STRA/IN™ · Stress to Rest Arm per Inning

Pitch counts don't
measure cost.

Two pitchers throw 85 pitches on Friday. One cruised. One fought for every out.

The box score records the same number. Their arms do not. STRA/IN scores every inning on the situation inside it — two-strike counts, traffic on the bases, extended at-bats — and returns a single number for what that inning actually cost.

Open the live board Get the weekly board

What the data says

r² = .005
Pitch count explains almost nothing
Across 13,501 outings, pitch count explained less than half a percent of the variation in situational stress. The number every dugout manages by is nearly uncorrelated with what it's meant to protect. n = 13,501 · p = 3.8 × 10⁻¹⁶
0.76 mph
Velocity lost, Efficient to Heavy+
Pitchers throw measurably slower in the inning after an expensive one, and the effect climbs step by step across all five bands. Not a threshold that appears once — a gradient that holds the whole way up. r = −0.179 · monotonic across 5 bands
4.49×
Adjusted odds of arm injury
Pitchers who repeatedly crossed into the Heavy band reached the arm injured list at 18%, against 8% for those who didn't. OR 4.49 · 95% CI 2.40–8.39 · p = 2.6 × 10⁻⁶ · 103 of 118 cases

Three inputs. No new hardware.

STRA/IN runs on scorekeeping data — counts, outcomes, runners. That's it. No TrackMan, no Hawk-Eye, no cameras, no wearables. If your program keeps a scorebook, you already collect everything the core score needs.

It is a monitoring metric, not a decision engine: a deterministic formula that reports what an inning cost, reviewed between innings and between games. It does not call pitches and it does not recommend substitutions.

Where it misses

Every claim above survived testing. Five others didn't, and were cut: a hard-thrower differential, a band gradient, two circular results, and an exposure artifact that looked like a 91% clearance rate until it was examined properly.

STRA/IN also missed on individual pitchers — Eflin and Herrera among them — and those misses stay in the paper. A model that only reports its wins isn't a model, it's marketing.

Get the weekly board

One email a week: the innings that cost the most across the league, what moved, and what the model is learning. No charge, no pitch.

STRA/IN Baseball LLC · Longhunter OS · michael@strainbaseball.com · @strainbaseball Validated on 31,000+ innings across 767 pitchers and 30 teams. The live board updates every morning from Baseball Savant. Not affiliated with Major League Baseball.
0