UFC main card, where the price looks wrong
Fight markets price narrative more than any other sport, because there is no league table to argue with and the sample per fighter is tiny.
MMA is the hardest sport on this board to model honestly. A fighter's entire professional record might be twenty contests spread over a decade, against opponents whose quality varies more than in any league, with rule sets and weight classes that changed along the way.
What that does to a number
Small samples produce wide error bars, and wide error bars mean that most disagreements between a model and a price are not real. In football a two point gap can be meaningful. In MMA, two points is inside the noise on almost any estimate, and treating it as a finding is how a model manufactures bets it should not take.
So the bar here is deliberately higher: only gaps large enough to survive the uncertainty are published at all, and on most cards that is one fight or none.
This card
One qualifies. A favourite at 1.36 where our number is nearer 1.60. The
market appears to be pricing a five fight winning run; the research half found
that three of those five were against opponents who have since lost four in a
row between them, and that the fifth was a decision several outlets scored the
other way.
That is not a strong claim. It is a claim that the record overstates the run,
and at 1.36 the price has no room for the record to be overstated.
Where narrative enters
There is no league table in fighting, so the public record is mostly commentary, and commentary is written to be read. A fighter with a compelling story is priced shorter than one without, and that is the closest thing this sport has to a systematic market bias. It is also very hard to measure, which is why we do not stake it heavily.
The general discipline is the one in how to bet on football with a model: the number is inert until it disagrees with a price by more than its own error.
What this does not tell you
Fights are decided by single moments in a way league sports are not. A model that is right about who is better is frequently wrong about who wins, and no amount of process fixes that.
Common questions
Why are MMA odds harder to model than team sports?
Because a fighter's professional record might be twenty fights across ten years, against opponents of wildly varying quality. That is a tiny sample by any modelling standard.
Does styles make fights hold up statistically?
The matchup effects are real but much smaller than commentary suggests, and they are very hard to estimate from a handful of fights against different opponents.
Why do MMA prices move so far on fight week?
Because weight cut news, a missed weigh-in or a visible decline in condition are genuine information arriving very late, and there is little liquidity holding the line still.