The patterns
behind the fights.
Fight Atlas turns complete professional careers into transparent research. Every finding below includes its sample size, calculation rule and limitation.
Decision is only one part of the record.
Each bout is counted once even when both fighters have a compiled career. Methods are grouped from their source-listed descriptions.
Submission
Decision
DQ / technical
Other / unknown
Classification rule: decisions are detected first; submissions include named locks and chokes; striking stoppages include KO, TKO, punches, kicks, knees, elbows and slams. Unclassifiable source labels remain visible as other or unknown.
The finish curve,
round by round.
Only non-decision bouts with a recorded round from one through five are included.
Sample: 7,674 recorded stoppages.
Recovery is measurable.
For every current UFC roster career, the next chronologically recorded professional bout after a loss is classified as a win, loss, draw or no contest.
This is a historical roster-career pattern, not a forecast. Fighters without another recorded bout after a loss are excluded from the denominator.
Days between bells.
Intervals use consecutive dated professional bouts in current UFC roster careers, including fights before an athlete joined the promotion.
Finish rates by division.
These are full-career wins belonging to fighters currently listed in each UFC division—not UFC-only results and not a ranking of divisional quality.
Most connected fighters.
Connection count measures indexed opponent links in the current Fight Atlas graph. It rewards data depth and career breadth—not competitive greatness.
Numbers readers can audit.
Deduplication: mirrored career rows are collapsed using date, fighter identities, event, method, round and time.
Coverage: unresolved opponent identities remain excluded from identity-based research but stay disclosed in the overall archive.
Updates: findings recalculate whenever the underlying career dataset changes. Last checked August 29, 2026.