The Fight StoryClick here to expandThe story behind the stats.
As Kayla Harrison steps into the octagon against Amanda Nunes at UFC 334, fans can expect a clash of titans that mirrors some of the most electrifying showdowns in UFC history. Both fighters boast impressive records, with Harrison at 3-0 and Nunes holding a formidable 16-2. While they share the same height and weight, Nunes has a slight reach advantage at 69 inches compared to Harrison's 66 inches. Harrison’s striking accuracy of 62.1% gives her a technical edge, but Nunes has an impressive knockout win rate of 43.8%, showcasing her ability to end fights decisively, just like Nate Diaz did against Tony Ferguson at UFC 279.
Diving into their fighting styles, Harrison’s grappling prowess shines through her control time of 161.4 seconds per round, while Nunes has shown ferocity with a takedown accuracy of 56.7%. With both fighters in their prime and a win streak in Harrison's favor at four, this matchup is poised to be a tactical battle. The stakes are high, as a notable odds gap indicates that Nunes is favored, but Harrison’s undefeated record and a higher sub-win rate suggest that she is more than capable of pulling off an upset. This fight could very well be the beginning of a new era in women’s bantamweight, reminiscent of previous legendary clashes.
Prediction IntelClick here to expandOur call — and the case for it.
Stats Model Prediction
🔒 Prediction locked in before the fight · Predicted Oct 24, 2026, 00:04 UTC
Model Breakdown
Kayla HarrisonAmanda Nunes
Deep Learning Model α 26% weight71.1%28.9%
Deep Learning Model ξ 25% weight61.4%38.6%
Statistical Model δ 13% weight59.3%40.7%
Statistical Model β 21% weight58.0%42.0%
Statistical Model ε 15% weight40.8%59.2%
Ensemble63.9%36.1%
AI Reasoning (why the models picked this)
The ensemble prediction favors Kayla Harrison, driven by factors like both fighters being past their prime and Amanda Nunes's age and significant strikes absorbed. Despite Model ε favoring Amanda Nunes due to her Elo rating, calibration and weighting led to the ensemble's final prediction for Kayla Harrison.
Statistical Model β: Model β favors Kayla Harrison, citing that both fighters are past their prime and Kayla Harrison has absorbed fewer significant strikes.
Statistical Model ε: Model ε favors Amanda Nunes, primarily due to her higher Elo rating, a chess-style skill score.
AI-generated from the model outputs above — wording may occasionally be imprecise.
Predicted Winner
Kayla Harrison
63.9%
Confidence
* Model output, not betting advice. Bet responsibly.
*New fighters, prediction made with limited information
Predicted Duration
3rd Round at 1m15s (11m15s total)
Chance the Fight Is Still Going
79%61%48%42%38%
Rnd 1Rnd 2Rnd 3Rnd 4Rnd 5
38%
Chance of Decision
The dashed line is 50%. The round where the curve crosses it is the round named above — the midpoint of the distribution, not necessarily its single most likely outcome.
Matchup IntelClick here to expandPast fights with a similar statistical shape, plus how fights like this one typically end.
How Similar Fights Ended
288 historical bouts
Every past fight is placed by the SHAPE of its mismatch — how far apart the two fighters were on each of 37 career measures, setting aside which one was ahead. Fights that sit close together form a profile. Below is how the 288 bouts in this bout's profile actually ended.
53%Reached the judges
47%Ended early
31%KO/TKO16%Submission5:11Median finish, round 2
When those finishes came
Rnd 168
Rnd 240
Rnd 325
Rnd 42
Rnd 51
This is a base rate over similar past matchups, not a prediction about these two fighters. It is drawn from what happened historically, and the profile is blind to which corner is which — so it can say how fights like this one tend to go, and nothing at all about who wins.
Fights with a similar mismatch
Every fighter is reduced to a profile of 37 career measures as they stood going into the fight, standardised so that no single measure dominates merely for having bigger units — a far wider read than the six charted below. The bout is then described by the gap between the two: how far apart they were on each of those measures, setting aside which one was ahead. The fights listed here are those whose gaps line up most closely with this one. Because the match is on the shape of a mismatch rather than on a weight class or an era, a neighbour can come from an entirely different division.
80%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
80%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
78%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
77%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
77%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
77%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
75%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
75%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
74%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
74%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
74%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
73%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
73%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
72%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
72%How closely the gap between these two fighters resembles the gap in that historical bout — the size and shape of the mismatch, not the fighters themselves. Two completely different pairs can produce the same mismatch. 100% would be an identical gap; two fights picked at random score about 36%, so a number near that means there is no real comparable on record.Match
Statistics IntelClick here to expandA full head-to-head stat comparison — per-round output and career tendencies — against the UFC average.
Head-to-Head Statistics
The same per-round and advanced numbers each fighter's profile shows, side by side. The stronger value in each row is highlighted.
Stat
Kayla Harrison
Amanda Nunes
UFC AVG
Per Round Average
Strikes
19.4
19.7
13.2
Strike accuracy
62.1%
52%
43.9%
Takedowns
0.9
0.8
0.4
Takedown accuracy
37.5%
56.7%
33.0%
Sub attempts
0.3
0.2
0.2
Control time
161.4s
89.3s
48.9s
Knockdowns
0
0.17
0.08
Knockdowns absorbed
0
0
0.13
Advanced Fighter Statistics
win rateWins divided by this fighter's total recorded fights.
92.3%
88.9%
45.8%
recent win rateWin rate over the last 5 recorded fights only — a form signal, not a career figure.
80.0%
80.0%
37.6%
win streakCurrent run: a positive number is consecutive wins, a negative number consecutive losses.
4
2
-2
total fightsTotal recorded fights, including bouts outside the UFC.
13
18
—
head strike shareShare of this fighter's landed significant strikes that went to the head. Head, body and leg sum to 100%.
71.3%
67.8%
—
body strike shareShare of this fighter's landed significant strikes that went to the body. Head, body and leg sum to 100%.
19.9%
13.4%
—
leg strike shareShare of this fighter's landed significant strikes that went to the legs. Head, body and leg sum to 100%.
8.8%
18.9%
—
sig strikes absorbedSignificant strikes opponents land on this fighter PER ROUND, counted from the round-by-round UFC stats across their scored rounds. Lower is better.
5
10.62
15.2
takedowns absorbedTakedowns opponents complete on this fighter PER ROUND, counted from the round-by-round UFC stats across their scored rounds. Lower is better.
0
0.13
0.4
cardio sig ratioThis fighter's significant strikes landed in the LATER rounds divided by the EARLIER rounds. 1.0 is neutral: above it their output grows late, below it they fade. Shown only for fighters with at least 3 UFC bouts on record.
0.25
0.13
0.52
cardio defence ratioWhat OPPONENTS land on this fighter in the LATER rounds divided by what they land in the EARLIER rounds. 1.0 is neutral. LOWER is better — a high value means opponents find them more as the fight wears on. Shown only for fighters with at least 3 UFC bouts on record.
0.33
0.23
0.56
cardio takedown ratioThis fighter's takedowns in the LATER rounds divided by the EARLIER rounds. 1.0 is neutral. Shown only for fighters with at least 3 UFC bouts on record.
0.25
0.57
0.42
is orthodoxWhether the fighter's recorded stance is orthodox.
No
Yes
—
is southpawWhether the fighter's recorded stance is southpaw.
Yes
No
—
switch stanceWhether the fighter is recorded as switching stance.
No
No
—
finish rateShare of this fighter's WINS that came by finish, KO/TKO or submission. The denominator is wins, not fights.
58.3%
62.5%
46.5%
KO win rateShare of this fighter's WINS that came by KO or TKO. The denominator is wins, not fights.
25.0%
43.8%
27.9%
sub win rateShare of this fighter's WINS that came by submission. The denominator is wins, not fights.
33.3%
18.8%
18.7%
been stopped rateShare of this fighter's LOSSES that ended by finish — the denominator is LOSSES, not fights. A fighter who is 20-1 whose single loss was a KO therefore reads 100%.
0.0%
100.0%
52.3%
distance strike shareShare of this fighter's landed significant strikes thrown at distance. Distance, clinch and ground sum to 100%.
29.4%
68.9%
—
clinch strike shareShare of this fighter's landed significant strikes thrown in the clinch. Distance, clinch and ground sum to 100%.
17.6%
3.6%
—
ground strike shareShare of this fighter's landed significant strikes thrown on the ground. Distance, clinch and ground sum to 100%.
52.9%
27.6%
—
avg fight duration (m:ss)Mean length of this fighter's UFC fights, in minutes and seconds. "—" means we hold no round-time data for them.
10:34
11:21
—
Elo ratingA UFC-only Elo rating: it rises and falls with results, weighted by the opponent's rating. 1500 is the starting and dataset-average value.
1,554.6
1,677.7
1500
PageRankHow central this fighter is in the network of who-fought-whom. It reflects the quality of opposition they have shared a cage with, not their skill.
0.248
0.570
0.069
career stageForm over the last 4 fights before this one: 3 or more wins reads Peak, 2 or fewer reads Decline, and fewer than 4 prior fights reads Early.
Peak
Peak
—
Market IntelClick here to expandHow the betting market prices this fight compared to our model's number.
Betting Edge
Fighter
Win confidence
DraftKings odds
Break-even
Edge
Kayla Harrison
63.9%
1.49
67.1%
-4.8%
Amanda Nunes
36.1%
2.70
37.0%
-2.5%
Odds are the latest we have from DraftKings. Edge = win confidence × decimal odds − 1, so a positive edge means the price pays more than our model thinks the fighter is worth.
Break-even = 1 ÷ decimal odds — the win rate you would need at that price just to come out level. It is the raw price, with the bookmaker's margin still in it, so the two fighters' break-evens add up to more than 100%: that surplus is the book's cut, not a mistake.
Recommended stake4%of bankrollon Kayla Harrison at 1.51 (DraftKings)* Model output, not betting advice. Bet responsibly.
