REVERSE REPORT · Paul Skenes vs St. Louis

Thursday, September 24, 2026 · PNC Park · Game 823326 · Final start of the 2026 season
6.1 IP
3 H
1 R / 1 ER
2 BB
8 K
1 HR
96 P / 23 BF
Net RV −0.477 (good)
Process read: The surface line (1 ER, 8 K, 3 H) reads like a near-shutout, and the outcome grading agrees: zero pitches graded as outright execution mistakes all night, the first time that's happened all season. But the actual run-value math is closer than the box score suggests — St. Louis hit the ball hard all game (8 of 13 balls in play at 95+ mph), and two of those loud swings (Wetherholt's near-leadoff-homer flyout in PA1, Burleson's solo shot in PA3, plus Wetherholt's PA10 double) carry real weight in the actual per-pitch run-value model even though the location on all three was defensible, not a middle-middle miss. TTO1 and TTO2 both finish barely positive (bad) once weighted that way; the start is saved entirely by a dominant final stretch — TTO3's −1.362 over the last 5 batters, capped by a Skenes-started double play and a called strikeout to end the season. Mechanically this was the healthiest start of the year: fastball velocity climbed as the outing went on (95.9 in the 1st, 97.0–97.2 by the 4th–6th), extension did the same. A clean way to end the season, even if the final run-value tally is closer than "zero fixable mistakes" makes it sound.

Time-Through-The-Order

TTO1 · PA 1–9
+0.515
34 pitches · 9 batters
TTO2 · PA 10–18
+0.370
46 pitches · 9 batters
TTO3 · PA 19–23
−1.362
16 pitches · 5 batters (into the 3rd time through)
TTO1 and TTO2 both land slightly positive (bad) once weighted by the real per-pitch model — Burleson's HR (PA3) and Wetherholt's double (PA10) carry more value there than the same-night estimate showed, the same correction pattern seen in the 9/18 KC report. TTO3 is what saves the night: −1.362 across just 5 batters, the best per-batter rate of the season, capped by a double play to escape the 5th and a six-pitch strikeout to end the start and the year.

Verdict Split

0
Fixable
4
Soft-Fixable
3
Noise
16
Clean
Zero FIXABLE plate appearances — the first shutout of that category all season. Even the home run graded NOISE rather than FIXABLE: it came on a fastball up in the zone (an intended location), not a middle-middle miss.

Contact Quality

2
Barrels allowed
8
Hard-hit BIP (95+)
13
Total BIP
92.0
Avg EV allowed
This is the one number that doesn't look like a shutout-caliber start on paper — 8 of 13 BIP were hard-hit, well above the season rate. But hard contact landed almost entirely on the ground or right at fielders (Burleson's 110.9-mph 4th-inning grounder, Walker's 104.7-mph 6th-inning tapper) rather than in gaps. The two barrels were PA1 (Wetherholt, run down by Cruz in center for a sharp flyout) and PA3 (Burleson's HR) — the only real damage of the night, and it happened early.

Mechanical Check

MECHANICAL CHECK: clean bill of health — and trending up. No fade of any kind across 96 pitches and 6.1 innings. If anything, stuff got stronger as the start (and season) wound down.
InningFF VeloSeason FF AvgFF ExtensionSeason Ext
195.996.86.646.77
296.66.73
396.46.72
497.06.81
596.66.80
697.26.83
795.96.60
Fastball velocity sat at or above the season average from the 2nd inning on, peaked in the 6th (97.2 avg, 97.4 max) deep into the outing, and extension followed the same shape — both innings 4 and 6 topped the season mark. Inning 7 (2 pitches, small sample) ticks back toward the inning-1 numbers, unsurprising on the final two batters of a 96-pitch night. Release point held station throughout (x range −2.58 to −2.82 ft, z range 5.56–5.69 ft), no glove-side or arm-side drift of the kind flagged earlier this season (DET 8/19, SF 9/1). No mechanical follow-up needed — ends the year on the healthiest note of the sample.

Plate Appearances

All 23
Fixable 0
Soft 4
Noise 3
Clean 16

Bottom-Line Math

TTO1 (PA1–9)+0.515
TTO2 (PA10–18)+0.370
TTO3 (PA19–23)−1.362
NET RV (game, official)−0.477
Runs allowed (actual, official)1 ER
Run explained by the lone NOISE-graded HR (PA3)1 of 1
FIXABLE PAs on the night0
Run values are the indexed pitch-level dataset's actual run-expectancy deltas (summed per plate appearance), pulled once the game finished indexing the next day; this replaces the same-night live-feed RE24 estimate used at build time. Both passes agree on pitch counts, sequencing, verdict grading, and every contact-quality number (EV/LA/xBA/barrels) — the shift is in the run-value split, same correction pattern as the 9/18 KC report: the simplified 24-state estimate underweights extra-base hits (the HR and the double here) relative to the real per-pitch model.