Part of the page failed; a hard reload (Ctrl+F5) usually fixes it.
MMA Career Rankings
Rank fighters by career (All-time), best ten years (Prime), or recent form (Current). One model rates every recorded professional MMA fight, from every promotion. All-time and Prime rank careers from the UFC and the other major promotions, so a fighter can rank without a UFC fight. Current and weight-class boards use UFC bouts.
How to read the scores
How All-time is calculated
The score combines championships (30%), career skill (17.5%) and contender wins (52.5%). Title and contender wins are weighted by opponent quality. Career skill measures active years above the division's contender level.
Each part is divided by the average of its own 100 highest positive values, then square-rooted and weighted. The total is multiplied by 1,000. The square root limits how much one extreme part can dominate. These weights are chosen scoring rules. A score is a points total, not a percentage advantage over another fighter.
Why a rating can fall after a win
The career chart shows estimated skill at each bout, not accumulated points. The model fits whole careers, so later results can revise earlier estimates.
The chart also adjusts each year against that year's contender baseline. If the baseline rises faster than a fighter's raw rating, the plotted line falls even after a win. This happens in GSP's career. A downward step does not necessarily mean the win hurt him.
The top-100 normalization applies to the All-time score, not this line. Between releases, a retired fighter's score can fall as the comparison pool improves or the model revises past ratings.
Rankings
The published boards. A weight class keeps each board's order and shows only the fighters whose UFC career was mostly fought in that class; a fighter who moved divisions stays in the class of most of their fights. A blank Prime rank means the fighter did not qualify. Click a name to add it to Fighters.
Score breakdown
1000 × (a·√championships + (1 − a)·(s·√career skill + (1 − s)·√contender résumé)), each
part scaled by its own top-100 mean before the square root, so a career is paid for breadth more than for one
extreme part. Ratings are read against their own era. Move the rates to re-rank.
Score by part
The math, in plain terms
Whole-History Rating
A chess-style rating fitted to every career at once (Coulom, 2008). The gap between two ratings gives each fighter's chance of winning. Because nothing is fixed one fight at a time, a win counts for more when the opponent later proves to be great, and fighters from different decades share one scale.
A neutral starting bout
Each fighter starts with the weight of one imaginary fight against an average opponent, half won and half lost. Without it, a short unbeaten record has no upper limit and could top the board.
Graded results
A finish counts as a full win; decisions and disqualifications count slightly less, and a draw counts as half. A knockout says more about who was better than a split decision does.
A contender line for each year
Every rating is read against its division's fifth-best fighter that year. The rating scale drifts upward over time, so this compares a 2007 win and a 2023 win fairly.
Win probability raised to a power
A win is valued by the chance the beaten opponent would have beaten a contender: to the fourth power for a title win, the third power for other contender wins. One elite win is worth about five borderline ones, without letting a single win carry a career. The third power was chosen after testing powers 1 to 6.
Promotion weights, pulled toward average
The rating never looks at the promotion. Afterwards, skill and contender-win points are scaled by where they were earned, between 0.20 and 1.00 (UFC 1.00). A career with few promotion-labelled fights is pulled toward the average, so a handful of fights cannot decide it.
Square roots on a common scale
Titles, contender wins and skill are each measured against their top-100 average, square-rooted, then weighted. A career strong in all three beats one huge number in a single part.
Best ten-year window
Prime finds the ten-year stretch with the most wins over contenders and needs at least four. Each win beyond that adds to the score. A ten-year window keeps a peak from being diluted by a long career before or after it.
The model has not yet been shown to predict fights better than a simpler rating. These are measured judgements of careers, not betting odds.
Fighter comparisons
Estimated skill at each bout, up to six fighters: each year is shifted by how far its contender line sat from the average year's, so careers in different eras share one scale. Later bouts revise earlier ratings. This is a hindsight estimate, not a running points total; it can fall after a win.
Estimated skill at each bout, adjusted for era
Career scores by year
Career mass adds, for each active year, how far the fighter's average rating stood above that year's bar; no year subtracts.
Credit by year
Annual rating changes
Biggest rises and falls
UFC win streaks
Unbroken UFC win runs; a draw or loss ends one. Click a row to mark it.
Length against opposition
Biggest upsets
UFC wins by the fighter the betting market gave the lowest chance, from no-vig odds.
Winner's chance before the fight
Weight-class strength
UFC bouts only. These chips and years apply to this section.
Class strength by year
Top five per class
How fights end
Title reigns
Title history
Dominance in wins
Strike differential, control time and submission attempts per win. Not a rating input.
Mean dominance per win
Age and layoffs against the odds
UFC bouts with betting odds. Each fighter's win rate against the market's no-vig price for them. A gap means the market did not fully price age or time away.
Win rate and market price by age
Win rate and market price by time since the last fight
Integrity policy
Judgement-based deductions for wins with a confirmed PED violation, a disqualification or a missed weight. These do not affect the published boards.
Rank before and after the debits
Forecast accuracy
Each bout's pre-fight win probability against how often those fighters won. Only earlier bouts inform a forecast. The betting market is the benchmark a fight forecast has to beat.