ABOUT US

WHAT IS CIR?

CIR stands for Cage Index Rating. It brings several parts of a fighter’s record together into one score. A higher score means a stronger standing under our model at that point in time.

Think of it as a report card for a fighter’s UFC results. Two fighters can have the same number of wins but different scores because opponent strength, result reliability, championship results and time decay all matter.

SCORE VS. RANK

The CIR score is the rating. The P4P rank is the fighter’s position when eligible fighters are sorted by that rating across weight classes. A score of 90 does not mean a 90% chance of winning, and a higher-ranked flyweight is not being predicted to beat a heavyweight.

WHY CAN A SCORE CHANGE WITHOUT A FIGHT?

Time passing decays ability evidence and championship support. Other fighters’ results can move a rank even when a score stays the same. Opponent strength is frozen at the time of each fight: later results do not rewrite the value of an earlier opponent.

A simple example: imagine two fighters each have three wins. One beat established, high-rated opponents recently; the other beat lower-rated opponents several years ago. CIR looks at those differences instead of treating both records as identical. This is an illustration, not a fixed points award.

WHO QUALIFIES?

To receive a published rank, a fighter needs at least two UFC fights and a UFC appearance within the previous 18 months, and roster eligibility. “Not qualified” means they do not meet that gate at the selected date—not that they lack talent.

HOW TO READ RANK MOVEMENT

Current rankings compare against the pre-card baseline, with both dates shown above the table. Historical views use their own dated comparisons. ▲ means the fighter moved up in rank. ▼ means they moved down. The number is how many places they moved, not how many CIR points they gained or lost. A zero means no rank change; “N/A” means the fighter is newly qualified or a verified comparison is unavailable.

HOW THE HISTORICAL CALCULATOR WORKS

Choose a fighter and an available snapshot to explore their CIR and ranking at that time. The historical view is designed to use results available by that date, not later achievements. The line chart connects the available snapshots; it does not represent a new calculation for every day between them.

HOW CIR v2.8 CALCULATES A SCORE

CIR starts with a full-range opponent-relative ability rating. Championship support adds a limited bonus; no part of the rating scale is reserved for title winners.

Ability starts at 50 and changes with the surprise of the result against the opponent’s frozen pre-fight ability. Unanimous decisions and ordinary finishes receive the same result weight. A split decision receives one-third weight, and a majority decision two-thirds, reflecting agreement among the judges. These are design weights, not calibrated win probabilities. A unanimous decision can be close; the rule measures agreement about the winner, not dominance.

Each bout can at most double or halve ability odds, with smaller changes for expected outcomes. A source-labeled injury stoppage receives one-quarter weight on both sides; a disqualification receives half weight. The official result remains unchanged. An injury label does not establish accidental causation, and missing injury information is not inferred.

Judge totals are oriented only when three scores agree with the recorded decision type. The parser preserves dissenting judges in split decisions. Incomplete or contradictory totals do not supply dominance evidence. Performance report-card bars describe the record; they add no separate finishing bonus to CIR.

Competitive championship support starts at zero in each division. A title win moves that division’s support toward 100; a title loss moves it toward zero. The fraction updated is 1 − exp(−|e|), where e is the same signed evidence used for ability. Unexpected wins earn more support, while divided decisions and losses to stronger opponents change it less. Interim and undisputed title bouts use the same rule. Ordinary bouts, BMF titles, tournament titles, promotions, draws and no-contests do not update support at the bout. Every historical title result remains in the ledger.

Division records combine with diminishing additional credit: C = 100 × [1 − product(1 − C_division / 100)]. Success in another division can add support, but the combined championship bonus stays within the same limit. An unsuccessful title challenge in another division still lowers ability, but does not remove support earned in the original division. All division support continues to age with time.

Let A be ability and C be competitive championship support. Multiply ability odds A / (100 − A) by 2^C/100, then convert back to a 0–100 base score. The title multiplier can never exceed two, and a fighter with no title support retains the full ability range. Ability and each division’s support age continuously at about 3% per year; the combined support is then recalculated.

CIR is the combined base score directly. There is no square-root display curve, percentile adjustment, target average, or achievement-tier calibration. A one-point difference in the combined score remains a one-point difference in published CIR.

The underlying evidence and championship calculations remain bounded and nonlinear; removing the display curve does not remove those model mechanics. Historical snapshots are restated under the same v2.8 calculation. Comparing a v2.7 score with a v2.8 score is not a measure of weekly performance.

Wins cannot lower the score at the time of the bout. Losses cannot increase it; title losses reduce support in the contested division, without removing support earned in other divisions. Draws and no-contests are neutral at the bout, while elapsed time can still reduce dated snapshots. All 8,709 recorded losses passed the non-increase check, including debut losses and every per-fighter ordering of same-day bouts.

The historical calculator shows ability, the championship bonus, and their combined CIR score. Other report-card bars summarize the record and are not extra scoring weights. Every historical snapshot is replayed with v2.8. Changes from an older model are formula changes, not new fight results; weekly movement compares two v2.8 snapshots.

READING THE SCALE

90–100: Exceptional. 80–under 90: Very strong. 70–under 80: Strong. Below 70, we show the score without an achievement label. These descriptions apply to model standing at the selected moment, not a fighter’s entire career, official title status, or likelihood of winning.

The score is calculated first; its description is applied afterward. Tier boundaries do not change scores. There is no required average or required number of fighters in a band.

Fighters become eligible after two UFC appearances, with the same 18-month activity window. A two-fight sample contains less evidence, and the historical result identifies it explicitly. Eligibility does not award bonus points. The eligible population is not a verified contracted roster.

WHAT CIR IS—AND ISN’T

CIR is a consistent way to compare recorded results. It is not an official UFC ranking, a guarantee of who wins next, or a complete measure of a fighter’s skill. Every model has assumptions and limitations, and reasonable fans can disagree with its conclusions.

The formula—not an AI opinion or a manual vote—determines the published order. The Cage Index is independent and is not affiliated with or endorsed by UFC.