PropBetEdge Fight DNA · proprietary fighter intelligence
Fight DNA. From fight records to fight intelligence.
Fight DNA transforms normalized event, bout and round-level fight records into proprietary metrics covering pace, opponent stance, striking geography, grappling efficiency, finishing patterns, round progression and matchup context. Every metric carries its supporting sample, provenance, confidence and definition version.
Raw data tells you what happened.Fight DNA describes the fighter the data reveals.
Fight DNA is PropBetEdge's proprietary analytics layer above the normalized UFC record. The inputs are ordinary source facts: an event date, a bout result, strikes landed and attempted, takedowns, control time, per-round rows from UFC Stats. The product is what PropBetEdge builds from them: a versioned, reproducible, as-of feature system that turns a fighter's event-dated history into fighter-specific and matchup-specific intelligence.
Every Fight DNA metric keeps its definition, sample size, provenance, as-of date, confidence tier, coverage state and definition version. That packet of context is part of the product, not a footnote.
In one line
Not simply republished source statistics. Derived by PropBetEdge from the normalized fight record, and shown with receipts.
02
What it is not
Not an official UFC statistic, and not affiliated with the UFC, Zuffa LLC, TKO Group, ESPN or any sportsbook.
Not a prediction. Fight DNA describes what the record shows; it does not claim a fighter will reproduce a number in the next fight.
Not a grade. There are no letter scores, gauges, percentiles or “elite” labels, because no benchmark in the system would justify them.
Not a career snapshot copied from a source profile. Career-to-date figures on a fighter row are display fields; Fight DNA is rebuilt from event-dated bout and round rows.
Not betting advice. PropBetEdge Pro sells intelligence, not guaranteed outcomes.
03
The data pipeline
The moat is the system, not any single stat. Six stages, each reproducible from the one before it.
01
Source record
events
bouts
results
round stats
Source
02
Normalized fight history
identity-matched fighters
event-dated bouts
per-round rows for both corners
source URL + capture time on every row
03
As-of reconstruction
only bouts before the date
one snapshot per event date
no future-fight contamination
04
Fight DNA feature system
stance
striking
grappling
finish
round
context
PBE Derived
05
Confidence + provenance
sample bouts · rounds · seconds
confidence tier per metric
definition version
source families
06
Matchup intelligence
paired comparisons
supported observations
counter-case warnings
04
Why as-of matters
Fight DNA can reconstruct a fighter using only the information that existed at a particular point in time. A snapshot dated the morning of a card includes every bout before that date and nothing after it, so a historical matchup read is never contaminated by fights that had not happened yet.
Example
A fighter has ten archived bouts. The snapshot for a fight in year three is built from the bouts before that date only. When you open an archived matchup page, the Fight DNA you see is the version that could have been known then, and the as-of date is printed beside it.
05
Why normalization matters
Round stats arrive as raw counts: 32 significant strikes landed in a round, 90 attempted, 1:45 of control. Counts are hard to compare across fights of different lengths, so Fight DNA normalizes them to observed time. Per-minute rates divide by the minutes actually covered by round stats; per-15 rates scale to the length of a standard three-round fight. A 90-second finish and a 25-minute decision become comparable, and the observed minutes are always shown.
Example
td_landed_per_15 = takedowns landed × 900 / observed seconds. Three takedowns across 20 observed minutes (1,200 seconds) is 2.25 per 15.
06
Why sample size matters
A fighter with one observed round should not be interpreted the same way as a fighter with many fully captured fights. A 100% finish rate from two wins is a fact about two fights, not a tendency. That is why every Fight DNA number is displayed with the bouts, rounds and minutes behind it, and why the DATA COVERAGE bar sits at the top of every profile.
10Completed bouts
11Rounds with stats
23 minObserved
Sep 7, 2026As of
v1Definition
MediumConfidence
Normalized recordsSources
Missing round coverage is never rendered as a zero. If a fighter's bouts have results but no round stats yet, the striking and grappling families are withheld until observations exist.
07
What confidence means
Confidence reflects the amount of supporting sample for this metric. It does not mean PropBetEdge is predicting the fighter will reproduce the number in the next fight.
Highlarger supporting sampleResult metrics: 8 or more bouts. Rate metrics: 5,400 or more observed seconds (90 minutes).
Mediumuseful but still limited sampleResult metrics: 3–7 bouts. Rate metrics: 1,800–5,399 observed seconds (30–89 minutes).
Lowsmall sample — interpret carefullyThe metric's minimum is met, but result metrics have fewer than 3 bouts or rate metrics fewer than 1,800 observed seconds (30 minutes).
Insufficientnot enough observed evidence to publish the metricBelow the metric's registry minimum (minimum bouts, rounds or seconds), or no denominator at all. The value is withheld rather than shown as zero.
Thresholds are the definition v1 rules recorded in the Fight DNA contract; they are applied by the builder, not by this page. Result metrics count bouts; rate metrics count observed seconds.
ConfidenceConfidenceConfidence reflects the amount of supporting sample for this metric. It does not mean PropBetEdge is predicting the fighter will reproduce the number in the next fight.Thresholds are the definition v1 rules recorded in the Fight DNA contract; they are applied by the builder, not by this page.Highlarger supporting sampleMediumuseful but still limited sampleLowsmall sample — interpret carefullyInsufficientnot enough observed evidence to publish the metricConfidence tiers explained →What does confidence mean?
08
Striking DNA
Pace, efficiency, defensive exposure, targets and where exchanges occur.
Significant strikes
UFC Stats separates significant strikes (strikes at distance, plus power strikes in the clinch and on the ground) from total strikes, which include short ground-and-pound and clinch taps. Fight DNA's striking family is built on significant strikes because that is the category the source records consistently per round.
Pace vs accuracy
Sig. landed / min is pace: how many significant strikes landed for every observed minute. Sig. accuracy is efficiency: the share of attempts that landed. They move in different directions. A fighter who throws constantly can land a lot at a modest accuracy; a counter-striker can post high accuracy at low volume. Read them together.
Example
7.63 significant strikes landed per minute across 23 observed minutes, at 68% accuracy. That is a description of output and efficiency in the observed sample, with no claim about how it compares to other fighters.
Striking defense
Sig. absorbed / min is exposure: strikes taken per observed minute. Sig. defense is the share of opponents' attempts that missed. Absorbed and landed are different numbers about different people: landed is what this fighter did, absorbed is what opponents did to them.
Strike targeting
Head, body and leg shares split this fighter's own significant-strike attempts by target. They sum to roughly 100% and describe where the offense went, not how much landed.
Distance, clinch and ground
Distance, clinch and ground shares split the same attempts by where they were thrown from. A fighter with a high ground share spends offensive time on top; a high distance share means the exchanges happened on the feet at range.
09
Grappling DNA
Takedown activity, success, control and submission pressure.
Takedown metrics
TD attempts / 15 is how often a fighter shoots; TD landed / 15 is how often it works, both scaled to 15 observed minutes. TD accuracy is landed divided by attempted. Read the three together: a fighter with one attempt and one completion shows 100% accuracy from a single shot.
Control
Control / TD is the recorded control time that followed each successful takedown, in seconds. Control share is the share of all observed time spent in control. The source records control per round for each fighter; Fight DNA does not split it by position.
Submission activity
Sub attempts / 15 counts submission attempts per 15 observed minutes. It measures pressure, not success; submission wins live in Finish DNA.
10
Finish DNA
How this fighter's recorded wins end and when they tend to end.
Finish rate is the share of recorded wins that ended by KO/TKO or submission rather than decision. KO/TKO share and Submission share split those finishes by method. Median finish time is the elapsed fight time of the typical finish, counted from the opening bell across rounds, so a 2:30 finish in round two reads as 7:30.
Example
Stored finish_rate = 100% displays as “All archived wins in this sample ended before a decision.” The sentence states the fact and its scope; it does not call the fighter dangerous.
11
Round DNA
How pace and defensive performance change from round to round. Each round shows attempts, landed and absorbed per minute for that round number, pooled across the sample, with the rounds and minutes observed.
Pace retention divides round-two (or round-three) attempt pace by round-one pace over bouts where both rounds are covered; 100% means the later round matched the first. Championship-round delta is rounds 4–5 pace minus rounds 1–3 pace over bouts that reached round four. Defensive drift is strikes absorbed per minute in round three minus round one.
Because these use only bouts where both rounds carry stats, early finishes are excluded by construction. A fighter who ends fights in round one will have a small late-round sample, and the confidence tier will say so.
12
Stance DNA
Historical results against opponents with different listed stances.
Historical split, not causation.A record against southpaws describes what happened in those fights. It does not prove that stance caused the result, and a split built on two appearances describes two fights.
Each row groups completed bouts by the opponent's listed stance at the source (orthodox, southpaw, switch, open stance, sideways or unlisted). Open-stance rows pool bouts where one fighter was listed orthodox and the other southpaw; same-stance rows pool matching stances. Where round stats exist for the split, the strike differential per minute and takedowns per 15 inside that split are shown with the number of bouts that carry stats.
13
Context DNA
How the recorded results differ across fight formats and situations. Records are grouped by three-round and five-round scheduling, title bouts, main events and verified short-notice bouts. A missing split means the record holds no bouts of that kind, not that the fighter has never been in that situation.
14
Matchup DNA
Two fighter histories. One matchup-specific intelligence layer. Fight DNA compares relevant historical characteristics from each fighter rather than treating career averages as isolated numbers: stance interaction (each fighter's record against the other's listed stance), pace differential, distance tendencies, takedown pressure against takedown defense, finish patterns, round progression and contextual experience, where the record supports each.
A supported observation is emitted only when the underlying metric clears its sample threshold. Anything that does not clear it goes to the counter-case column as a warning, so a thin sample is stated instead of hidden. Every visible matchup conclusion traces to stored evidence.
15
PBE Derived explained
PBE derivedPBE DerivedCalculated by PropBetEdge from normalized event, bout and round-level source records using a versioned Fight DNA definition. It is not an official UFC statistic, is not simply copied from a source page, and is not a prediction.Definitionversioned Fight DNA registrySampleshown beside every numberConfidencetier set by the sampleAs ofsnapshot datePBE Derived explained →What does confidence mean? Calculated by PropBetEdge from normalized event, bout and round-level source records using a versioned Fight DNA definition. It is not an official UFC statistic, is not simply copied from a source page, and is not a prediction.
Wherever the badge appears, the definition, sample, confidence and as-of date sit behind the explainer. SOURCE marks a raw fact republished from ESPN or UFC Stats. LICENSED and MODEL labels are reserved for data that does not yet ship publicly.
16
Source data vs derived features
The source records are building blocks. The derived feature system is the product.
Source Source facts
Event date and venue
Bout result, method and round
Significant strikes landed and attempted
Takedowns landed and attempted
Control time, knockdowns, submission attempts
Per-round UFC Stats rows
Republished with their source URL and capture time. Where two sources disagree, the disagreement is shown.
PBE derived PropBetEdge derived
Stance-specific records and finish rates
Significant-strike differential by opponent stance
Head / body / leg target shares
Distance / clinch / ground position shares
Pace retention round to round
Finish-round distributions and median finish time
Per-15-minute normalized takedown, knockdown and submission rates
Control time per takedown and control share
Matchup comparisons and supported observations
Only metrics that are implemented today are listed. Position profiles from licensed data are reserved until validated.
17
Methodology and versioning
Definition v1. Fight DNA metrics are versioned so the same historical record can be rebuilt consistently as the feature system evolves. When a definition changes, its version changes, and a snapshot always states which version produced it. That is a strength: an as-of snapshot from last year and one from today can be compared on equal terms.
Every derived number can expose its metric key, value, unit, numerator and denominator, sample bouts, rounds and seconds, as-of date, definition version, confidence, coverage status and source families. Internal infrastructure details stay internal; the receipts are public.
18
What the numbers cannot tell you
They cannot say who wins. A profile is a description of the observed record, not a probability.
They cannot see camp, injuries, weight cuts, referee assignment, judging or motivation. Fight Week and the newsroom carry those only when a verified source exists.
They cannot separate a fighter from their opponents. A high absorbed rate can mean a porous defense or a run of high-output opponents; the counter-case column exists for exactly that reason.
They cannot fix a small sample. Two fights are two fights, and the confidence tier will keep saying so.
They cannot substitute for watching the fights. Fight DNA tells you where to look; the tape tells you why.
19
Glossary
Every technical label used in Fight DNA, with the definition it maps to. Popovers on fighter and matchup pages link here.
Striking DNA
Significant strikes landed per minuteSig. landed / min · rate metric
Offensive striking pace normalized by observed fight time: how many significant strikes this fighter landed for every minute the round stats cover.Per minute of observed, stat-covered fight time.Caution: More volume is not automatically better. Accuracy, defense, position and opponent context matter too.sig_str_landed / observed_minutes
Significant strikes absorbed per minuteSig. absorbed / min · rate metric
How many significant strikes opponents landed on this fighter for each observed minute.Per minute of observed, stat-covered fight time.Caution: This is exposure, not a complete measure of defense. A high-pace fighter can absorb more simply by being in more exchanges.opponent_sig_str_landed / observed_minutes
Significant strike differential per minuteStrike differential / min · rate metric
Significant strikes landed minus significant strikes absorbed, per observed minute. Positive means this fighter landed more than they took.Per minute of observed, stat-covered fight time.(sig_str_landed − opponent_sig_str_landed) / observed_minutes
The share of significant-strike attempts that landed.A percentage of this fighter's own attempts.Caution: Accuracy rises when a fighter throws less and picks shots; read it together with pace.sig_str_landed / sig_str_att
The share of opponents' significant-strike attempts that did not land.A percentage of the opponents' attempts.Caution: Defense counts avoided attempts, not damage. It says nothing about the strikes that did get through.1 − opponent_sig_str_landed / opponent_sig_str_att
Head attack shareHead share · rate metric
The share of this fighter's significant-strike attempts aimed at the head.A share of this fighter's own significant-strike attempts, so the six shares describe where the offense went, not how much landed.head_att / sig_str_att
Body attack shareBody share · rate metric
The share of this fighter's significant-strike attempts aimed at the body.A share of this fighter's own significant-strike attempts, so the six shares describe where the offense went, not how much landed.body_att / sig_str_att
Leg attack shareLeg share · rate metric
The share of this fighter's significant-strike attempts aimed at the legs.A share of this fighter's own significant-strike attempts, so the six shares describe where the offense went, not how much landed.leg_att / sig_str_att
Distance attack shareDistance share · rate metric
The share of this fighter's significant-strike attempts thrown at distance (on the feet, not in the clinch).A share of this fighter's own significant-strike attempts, so the six shares describe where the offense went, not how much landed.distance_att / sig_str_att
Clinch attack shareClinch share · rate metric
The share of this fighter's significant-strike attempts thrown in the clinch.A share of this fighter's own significant-strike attempts, so the six shares describe where the offense went, not how much landed.clinch_att / sig_str_att
Ground attack shareGround share · rate metric
The share of this fighter's significant-strike attempts thrown on the ground.A share of this fighter's own significant-strike attempts, so the six shares describe where the offense went, not how much landed.ground_att / sig_str_att
Knockdowns per 15 minutesKnockdowns / 15 · rate metric
Knockdowns scored, normalized to 15 minutes of observed fight time.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).Caution: Knockdowns are rare events; small samples move this number a lot.kd × 900 / observed_seconds
Knockdowns per 15 minutesKnockdowns / 15 · rate metric
Knockdowns scored, normalized to 15 minutes of observed fight time.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).Caution: Knockdowns are rare events; small samples move this number a lot.kd × 900 / observed_seconds
Times this fighter was knocked down, normalized to 15 minutes of observed fight time.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).opponent_kd × 900 / observed_seconds
Times this fighter was knocked down, normalized to 15 minutes of observed fight time.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).opponent_kd × 900 / observed_seconds
Takedown activity normalized to the length of a standard three-round UFC fight.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).td_att × 900 / observed_seconds
Successful takedowns normalized to 15 minutes of observed fight time.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).td_landed × 900 / observed_seconds
Takedown accuracyTD accuracy · rate metric
The share of takedown attempts that were completed.A percentage of this fighter's own attempts.Caution: A fighter who rarely shoots can post a high accuracy from a handful of attempts.td_landed / td_att
Control time per takedownControl / TD · rate metric
How much recorded control time followed each successful takedown.Seconds of control for every takedown landed.Caution: Control time is recorded per round by the source; it is not split by position or by who initiated the grappling.ctrl_sec / td_landed
Control share of observed timeControl share · rate metric
The share of observed fight time this fighter spent in recorded control.A percentage of all observed, stat-covered seconds.ctrl_sec / observed_seconds
Submission attempts normalized to 15 minutes of observed fight time.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).sub_att × 900 / observed_seconds
Reversals per 15 minutesReversals / 15 · rate metric
Position reversals (as recorded by the source) normalized to 15 minutes of observed fight time.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).rev × 900 / observed_seconds
Finish DNA
Finish rate of winsFinish rate · result metric
The percentage of recorded wins ending by KO/TKO or submission rather than decision.A percentage of wins, not of all bouts.Caution: A fighter with two wins and two finishes shows 100%. Read the sample beside it.(ko_tko_wins + submission_wins) / wins
KO/TKO share of winsKO/TKO share · result metric
The percentage of recorded wins that came by knockout or technical knockout.A percentage of wins.ko_tko_wins / wins
Submission share of winsSubmission share · result metric
The percentage of recorded wins that came by submission.A percentage of wins.submission_wins / wins
Median finish timeMedian finish time · result metric
The median elapsed fight time of this fighter's finish wins, counted from the opening bell across rounds.Minutes:seconds of total fight time; a 2:30 finish in round two shows as 7:30.median((round − 1) × 300 + time_sec) over finish wins
Finish round distributionFinish wins by round · distribution
How many of this fighter's finish wins ended in each round.count finish wins by round
Finished-by round distributionFinished by round · distribution
How many of this fighter's KO/TKO or submission losses ended in each round.count finish losses by round
Round profile
Significant-strike attempts per minute (this round)Attempts / min · rate metric
Significant strikes attempted per observed minute of this round number, pooled across the sample.Per minute of observed, stat-covered fight time.rN_sig_str_att / rN_observed_minutes
Round-two pace retentionR2 pace retention · rate metric
Round-two significant-strike attempt pace divided by round-one pace, over bouts where both rounds are covered. 100% means round two matched round one.A ratio shown as a percentage of the round-one pace.Caution: Only bouts with both rounds covered count, so early finishes are excluded by construction.r2_sig_att_per_min / r1_sig_att_per_min
Round-three pace retentionR3 pace retention · rate metric
Round-three significant-strike attempt pace divided by round-one pace, over bouts where both rounds are covered.A ratio shown as a percentage of the round-one pace.Caution: Only bouts that reached round three with stats count.r3_sig_att_per_min / r1_sig_att_per_min
Championship-round pace deltaChampionship-round delta · rate metric
Rounds 4–5 significant-attempt pace minus rounds 1–3 pace, over bouts that reached round four. Negative means output fell in the championship rounds.Attempts per minute, as a difference.late_round_sig_att_per_min − early_round_sig_att_per_min
Defensive drift, round three versus round oneDefensive drift R3 vs R1 · rate metric
Significant strikes absorbed per minute in round three minus the round-one rate, over bouts with both rounds covered. Positive means more strikes were getting through late.Strikes absorbed per minute, as a difference.r3_absorbed_per_min − r1_absorbed_per_min
Stance DNA
Record versus opponent stanceRecord · record
Wins, losses, draws and no-contests against opponents with this listed stance.Caution: Historical split, not causation. Stance is the opponent's listed stance at the source, not observed switching.aggregate outcomes grouped by opponent_stance
Appearances versus opponent stanceAppearances · record
Completed bouts against opponents with this listed stance.
Finish rate versus opponent stanceFinish rate · result metric
KO/TKO plus submission wins divided by wins against this listed stance.A percentage of wins in the split.(ko_tko_wins + submission_wins) / wins_vs_stance
KO/TKO rate versus opponent stanceKO/TKO rate · result metric
KO/TKO wins divided by completed appearances against this listed stance.A percentage of appearances, not of wins.ko_tko_wins / completed_appearances_vs_stance
Submission rate versus opponent stanceSubmission rate · result metric
Submission wins divided by completed appearances against this listed stance.A percentage of appearances, not of wins.submission_wins / completed_appearances_vs_stance
Significant strike differential per minute versus opponent stanceStrike differential / min · rate metric
Significant strikes landed minus absorbed, per observed minute, in bouts against this listed stance.Per minute of observed, stat-covered fight time.(sig_str_landed − opponent_sig_str_landed) / observed_minutes
Takedowns landed per 15 minutes versus opponent stanceTakedowns / 15 · rate metric
Takedowns landed per 15 observed minutes in bouts against this listed stance.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).td_landed × 900 / observed_seconds
Knockdowns per 15 minutes versus opponent stanceKnockdowns / 15 · rate metric
Knockdowns scored per 15 observed minutes in bouts against this listed stance.Normalized to 15 minutes of observed fight time (the length of a standard three-round bout).kd × 900 / observed_seconds
Context splits
Three-round record3-round bouts · record
Results in bouts scheduled for three rounds.aggregate outcomes where scheduled_rounds = 3
Five-round record5-round bouts · record
Results in bouts scheduled for five rounds.aggregate outcomes where scheduled_rounds = 5
Title bout recordTitle bouts · record
Results in bouts contested for a title.aggregate outcomes where is_title
Main event recordMain events · record
Results when the bout was the last on its card (highest bout order).aggregate outcomes where is_main_event
Short-notice recordShort notice · record
Results in bouts that carry a verified short-notice value in the record.Caution: Only bouts with a verified short-notice value are counted; the absence of a split does not mean a fighter never took short notice.aggregate outcomes where short_notice_days is not null
PropBetEdge is an independent sports intelligence product and is not affiliated with the UFC, Zuffa LLC, TKO Group, ESPN or any sportsbook. Fight DNA is not an official UFC statistic. Nothing on this page is betting advice or a prediction of fight outcomes.
