven’t been demonstrated against UFC-caliber competition. At heavyweight in particular, sparse regional footage and quick finishes can conceal large portions of a fighter’s game. The appropriate response is wider uncertainty—not artificially large probability gaps.

Liu’s most obvious path is also highly credible: physicality and early damage. He has the size and power to make Rodrigues pay immediately for defensive mistakes, and heavyweight finishing variance means that advantage cannot be treated as a minor consideration. Under DPO + RFT, regional knockouts deserve some translation into UFC-level finishing probability, but not enough to assume Liu can reproduce them against substantially better resistance.

Rodrigues gets the slight overall advantage because his offensive options appear broader. He can threaten in transitions and doesn’t necessarily have to remain in prolonged pocket exchanges with Liu. But SAR/SCV are important here: having submission ability isn’t sufficient for me to predict a submission unless Rodrigues demonstrates reliable positional access and conversion against comparable opponents. I’m therefore treating grappling primarily as something that improves his overall win probability rather than making submission the predicted outcome.

The Early Survival Gate is particularly important. Rodrigues’ advantage increases if this gets beyond the initial heavyweight firefight, whereas Liu’s most dangerous window should be early. That means Rodrigues cannot simply be favored because I expect him to be the better fighter over an extended contest—the model has to price the probability that Liu damages or finishes him before that superiority becomes relevant. That keeps Liu’s equity unusually high at 45%.

Ultimately, I see a volatile heavyweight fight rather than a matchup with a dependable favorite. Rodrigues has slightly more ways to solve problems as the fight develops, while Liu possesses the simpler and potentially devastating early path. If Rodrigues survives Liu’s initial pressure without absorbing major damage, I favor him increasingly as Round 1 progresses and particularly entering Round 2.

Model confidence: Low–Moderate

Liu 45% vs Rodrigues Jr. 55%

The first thing that stands out is that these are two undefeated 9-0 finishers with identical seven-KO totals, so I don’t want the model making the mistake of treating Rojas as merely an opponent being brought in for Hasan. Under USN + RFT, Rojas’ regional success represents legitimate evidence of offensive ability; what remains uncertain is how well it translates when the speed, defense and countering quality increase.

Hasan nevertheless has the stronger competition-validation signal. He became CFFC flyweight champion and defended the belt, including a fourth-round TKO of Jason Eastman after successfully dealing with Eastman’s takedown and submission attempts. More recently, Hasan stopped Renaldy Manse and then destroyed Saikia in 45 seconds on DWCS. That progression matters under PMO: rather than simply accumulating regional wins, Hasan appears to be getting more dangerous as the opposition improves.

The physical geometry also favors Hasan. He is listed at 5’7” with roughly a 70–71-inch reach, whereas Rojas is 5’5” and one source lists his reach at 65 inches. I wouldn’t build the prediction around the uncertain Rojas reach measurement, but Hasan’s height and demonstrated ability to attack from range should make Rojas work through a dangerous layer before establishing his preferred exchanges.

There is an important DPO/CDI interaction here. Rojas has genuine power, so Hasan cannot freely exchange simply because I consider him the cleaner striker. Seven knockouts in nine victories demand respect. Rojas’ recent body-kick-and-punch TKO over Wallafy Fernandes is also useful evidence that his finishing isn’t dependent on one weapon. If Hasan becomes overconfident after his 45-second DWCS knockout and turns this into a firefight, Rojas’ upset probability rises considerably.

Where Hasan separates himself is offensive diversity and demonstrated defensive response. He has already won by RNC, and against Eastman he showed the ability to defend grappling attacks before taking over with strikes. Rojas has zero professional submission victories. Following USN, that doesn’t mean Rojas cannot grapple; it means we shouldn’t manufacture a grappling advantage without evidence. Hasan therefore has the better validated secondary path if the striking becomes uncomfortable.

I considered pushing Hasan into the 73–75% range, but that would violate several of the corrections we’ve made to the model. Rojas is undefeated, in his athletic prime, has substantial finishing power and has not yet provided evidence of a persistent defensive weakness. USN + UEC + RFT specifically warn against creating huge gaps in low-data prospect fights merely because one fighter comes through the better-known developmental pipeline. The appropriate adjustment is to favor Hasan clearly while preserving meaningful Rojas knockout equity.

The method is more aggressive than the winner probability. Rojas’ forward offense should give Hasan countering opportunities, but I don’t necessarily expect an immediate collapse. If Rojas survives Hasan’s initial speed and power, the accumulating reads and Hasan’s broader offensive toolkit should become increasingly important. That makes Round 2 KO/TKO my strongest individual outcome, with Hasan by decision and Rojas by KO/TKO the principal alternatives.

Model confidence: Moderate-High

Hasan 69% vs Rojas 31%