Observation 11. Champion beats naive. Guerrero validated the sole Edge call at .636 actual. The model under-predicted team output because Jays scored 10, nobody predicted a blowout.
H1 · Champion vs Naive-1
Champion MAE: 0.216 · Naive-1 MAE: 0.228 · Champion wins by 0.012
Five of six hitters the champion wins. Schneider is the miss, the model had him Neutral at .303, he produced .636. The Slider assumption failure partially explains it: WR leaned on the Slider heavily against RHH when the game unraveled, and Schneider was one of the beneficiaries.
H2 · Ranking signal
Spearman rho: +0.395 · Top-3 overlap: 2/3
Guerrero and Varsho in the actual top-3, both predicted. Schneider the miss. Positive Spearman in a blowout is meaningful, the model correctly ordered the lineup for the hitters WR actually faced.
Bias note: Bias is −0.115, meaning the model under-predicted by that amount on average. The Jays scored 10 runs. The under-prediction is expected when a starter gets knocked out in the 4th, the model predicts against a full career-file performance, not a collapse. This is not a flaw in the model. It is the correct behavior when a pitcher underperforms his profile.
Guerrero validated. The corrected Edge call, Guerrero at .397 instead of Springer, produced an actual wOBA of .636. The Slider wOBA split (.389 in 188 PA) that drove his Edge placement was the right signal. This game is a direct vindication of the model revision.
Running study metrics (11 games)
Champion hitter MAE
updating
Game-level MAE this game
0.115
Spearman rho this game
+0.395
Running n
~76 hitter-games
The Slider failure
WR threw the Slider 48.8% of the time to RHH, expected 35.1%. That is a +13.7pp failure, outside the 12pp tolerance. When a pitcher's game unravels, he tends to reach for his best pitch. The Slider is WR's RHH weapon. Leaning on it more than usual is the behavioral response to being hit hard. The model knew the Slider usage was coming, it did not know WR would over-rely on it under pressure. This is exactly the kind of pattern the new research project would capture.
Three of four assumptions held. The Slider to RHH failed when WR leaned on it as his game fell apart.
Pattern to track: This is the second assumption failure in three games. Yamamoto's Splitter failed Apr 7 (−16pp). WR's Slider failed Apr 10 (+13.7pp). Both failures happened when the pitcher was in trouble. A pitcher under pressure reverts to his best pitch more than his career average suggests. The new research project maps exactly this.
Running study after 11 observations. Jays win 10-4. WR knocked out in the 4th.
Game log
1 · Feltner Apr 2024
0.144
2 · Freeland Aug 2025
0.113
11 · Woods Richardson Apr 10
0.115
Emerging patterns
Guerrero Edge validated. The model revision that moved him from Neutral to Edge was correct. Actual .636 confirms the Slider split signal.
Assumption failures correlate with pitcher distress. Yamamoto and WR both failed pitch-mix assumptions when they were getting hit hard. Career-file averages do not capture in-game behavioral shifts under pressure.
Champion beats naive this game. H1 supported at this observation. Running bias still elevated from 2025 prior generosity, but narrowing as 2026 season data accumulates.
Study status after observation 11
Eleven observations. H1 supported in this game. Ranking signal positive (+0.395 Spearman). Running n approaching 80 hitter-games. No inference before n=200. The season is the validation.
Post · X / Twitter
Jays Matchup Intel retro · 4-10-26. TOR 10, MIN 4.
WR: 4 IP, 68 pitches, knocked out in the 4th. Champion beats naive (MAE 0.216 vs 0.228). Spearman +0.395.
Guerrero Edge call validated, actual wOBA .636. The Slider split (.389 in 188 PA) was the right signal.
One assumption failed: Slider to RHH jumped to 48.8% (expected 35.1%). WR reached for his best pitch as the game unraveled. Third time this series a struggling pitcher over-relied on a signature pitch. That pattern is now on the research agenda.
Running n: ~76 hitter-games, 11 observations. No inference before n=200.
scherrology-sys.github.io/jays-matchup-intel · #BlueJays