Can UFC Faceoffs Predict the Winner? What Data Says

Quick answer

UFC faceoffs cannot reliably predict the winner. One 2013 study of 76 UFC bouts found that fighters who smiled more intensely before fighting lost more often. The link survived controls for height and betting odds. Yet the study was small, male-only, and never produced a simple win rate for smiling fighters. Later research on fighter faces ranges from 55% prediction accuracy to no link at all. Treat a staredown read as a story, not a model input.

Data snapshot: The direct smile study covers 76 UFC bouts from 2008-2009. Later studies cover 114 fight pairs in 2015 and 44 MMA fighters in 2019. Two 2022 replications cover 520 fighters and 1,367 bouts in one sample, and 516 UFC fighters in another.

What would a useful UFC faceoff prediction need to show?

A useful cue must predict a specific bout before the result is known. It must also add information beyond records, skill, size, and market odds.

That sounds obvious. Most viral staredown analysis fails both tests.

Fans usually mix three different questions:

• Can a brief expression, such as a smile, predict tomorrow's result?
• Can a neutral face reveal a fighter's strength or career ability?
• Can viewers read fear, confidence, or fatigue from live body language?

Those questions need different data. A study of facial shape doesn't validate a claim about blinking. A study of still photographs doesn't test who stepped forward at ceremonial weigh-ins.

The closest direct evidence comes from one old UFC smile study. The wider face-perception research tests a related claim: whether appearance carries enough information to beat chance.

What did the UFC smiling study find?

Michael Kraus and Teh-Way David Chen studied 152 male UFC fighters across 76 bouts in 2008 and 2009. The photographs came from faceoffs held one day before each fight.

Four coders, unaware of the study's goal, rated each fighter's expression. The scale ran from 0 for neutral to 2 for a teeth-baring smile. Most smiles didn't include the eye movement linked with a spontaneous smile.

Winners smiled less intensely than losers. The group difference was statistically significant, t(150) = -2.69, p < .01. The association remained after the authors controlled for betting odds and height, F(1,148) = 5.17, p < .05.

That result is interesting. It isn't a betting system.

The paper doesn't report a clean win percentage for smiling and non-smiling fighters. You can't honestly turn its test statistics into “smilers lose X% of the time.” It also found no link between smile intensity and later career win percentage, r = .02.

The authors ran a second experiment about perception. They recruited 178 online participants, then removed 16 who recognized a fighter. The remaining 162 rated the same fighter as less dominant when smiling, 6.29 versus 6.80. They also rated him less aggressive or hostile, 4.12 versus 6.47.

That helps explain why fans read smiles as submission or nerves. It doesn't prove that the expression causes poor performance. A smile can reflect personality, promotion, tension relief, or a deliberate attempt to unsettle someone.

Can people pick MMA winners from faces alone?

The best answer is barely, in one experiment.

Anthony Little and colleagues showed observers 114 pairs of male MMA fighters. Each pair contained the winner and loser of a real bout. The images were neutral faces, not moving faceoffs.

The 69 participants assigned to predict winners chose correctly 55.0% of the time. That cleared the 50% chance mark, Z = 2.35, p = .019. It is still only five extra correct calls per 100 choices.

The design also gives viewers information that a fair forecasting test should handle carefully. Faces may reveal age, size, body fat, and other broad physical traits. Those traits can overlap with real competitive differences. The study didn't show that viewers detected confidence or fear.

Later work makes the signal look less stable.

Why the facial-structure findings conflict

The 2019 Czech study used far better-controlled images than a promotional faceoff. Researchers photographed 44 fighters through a standardized 360° sequence. Ninety-four people rated their fighting ability.

Viewers thought heavier fighters and those with higher anaerobic performance looked more capable. Yet perceived ability didn't match actual competition success. The sample was small and came from one country, but the null result matters.

The 2022 facial-ratio replication went much larger. Neil Caton and colleagues tested 520 fighters. One manually measured eyebrow-based ratio predicted career success after controls for body mass index and total fights, p = .004.

That sounds stronger until the target becomes an individual bout. Across 1,367 fights, no facial-ratio measure predicted the winner. A career association isn't the same as a next-fight signal.

A separate preregistered project used 516 UFC fighters and 36,636 facial landmarks. Two panels of 500 people rated either aggressiveness or fighting ability. Masculine-looking faces seemed more aggressive. The perceived fighting-ability effect disappeared after body-size controls, and facial structure didn't predict success.

Measurement choices also changed results. Eyebrow landmarks, eyelid landmarks, manual coding, and automatic coding didn't behave alike. If a cue works only under one reasonable measuring method, you shouldn't trust it at cage side.

What the Makhachev-Garry faceoff actually tells us

UFC 330 gives the current version of this debate. Welterweight champion Islam Makhachev defends against Ian Machado Garry on August 15, 2026, in Philadelphia.

After their first faceoff, Garry said he sensed “nervous energy” from Makhachev. He also said the same promotional faceoffs were fake, made for social media and contractual obligations.

Both statements can fit together. Garry may believe he read a real moment inside a staged task. An outside viewer still can't test that read from a short clip.

The bout result won't validate the method either. If Makhachev wins, fans can call his stillness confidence. If Garry wins, the same clip becomes evidence of nerves. That is outcome bias: the ending rewrites the earlier image.

The honest pre-fight position is simpler. Neither man's expression supplies a measured probability adjustment. Their records, matchup, recent performance, injuries, and odds carry testable information. The faceoff supplies a memorable narrative.

How to read UFC staredowns without fooling yourself

You can enjoy faceoff psychology and still keep the analysis clean.

Compare the fighter with his own baseline

A behavior is more useful when it differs from that fighter's usual routine. Even then, the change may have many causes. Compare several past appearances before calling it fear.

Write the claim before the fight

Record exactly what you saw and what you think it predicts. “He looked off” is too loose. A dated note stops you from changing the story after the result.

Separate official facts from visual guesses

A missed weight, confirmed injury, or late opponent change is a fact. Sunken cheeks, eye contact, and posture are interpretations. Don't let the second group borrow certainty from the first.

Ask whether the cue changes your probability

If you can't state a repeatable rule or show past accuracy, don't move the number. A prediction model needs inputs that can be defined before every fight.

Never diagnose a fighter from a clip

A camera angle, lighting, editing, dehydration, and ordinary stress can change appearance. Faceoff video cannot establish a medical or mental condition.

How solid is this data?

The evidence is real but indirect. Only the 2013 paper tests an expression during UFC faceoffs against the next bout. It covers 76 bouts, all involving men, from two seasons more than 15 years ago.

The 2015 study predicts paired winners from static faces. It doesn't test spontaneous behavior. Its 55.0% result is statistically above chance but weak for real forecasting.

The other studies examine career records, physical performance, or facial geometry. They use samples from 44 to 520 fighters. Their mixed findings show how sensitive the answer is to the photograph, outcome, body-size control, and measurement method.

No source here tests modern faceoff video out of sample. None shows that a staredown cue improves a model already using odds and fight data. That missing test is the one analysts need.

FAQ

Do fighters who smile at UFC weigh-ins usually lose?

One 76-bout study found that winners smiled less intensely, even after height and odds controls. It didn't publish a simple loss rate, and no modern direct replication confirms the pattern.

Does looking away during a UFC faceoff mean fear?

No reliable MMA study shows that looking away predicts defeat. Eye contact can reflect habit, instructions, promotion, culture, or strategy, so one clip cannot identify fear.

Can body language predict a UFC fight?

Current evidence doesn't support body language as a standalone UFC predictor. Static-face studies range from 55.0% accuracy to null results, while direct faceoff research is limited to one small smile study.

Are UFC staredowns staged?

They are promotional events arranged for cameras, media, and fans. Fighters can still react naturally, but the setting encourages performance and makes isolated gestures hard to interpret.

Should faceoffs affect a UFC prediction model?

Not without a defined cue and out-of-sample validation. Records, opponent quality, age, pace, grappling, striking, injuries, and market price offer clearer inputs than subjective staredown readings.