Alexander Hernandez beat Austin Hubbard at UFC 307 on 5 October 2024, but the split-decision scorecards offered a more complicated story than the result. Two judges scored it 29–28 for Hernandez; the third awarded Hubbard a 30–27 victory. Hernandez entered the lightweight bout at No. 97 in the computerized rankings, with Hubbard at No. 143. That substantial gap made a close victory less impressive than another win on the record might suggest.
A ranking is not a count of raised hands. The question is how much a result changes the estimate of a fighter’s ability, given the opposition and the expected outcome. A narrow win can offer less evidence than supporters assume. In some circumstances, the defeated fighter provides the more surprising performance.
When the favourite wins, but proves little
For a favourite, victory is already part of the expectation. Winning decisively against a respected opponent reinforces a high rating; scraping past someone placed far below can do much less. Hernandez’s bout is a useful case because the result did not erase the difference between the fighters’ pre-fight standing. Hubbard lost officially, yet competing closely with a higher-ranked lightweight complicated the idea that he belonged so much farther down.
CIRRS, the Combat Intelli-Rating and Ranking System, accounts for opposition quality, recent results and other considerations beyond the latest win-loss entry. Its principles allow a narrow winner to lose rating points against substantially weaker opposition, while the defeated fighter gains. That is not a claim that every split decision triggers a penalty. It is a recognition that performances carry information about the level of both competitors.
What a split decision actually tells us
A split decision means two judges chose the winner and one chose the other fighter. In a majority decision, two favour the winner and the third scores a draw. A unanimous decision means all three agreed, although the scorecards can still show a narrow fight. Against Hubbard, Hernandez needed two 29–28 cards to prevail, while the dissenting judge gave him no rounds. The label alone hides that disagreement.
The distinction matters differently across rating models. Modified Elo gives a split-decision winner a match impact of 0.667, a majority-decision winner 0.833 and other winners 1.0. Glicko-1 instead uses 0.55, 0.61 and 0.91 for split, majority and unanimous decisions, respectively. Standard Elo treats every win as 1.0. Those figures belong to separate models; they must not be mistaken for CIRRS coefficients.
Whole-History Rating also considers information about an opponent that emerges after the fight, and it uses valid scorecards to assess decisions. The distinctions belong to a longer history of rating ideas that moved from chess to combat sports. A method calibrated for prediction may not reward the same result exactly as a system designed to describe current competitive standing.
The opponent can change the verdict’s value
The models had already disagreed before Hernandez and Hubbard touched gloves. The pre-fight program favoured Hubbard under standard Elo, modified Elo and Whole-History Rating, while Glicko-1 gave Hernandez a narrow edge. That is a sharper warning against reading one place in one table as a universal verdict. Different estimates of starting strength make even an identical split decision mean something different.
The same distinction exists in simpler sports records. A football score on rubiscore.com records the outcome, but it says little on its own about whether the favourite met expectations. MMA’s judges’ cards and stoppage methods supply more detail, yet they still cannot replace opposition quality. Losing narrowly to a proven contender can tell us more about a prospect than easily beating someone with a thin record.
A finish needs context too. Knocking out an established contender can strengthen a rating more than stopping a newcomer. An injury-related TKO counts as a TKO under CIRRS, even if observers debate what it demonstrated about the fighters. Neither the method nor the margin should be treated as a universal shortcut to assessing the quality of a performance.
A place on the list is not a points total
A fighter can gain rating points without climbing a place if the nearest rivals also improve. Another can move upward because someone else falls. The ranking disagreements across models are informative precisely because they reveal the assumptions behind each estimate. A one-place change is the visible outcome of a calculation, not a complete account of what a bout contributed.
Time introduces further complications. Current ranking eligibility requires a professional bout within the preceding 450 days, and inactivity penalties can reduce points. Changing division can also affect the rating, as can revisions to older fight records. Comparing an old result with a present-day list therefore risks crediting or blaming that one bout for changes caused elsewhere in the division.
Hernandez’s victory is not in doubt. What remains open to interpretation is how much it proved against Hubbard. The rankings become more useful once the opponent’s level, the scorecards and the gap between expected and actual performance are read together. One official win can be a modest competitive achievement, and one official loss can still reveal an improving fighter.