Tell 02
The Odometer
Everyone reads high mileage as a countdown. We checked which way it counts.
30% less likely to be stopped. The heavy odometer belongs to the harder chin.
- Sample
- Every UFC bout since 2005, 8,799 of them
- Confidence
- 19% to 39% less likely
- The catch
- UFC miles only
Last run after UFC Fight Night: Gamrot vs. Salkilld, Aug 21, 2026.
What everyone believes
Max Holloway has eaten 1,559 recorded head strikes across 33 UFC fights, more than anyone in the promotion's history. The whole sport watched that number climb and called it borrowed time. The borrowed time ran 29 fights before Ilia Topuria became the first to stop Holloway, in October 2024.
Cornermen talk about miles on the chassis. The betting chat docks a fighter for every war on the record. It sounds like physics, and it smuggles in a selection problem: to show up deep into a career with heavy miles and no KO on the card, a fighter had to walk through every one of those strikes still standing. The odometer you can read has already been proofread.
So the fair test is narrow. Two fighters, same age, same division, same era, neither ever stopped by strikes, one carrying a heavy odometer and one a light one. Which chin goes first?
The comparison
Share of fights that end in a first KO or TKO loss from strikes.
Light mileage
15.6%
Heavy mileage
11%
-30% more likely-39% to -19%
- Age
- Light mileage29.6
- Heavy mileage29.6
- UFC fights
- Light mileage0.6
- Heavy mileage7.2
- Head strikes eaten
- Light mileage4
- Heavy mileage188
- Fought around
- Light mileageJun 2018
- Heavy mileageJul 2018
Striking differential, per 15 minutes
Light mileage
0
Heavy mileage
2.5
The heavy odometer also outlands its opponents by 2.5 strikes more per 15. The miles came with craft.
What we measured
Every UFC fight since 2005: 8,799 bouts across 783 events. The pool is 10,810 fighter-fights from 2,419 fighters never yet stopped by strikes, because our chin decay study owns everything after a first KO and the two findings had to stay out of each other's samples. Here the outcome is the first stop itself.
Four odometer readings, tested one at a time and never blended: significant strikes absorbed, head strikes absorbed, knockdowns absorbed, and plain cage seconds, each as of the night before the fight. Mileage bands are drawn against fighters the same age, so nobody gets compared to a prospect a decade younger.
Every damage reading came out safer, and none of them beat plain cage time. At the same cage seconds, extra head strikes still read 24% less likely to be stopped. The one reading that leans dangerous is knockdowns already absorbed, 17% more likely at the same cage time, and its interval crosses no difference. Whatever the odometer is measuring, tenure fits the numbers better than damage.
The shape of the odometer
Each row is a mileage band for a fighter's age, against the lightest fifth. The line at 1 is no difference, and left of it is safer.
Second fifth
1,705 fights
Middle fifth
2,113 fights
Fourth fifth
2,125 fights
Heaviest fifth
2,131 fights
Why the needle runs backwards
The whole effect lives in the first step. A fighter with almost no UFC miles is the likeliest first KO on the card, one band of real history later the risk has already fallen through the floor, and from there the line is flat. Chin decay found a one-time step down after the first KO. This is the same picture in a mirror: a one-time step up for being new, and nothing that compounds.
Put the two 33 year olds side by side. One has eaten 8 head strikes in the UFC, the other 260. The folk theory backs the fresh chin. The fresh chin takes a first KO in 17.1% of fights and the heavy one in 7.9%, which is 54% less likely. Throw the newcomers out entirely and race the heaviest fifth against the second fifth, and the gap shrinks to 8% less likely with an interval that crosses no difference. The miles themselves add nothing we can measure in either direction. What they mostly do is certify who survived them.
What the odometer does predict
Out of every 100 fighters at their first heavy mileage fight, how many never fought three more times in the UFC.
20Light mileage
33Heavy mileage, same age
Count it again: the fresh chin is the one that breaks.
What this can't tell you
The odometer here only counts UFC miles. A fighter who arrives at the same age with a light UFC record often spent a decade getting hit somewhere the database can't see, which flatters exactly the group that looks safest in this design. It can't be the whole story though: rerun the test on fighters who debuted young enough that the UFC saw nearly the whole career, and the heavy odometer comes out 43% less likely, a bigger gap than the headline.
This design also can't split two explanations. Maybe absorbed punches genuinely don't set up the first stop. Or maybe they do, and the sport removes fragile chins so efficiently that the survivors' numbers hide it. Both readings make the same prediction for the next card, which is what a Tell is for. They disagree about why, and the data stays silent on why.
Inside a single career the strikes don't look like rust either. Hold age still and a fighter's own striking differential rises 3.45 per 15 minutes for every hundred head strikes added to the odometer, among fighters who lasted five or more fights, which is its own survivor filter. And with previously stopped fighters included in the pool, heavy mileage still runs 16% less likely.
Since we published
Nobody new has fallen into this one yet. Check back after the next card.
The fine print
Sample
UFC fights 2005-01-01 to 2026-08-08. Primary pool: 10810 never-yet-KOed fighter-fights from 2419 fighters. Primary contrast: 3540 heavy-mileage fighter-fights vs 3299 matched light-mileage fighter-fights (Kish effective N 1996) across 130 strata; 19 pruned off common support. Attrition cohort (24mo window): 686 fighters at their first heavy-mileage fight.
Method
Coarsened exact matching within the never-yet-KOed risk set: fighter-fights in the heaviest vs lightest tertile of the component, tertiles computed WITHIN each 2yr age band (pool-wide bands have no common support at fixed age), strata = age band x weight class group x era, control weights treatedN_s/controlN_s, all covariates as-of pre-fight from dcc.asof_fighter_stats. Not stratified on UFC fight count: fight count is itself mileage. Four components tested separately, never composited; ranking read from the contrast WITHIN cage-time strata plus head share at fixed sig absorbed. Outcomes: first KO/TKO-by-strikes loss rate per fight, and striking differential per 15 minutes (ratio of sums; per-fight-mean and decision-only sensitivities). Discrete-time hazard logit with age-band/weight/era dummies and fighter-clustered sandwich SEs as the continuous complement. Fighter-cluster bootstrap, 2000 replicates with bands and matching recomputed per replicate, percentile 95% CIs. Attrition matched additionally on fight-count bin and index-fight result, 24-month edge exclusion. Pre-committed sensitivity: UFC debut at 25 or younger. Runner: npm run analyze:mileage-index.
Assumptions
All mileage is UFC-recorded mileage, a lower bound on true mileage: regional and amateur damage is invisible, which misclassifies low-mileage arms at fixed age (late arrivals) and biases toward null. Reaching heavy mileage without a KO selects durable fighters, which also biases toward null; a null therefore cannot separate "no damage effect" from "selection offsets it". Per-fight rates in a first-event pool condition on having survived un-KOed to that fight; the hazard logit handles that structure exactly, the band rates approximate it. High mileage marks risk rather than being shown to cause it. Chin event = KO/TKO loss by strikes only, same classifier as the chin decay study (audit table scripts/analysis/chin-decay-method-map.json). Fighters are treated as independent sampling units although they fight each other.
Every mile on the odometer is a punch that didn't end the fight. That's the record it keeps.