How the rankings and predictions are computed
I built this ranking system because I was frustrated with subjective, media-driven UFC rankings. No opinions. No popularity contests. Just math, applied consistently to every fighter. Here's exactly how it works.
The "Best Right Now" Philosophy
The rankings answer one question: "Who is performing best right now?"
A rolling 36-month window supplies the record, activity and strength-of-schedule terms of the score, and a fighter must have fought in the last 18 months to be ranked at all — retired and long-injured fighters drop off. The rating itself is earned over a whole career, so a champion does not start from zero at the window boundary, but only recent results decide the rest.
Glicko Rating System
Every UFC bout in the database, oldest first, updates both fighters' rating and rating deviation. Nothing is re-seeded at the window boundary: a fighter enters the 36-month window carrying the rating their whole record earned.
Every rating carries an uncertainty. A debutant starts wide (350), it narrows with each bout, and it widens again with time away. A short unbeaten run cannot outrank a long proven record.
Rankings use rating minus two deviations. Beating a higher-rated opponent moves you more than beating a lower-rated one; that is built into the update, not bolted on.
The average rating of your last 36 months of opponents, as they were going into each bout, is a separate term of the score. Quality of competition matters.
Final Score Calculation
Four terms, weighted. The conservative rating carries most of the score.
Data Sources & Transparency
- Win/loss results from official UFC fights
- Fight dates and methods (KO, submission, decision)
- Opponent identity for strength of schedule
- Division and weight class assignments
- Media votes or journalist opinions
- Social media following or popularity
- PPV sales or promotional push
- Subjective "eye test" assessments
All fight data comes from public fight records. See the data sources page for our data policy.
Fight Predictions
The predictions are separate from the rankings. They come from a statistical model that looks at each fighter's public fight record and estimates a win probability for the matchup. It is a model, not a tipster, and it is early-stage.
- Recent form and activity: results and time since the last fight
- Finishing rate: how often wins come by KO/TKO or submission
- Striking and grappling volume from recorded fight statistics
- Division experience: how established each fighter is at this weight
- Each number is the model's estimate of a fighter's chance to win, fitted on past UFC results
- A 60% favourite is expected to lose about four times in ten. That is the model working, not failing
- When the two fighters come out even, the page shows a toss-up rather than picking one
- Late injuries, weight-cut problems, and judging are outside what the model can see
No live accuracy figure has been published for this model, and until one is you should treat its reads as one input among many, not as advice. Results on past cards are being recorded and a track record will be published here once there is enough of one to be meaningful.
FightIndex.io is not a sportsbook and nothing here is betting advice. See the responsible gaming page.
Built by Moetive. I'm building FightIndex.io solo because the MMA community deserves better than popularity-based rankings. This methodology will continue to evolve — I'm always looking for ways to make the system more accurate and fair. Follow @moetive_ · Support the project
Check out the rankings and see how objective, data-driven analysis compares to the official UFC rankings.