VERIFIEDcounts only · lead-left

Keenan Lewis

logged strikes

≥194

reviewed attempts (sweep, lower bound)

fights on file

verified fights

machine fights

Ask the Analyst

No tendency-grade fight yet — this fighter's corpus is counts-only. The Tale of the Tape tab shows exactly what exists and what would unlock more.

limitedDefense telemetry is early: the signal is real and eyeball-confirmed, but capture coverage — not the metric — is the blocker. More verified tape unlocks it.

defensive tell — guard recovery

insufficient data

0 of 25needs 25 more

guard telemetry coverage too thin under the frozen validity rules for this fighter

unlock: a defense-telemetry upgrade is pre-registered — these cells re-grade automatically after the next data pass

pre-registered gate: n ≥ 25 for shares/conditionals

fight volume

insufficient data

0 of 25needs 25 more

no swept events

pre-registered gate: n ≥ 25 for shares/conditionals

Verified verdict profile

Operator verdict-ledger evidence, kept separate from temporal tendencies. Complete reviews may support totals and gated rates; partial reviews remain labeled samples; reviews below the coverage floor expose only their withholding reason.

verdict evidence on file

No complete reviewed fight is eligible for pooled verdict totals, accuracy, mixes, or rates. Partial and withheld rows below remain coverage evidence only.

romero lewis

2363/3966 reviewed (60%) · Partial review

Reviewed sample only — not fight totals or accuracy: 194 own real attempts · 179 opponent real attempts.

Own clean-landed rate: complete review required for an accuracy rate · Opponent clean-landed-on-him rate: complete review required for a defense rate

Sample from an in-progress review; ordering is not random — rates may be inflated and early-fight-weighted.

weapon sample

elbow 2 (1%) · fist 142 (73%) · foot shin 43 (22%) · knee 7 (4%)

trajectory sample

arcing 43 (22%) · hooked 39 (20%) · overhand 8 (4%) · straight 91 (47%) · unlabeled 10 (5%) · upward 3 (2%)

side sample (physical, not stance)

left 78 (40%) · right 116 (60%)

target sample

body 26 (13%) · head 154 (79%) · legs 14 (7%)

outcome sample

blocked 27 (14%) · landed 77 (40%) · missed 56 (29%) · partial 21 (11%) · slipped 12 (6%) · unlabeled 1 (1%)

opponent outcome sample

blocked 12 · landed 77 · missed 56 · partial 15 · slipped 16 · unlabeled 3

53 unresolved attribution/validity verdicts — a floor on missingness.

The three frozen verdict gates

Per-fight mix shares require n ≥ 25 reviewed real attempts; below it, counts only.
Complete-fight pooled mix shares and own rates require n ≥ 25 real attempts.
Defense clean-absorb rates require a complete review and n ≥ 25 opponent real attempts.

Partial reviews are non-random samples: counts and eligible sample mixes are not fight totals or accuracy.
Physical side means left/right, not lead/rear stance. Jab/cross for swept strikes is stance-INFERRED from pose: a windowed ankle-vote at the strike instant when decisive (94.3% vs hand-log truth), else the fight-level lead; strikes where both abstain stay "straight" rather than guess.
A single reviewer produced these verdicts; unresolved attribution/validity verdicts are a floor on missingness.

Sweep analytics — reviewed sample

Counts from human-reviewed candidates only; they are lower bounds for the whole fight. Rates and temporal patterns stay withheld until a fight's review is complete.

lower bounds · review in progress

≥194

reviewed strike attempts

≥77

landed clean (reviewed)

≥179

opponent attempts faced

≤36

est. real strikes unreviewed

vs Anthony Romero1852/3047 in-scope reviewed · est ≤36 unreviewed

attempts ≥194 · landed clean ≥77 · partial 21 · opponent attempts ≥179

fist 142 · 73%foot_shin 43 · 22%knee 7 · 4%elbow 2 · 1%

landed 77 · 40%missed 56 · 29%blocked 27 · 14%partial 21 · 11%slipped 12 · 6%unlabeled 1 · 1%

head 154 · 79%body 26 · 13%legs 14 · 7%

Defense rates require a complete review: the unreviewed low-confidence tail is exactly where missed opponent attempts live, so a partial denominator would flatter the defender.

Coverage & provenance

packpack-v3

vocabv1.1

data through2026-08-10

regression gatePASS

vs Anthony Romero

partial assertion review: 2363/3966 (59.6%); in-scope 1852/3047 (60.8%), est ≤35.91 real strikes unreviewed (calibrated); counts are lower bounds — stream admitted at census grade or 100%; assertion review partial: 2363/3966 (59.6%); 53 unresolved — a floor on missingness; Sample from an in-progress review; ordering is not random — rates may be inflated and early-fight-weighted.

2363/3966 (60%)verified

Not measurable from broadcast

the single-feed footage can’t support these — they are not a locked upgrade, they simply aren’t recoverable. Saying so is the product.

not recoverable

Landed / grazed / missed

outcome carries ~no machine signal at this capture and is unreliable even by eye from a single broadcast feed

not recoverable

Fine punch subtype (jab / cross / hook)

capped by 72% annotator self-agreement and the single-feed capture wall

not recoverable

Clinch & ground interior

the corpus is standup-scoped — a single broadcast feed can't resolve grappling exchanges

not recoverable

Feint intent

intent-without-contact is not recoverable from broadcast video

Known limits

single-annotator ground truth; measured undercount ~8.8% on the deepest-swept fight — global estimate 11–15%; every count is a floor, not an exaggeration
tendency numbers are standup-scoped and in-round; clinch/ground exchanges are not measured
stance (lead/rear) is a per-fight constant from a pose asset; a mid-fight stance switch would mislabel lead/rear symmetrically
per-round counts sit on a Poisson floor of roughly ±18–25%; product language is trends and conditionals, never single-round precision