builtbykev
7ac6aa73e3
A player is a blend across independent axes, not one label. Skubal is a STARTER and a strikeout arm and a ground-ball arm and a control arm — four true things at once, and single-label classification threw three of them away. AXIS INDEPENDENCE WAS MEASURED, NOT ASSUMED. Correlations over the live store (467 batters, 531 pitchers); anything |r| >= 0.70 is one underlying trait and was collapsed so we never show one trait as two archetypes. Batter k% ~ whiff% +0.89, hard-hit% ~ exit velo +0.88, chase% ~ swing% +0.87, chase% ~ bb% -0.72; pitcher k% ~ whiff% +0.76, gb% ~ fb% -0.73 — all collapsed. The survivors are genuinely orthogonal, and one result is worth stating: pitcher velocity correlates +0.14 with K%, +0.07 with whiff% and +0.07 with GB%. Velocity is NOT a proxy for missing bats — a hard thrower who misses no bats is a real distinct type, so CANNON earns its own axis rather than being folded into STRIKEOUT. Pitcher K% ~ GB% is -0.10, so PUNCHOUT and SINKER are independent, which is exactly the multi-axis thesis. Cut-lines are the measured p75 (distinctive) and p90 (elite), per role where the tails differ even when the medians agree: reliever GB% p90 is 54.1 against a starter's 48.9, both with a median of 42.5. THE FALLBACK IS DELETED. classify() used to return FLEX (mlb) / SHIELD (wnba) / CONNECTOR (nba) at weight 1.0 when nothing scored — "could not classify" rendered as a fully-confident classification of a real archetype, with descriptive education copy attached. 8 of 18 MLB players carried it, and FLEX could never be earned because its only scoring input had zero writers. Every sport now does what MMA already did: unclassified is absent. Induced on real players. Skubal: STARTER, throws L, WHIFF + SEAM + PINPOINT, all elite. Judge: BOMBER + GRINDER + WHIFF RISK — elite power, patient, strikes out, three true things. Kwan: SURGEON + SNIPER + SLASH with NO power claimed (0.4 barrel% is absent, not "low power"). Josh Bell, who used to classify as DRIVER: empty blend, "No standout profile — league-average across every measured axis." Alan Roden, who was FLEX at weight 1.0 on 21 PA: every axis absent, "Not enough plate appearances yet — no profile claimed." Per-axis honest-absence holds: a velo-less pitcher keeps every other axis, and NO DATA is distinguishable from LEAGUE-AVERAGE rather than collapsing into one shrug. The full vector is stored for Layer 3; only the top three distinctive traits surface. Three existing tests asserted the fallback and were updated to assert absence. One of them surfaced a real robustness gap: classify(sport, null) threw, because an explicit null does not trigger a default parameter and every scorer dereferences its argument. Guarded. Every baseball name is accounted for in docs/ARCHETYPE-AXES.md — built, alias, tier, or shelved with its unlock condition. Zero orphans; cross-sport names left for their sport. Tests 3601 passed / 293 suites. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VCNgGSt5qvcLxaeQqa7Zpj
4.5 KiB
4.5 KiB
MLB ARCHETYPE AXES — Layer 2
A player is a blend across independent axes, not one label. Skubal is a STARTER and a strikeout arm and a ground-ball arm and a control arm — four true things at once. Single-label classification is lossy, and its failure mode is the FLEX disease: when nothing matches, invent a bucket.
There is no fallback. Unremarkable on an axis → absent on that axis. A genuinely average player surfaces nothing, and says so.
Axis independence — measured, not assumed
Correlations over the live store (467 batters PA≥50, 531 pitchers IP≥10). |r| ≥ 0.70 = one underlying trait → collapsed, so one trait is never shown as two archetypes.
| Pair | r | Decision |
|---|---|---|
| batter k% ~ whiff% | +0.89 | collapsed → one SWING-AND-MISS axis |
| batter hard-hit% ~ avg exit velo | +0.88 | folded into POWER as intensity |
| batter chase% ~ swing% | +0.87 | collapsed → one AGGRESSION axis |
| batter chase% ~ bb% | −0.72 | same axis inverted (discipline) |
| pitcher k% ~ whiff% | +0.76 | collapsed → one STRIKEOUT axis |
| pitcher gb% ~ fb% | −0.73 | collapsed → one signed tilt axis |
| batter barrel% ~ hard-hit% | +0.70 | at the line; barrel leads |
| pitcher velo ~ k% | +0.14 | INDEPENDENT — velo earns its own axis |
| pitcher velo ~ whiff% | +0.07 | independent |
| pitcher velo ~ gb% | +0.07 | independent |
| pitcher k% ~ gb% | −0.10 | INDEPENDENT — PUNCHOUT ⊥ SINKER |
| batter barrel ~ launch | +0.23 | independent |
| batter k% ~ chase% | +0.05 | independent |
The velocity result matters: velo is not a proxy for missing bats. A hard thrower who misses no bats is a real, distinct type.
Cut-lines
p75 = distinctive · p90 = elite, from the measured distribution — per
role where the tails differ even when medians agree (reliever GB% p90 = 54.1
vs starter 48.9, both median 42.5). Sample floors: PA ≥ 50 · IP ≥ 10.
Every baseball name accounted for
| Name | Status | Axis / reason |
|---|---|---|
| WORKHORSE | TIER | role=starter + high IP; the old hardcoded default is gone |
| PUNCHOUT / WHIFF | BUILT | strikeout (p75 / p90) |
| SINKER / SEAM | BUILT | ground_ball |
| FLY BALL / ELEVATOR | BUILT | fly_ball |
| SURGEON ARM / PINPOINT | BUILT | control (low BB%) |
| NIBBLER / SCATTERGUN | BUILT | wild (high BB%) |
| BAIT / TRAPDOOR | BUILT | chase |
| HITTABLE / BATTING PRACTICE | BUILT | contact_allowed |
| CANNON / HOWITZER | BUILT | velocity — unlocked by the 53%→99% velo fix |
| SIDEARM / SUBMARINE | BUILT | slot (low arm angle) |
| STARTER / RELIEVER / CLOSER / SETUP | BUILT | role, from real usage |
| SLUGGER / BOMBER | BUILT | power (barrel%) |
| TECHNICIAN / SURGEON | BUILT | contact (low K%) |
| WHIFF RISK / WINDMILL | BUILT | swing_miss |
| GRINDER / SNIPER | BUILT | patience |
| FREE SWINGER / HACKER | BUILT | aggression |
| TOPSPIN / LOFT | BUILT | launch |
| SLASH / BARREL FINDER | BUILT | line_drive |
| HAMMER | ALIAS | of chase/strikeout — best-pitch whiff r=0.50 with overall; not independent enough for its own axis |
| BRUSH | ALIAS | of contact (TECHNICIAN) |
| DRIVER | ALIAS | of power — RBI is lineup context, not a player trait |
| FLEX | RETIRED | it was the fallback, never earned: utility had zero writers |
| MIRROR | SHELVED | needs switch-hitter axis; bats='S' now joined — unlock when a platoon-split feed lands |
| GHOST / BURNER / LEG | SHELVED | speed: SB is statsapi, not yet joined into the aggregate store |
| CATALYST | SHELVED | table-setter: needs lineup slot (not ingested) |
| HYBRID | BUILT-ADJACENT | two-way = both role profiles exist (position TWP) |
| ALPHA | ALIAS | of strikeout+control combined; the blend expresses it |
| FASTBREAK · ENFORCER · CHAINMOVER · CONDUCTOR · ARCHITECT · TORCH · LINK · ARTILLERY · BELL COW · etc. | OTHER SPORT | NBA/WNBA/NFL names — correctly not built here |
Zero orphans.
Output shape
{
role, roleDetail, roleLabel, sufficient, sample, bats, throws,
vector: { axisKey: {tier,label,value,strength} | null }, // Layer 3 reads ALL
blend: [ {label, axis, tier, value, strength} ], // top ≤3 to surface
absent: [axisKey…], // NO DATA — distinct from unremarkable (vector null)
note // honest copy when the blend is empty
}
absent vs a null vector entry is a real distinction: absent means we could
not measure it, null means we measured it and he is ordinary.