A Missing Scorecard Is an Unknown Round, Not a Lost One

A blank “rounds won” field does not mean the fighter lost those rounds. The official fight result and the judges’ round-by-round ballots are separate records: the winner may be known even when the underlying scorecards are unavailable, incomplete, or irrelevant to the result because the bout ended by stoppage.

A defensible rounds-won statistic counts only documented judging data. It must also identify whether it measures physical rounds or individual judge ballots and classify everything else as unknown, incomplete, unjudged, not contested, or excluded.

The short answer: missing does not mean lost

An official result records how a bout ended—by decision, knockout, submission, disqualification, no contest, or another outcome. It does not necessarily make every underlying round score publicly available.

That distinction is especially important for older fights and stoppages. Some older completed bouts have no accessible scorecard in a particular archive. In a fight that ends by knockout during round two, judges may have scored the completed first round, but those ballots might not have been published or preserved in the dataset. The unfinished second round should be treated as unjudged unless an official partial-round score exists, while round three was never contested.

A community comparison of UFC points and rounds won illustrates how unavailable older cards and early finishes can change the amount of data eligible for a calculation. Its fan-produced totals are not official records, but the underlying data problem is valid.

Keep these statuses separate:

• Documented: Sufficient official ballots are available under the declared method.
• Unknown: The round occurred, but its ballots are unavailable.
• Incomplete: Some, but not all, expected ballots are available.
• Unjudged: No official score should be assumed, including for an unfinished round unless a partial-round score is documented.
• Not contested: The scheduled round never occurred.
• Excluded: The data exists but falls outside a stated methodological rule.

The UFC official scorecard index links to event-level scorecard pages, particularly for recent events. Its visible listings do not establish complete historical coverage. Failing to find an older card there does not prove that the card never existed or cannot be obtained from another official source.

Define what “rounds won” means before counting

“Rounds won” is a derived statistic, not a universally defined UFC record category. Any calculation must first specify its unit.

• A physical round is the single period contested by the two fighters.
• A judge ballot is one judge’s score and selection for that physical round.

Three judges can therefore produce three ballot observations for one physical round. Those ballots are not three separate fight rounds. Under the usual 10-point must structure, judges score rounds independently and their round scores are combined into final cards, subject to the rules adopted in the event’s jurisdiction.

Two aggregation methods are useful, but they answer different questions.

Majority-won physical rounds

Award one physical round to Fighter A when at least two available judges select Fighter A under the methodology’s minimum-ballot rule. With a complete three-judge card, a 2–1 split and a 3–0 agreement each produce one majority-won round for A.

This is an operational definition for analysis. It represents the majority view of the assigned judges, not proof of an objectively correct round winner.

Individual judge-ballot wins

Count each judge’s selection separately. If all three judges select Fighter A, A receives three ballot wins. If two select A and one selects B, A receives two ballot wins and B receives one.

The denominator can reach three ballots for every judged physical round. Unlike the majority-round method, this approach preserves judge disagreement.

The ABC’s July 2025 scoring clarification describes judging as a round-level evaluation of effective offense and impact, rather than a mechanical count of actions. A rounds-won field simplifies those scorecard decisions further, so its label must be precise.

“12 rounds won” is ambiguous. “12 majority-won physical rounds” and “32 judge-ballot wins” are auditable—and are not interchangeable.

Worked examples: how the two methods change the total

The same judging record can produce different totals depending on the chosen unit.

Scenario and available ballots	Majority-round result	Ballot result	Data status
A, A, B	One physical-round win for A	A wins 2; B wins 1	Complete
A 10-8, A 10-9, A 10-9	One physical-round win for A	A wins 3	Complete
Round-one ballots unavailable; KO during round two	No winner assigned	No ballot wins assigned	R1 unknown; R2 unjudged unless officially scored; R3 not contested
Only two ballots available: A, B	Apply the declared partial-card rule	A wins 1; B wins 1 among available ballots	Incomplete

In the second example, the 10-8 changes the point margin but not the binary rounds-won total. Fighter A still receives one majority-won physical round or three ballot wins. An analysis of dominance should preserve score margins in a separate field rather than asking a binary win variable to represent both winner and degree.

Incomplete cards require a rule chosen before totals are calculated. An analyst might:

• exclude every physical round without all three ballots;
• retain a round when two available ballots agree;
• require at least two ballots but classify a one-each split as tied or unresolved; or
• include every available ballot in the ballot metric while excluding the physical round from the majority metric.

Any of these rules can be implemented consistently. None should be left implicit.

Tied and unusual scores also need an explicit policy:

• A 10-10 ballot should be recorded as a tie, not a win.
• If tied ballots remain in the ballot denominator, they contribute zero ballot wins and must be reported separately.
• A physical round without a qualifying majority should be classified as tied or unresolved. The methodology must say whether it remains in the percentage denominator.
• A point deduction can alter the official numerical score without establishing that the other fighter won the underlying action. Store the point-adjusted result separately from any analytical interpretation.
• Drawn rounds, partial-round scores, and scores outside the chosen scope should be flagged rather than silently forced into an ordinary win-loss category.

Why scorecard datasets lose rounds

A blank value can arise at several stages:

1. Unfinished round: The fight ends during the round, and no official partial-round score is documented.
2. Unavailable completed-round ballot: A completed round may have been scored, but its ballots were not published or are not accessible.
3. Incomplete card: One or more expected judge entries are missing.
4. Archive-link gap: A card may exist without appearing in the archive being searched.
5. Failed parsing: A scan or image cannot be read reliably.
6. Failed fight matching: Databases use names, dates, or identifiers that do not join cleanly.
7. Methodological exclusion: The analyst deliberately removes ties, deductions, unusual scores, partial cards, or records outside the study’s scope.
8. Not-contested round: The bout ended before a later scheduled round began.

A non-decision result does not itself prove that judges produced no scores for completed earlier rounds. It establishes only that the fight was not decided by the final scorecards. Publication and availability must be checked separately.

These categories have different meanings. Converting all of them to zero turns uncertainty, technical failure, and deliberate filtering into apparent losses.

The scale of analytical attrition can be substantial. The third-party JudgeAI methodology began with 2,125 UFC decision fights from 2011 through 2020 and retained 1,670 fights containing 5,238 rounds after merging judging data and applying filters. Its exclusions included unavailable judging data and unusual scores.

That retained share is not a UFC archive-completeness rate. It describes one analytical pipeline with its own scope, data joins, and exclusions. The project also treated the judges’ majority as a modeling proxy rather than objective truth.

A separate methodology could require three complete judges’ cards for a particular calculation. Such a rule would change the sample without identifying why each omitted record was incomplete. The exclusion count and the underlying missingness reasons should therefore be reported separately.

Missingness can distort comparisons. If older eras, certain jurisdictions, or particular fighters have poorer scorecard availability, they may appear to have fewer rounds won simply because fewer rounds are documented. They may also disappear from rankings that require complete cards. Without a completeness measure, archival coverage can be mistaken for performance.

How to verify a missing scorecard

Use a bout-specific workflow rather than treating the first blank field as definitive:

1. Identify the exact bout. Record the event, date, fighters, official result, scheduled length, finish round, and finish time.
2. Check the UFC event scorecard page. Search the official index and the event’s scorecard article, while remembering that the visible index does not prove complete historical coverage.
3. Identify the governing athletic commission or regulatory body. Rules and procedures vary by jurisdiction, so establish which body regulated the event.
4. Locate the most authoritative available card. Prefer a commission-published or official event card over a third-party transcription.
5. Compare any transcription with the image. Check fighter order, judge names, round columns, point deductions, and totals.
6. Label secondary material. A transcription is not the original card, and a model estimate is not a submitted ballot.
7. Use fight statistics only as context. They can help verify the bout and describe activity, but they cannot supply a missing judge selection.

Athletic commissions generally appoint officials and control official results in their jurisdictions. The adopted rules and review procedures can differ, and ordinary disagreement with a judge’s assessment is distinct from a clerical, arithmetic, or rules error, as summarized in this MMA scoring and commission guide.

Do not reverse-engineer a ballot from strikes, takedowns, knockdowns, or control time. Public statistics do not reproduce a judge’s qualitative, round-level evaluation of effective offense and impact.

The UFC Record Book is not a substitute for scorecards. It presents selected UFC record categories rather than comprehensive round-by-round ballots, so it cannot validate a fan-created rounds-won total. Any separate historical-coverage statement should be verified directly on the live record-book page before publication.

A defensible reporting template for rounds won

Build the statistic from a bout-level ledger instead of maintaining only one cumulative field. Each record should contain:

• Event and date
• Opponent
• Official result
• Scheduled rounds
• Rounds completed
• Finish round and time
• Available ballots by judge and round
• Direct scorecard source
• Inclusion status
• Missingness reason
• Notes for ties, deductions, partial rounds, or unusual scores

Then publish the calculation rules.

Majority-round percentage

Majority-won physical rounds ÷ eligible documented physical rounds with the minimum number of ballots required by the declared rule

Judge-ballot percentage

Judge-ballot wins ÷ eligible documented judge ballots

The methodology must also state:

• the minimum number of ballots required for an incomplete physical round;
• whether a tied physical round remains in the majority-round denominator;
• whether 10-10 ballots remain in the ballot denominator with zero wins;
• how point-deduction ties and partial-round scores are stored; and
• which unusual scores are excluded.

For example, report “12 majority rounds won from 20 eligible documented judged rounds.” Do not divide by every scheduled round. Rounds that never happened are not losses, and unavailable ballots are not documented judge selections.

Maintain separate counts for:

• Documented rounds
• Unavailable cards
• Incomplete cards
• Unjudged rounds
• Rounds not contested
• Unmatched records
• Tied or unresolved rounds
• Methodological exclusions

When space permits, publish both majority-round and ballot-win figures alongside a scorecard-completeness rate. Define the completeness denominator according to the declared scope, excluding rounds that were unjudged or never contested. This allows readers to see whether a majority-round total hides frequent split ballots.

Date cumulative fighter statistics because new bouts, recovered cards, and corrections can change both numerator and denominator. If unavailable cards could materially reorder a ranking, publish a defensible range or withhold the ranking rather than presenting false precision.

Finally, define the competitive scope. A UFC-only calculation describes the selected UFC sample. It should not be labeled as the fighter’s complete professional MMA career unless the rest of that career has been collected and processed under the same rules.

Can UFCStats strikes, takedowns, or control time fill in a missing scorecard?

No. Those statistics can describe activity and help confirm that the correct bout and round have been matched, but they cannot recreate an official judge’s ballot.

Judging is not a fixed formula that awards rounds from strike, takedown, or control totals. A statistical model can estimate a likely round winner, but its output must remain labeled as an estimate rather than being inserted as an official score.

Who can verify or correct an official UFC scorecard?

Start with the governing athletic commission or regulatory body for the event. Jurisdiction-specific procedures differ, but the commission generally appoints the judges and controls the official result. UFC-published event scorecards may provide an image or transcription, while the relevant regulator may be able to address an apparent arithmetic, clerical, or rules-related discrepancy under its procedures.

A third-party database can correct its own transcription but cannot rewrite an official ballot. Ordinary disagreement with a judge’s evaluation does not by itself establish that the card is missing or contains a correctable recordkeeping error.

The audit rule is simple: count only what the cards document, name the unit being counted, and label everything else. A trustworthy rounds-won figure identifies whether it measures majority-won physical rounds or individual judge-ballot wins, shows its denominator, discloses missingness by reason, and never turns an absent scorecard into a loss.