MLB Analytics Counter-Examples Search: Managers Who Overrode the Model and Were Right

## Search Objective Find a recent MLB decision (2024–2025) where a manager overrode an analytics recommendation and was demonstrably correct—not lucky, but right because he saw something the model couldn't. ## Result: No Fully Verified Case; One Partial Match ### Closest Match: Carlos Mendoza, Mets vs. Cardinals — April 28, 2024 - **Situation:** Tie game, 8th inning. Starter Jose Quintana still pitching. Matchup data favored a right-handed reliever. - **Analytics recommendation:** "The book, the binder, the computer — they all suggested removing Quintana in favor of a right-handed pitcher." (MLB.com) - **Manager's choice:** Kept Quintana in. Mendoza walked to the mound, asked Quintana how he felt, and sent the reliever (Ottavino) back to his seat. - **Reasoning:** "The conviction, the way he asked for it, I was pretty confident that that was his game... I went against the matchup, but I'm glad it worked out." The stated basis is the pitcher's own assessment and body language — information a matchup model does not contain. - **Outcome:** Quintana retired the next batter, game stayed tied, Mets won 4-2 in extra innings. - **Assessment:** PARTIAL MATCH. The override is documented and the reasoning points to information a model lacks. But this is one plate appearance in one game. A single good outcome cannot separate "he saw something" from luck. No pre-play record shows how often this type of read is right. The reporter's framing ("ignored analytics") comes from narrative, not from any stated model output. - **Source:** https://www.mlb.com/news/carlos-mendoza-trusting-gut-feel-as-mets-manager ### Related Case: Carlos Mendoza, Mets NL Wild Card Game 3 — Oct. 3, 2024 - **Situation:** Elimination game at Milwaukee. Pre-game plan was to get ~12 outs from Quintana and not let him face hitters a third time. - **Manager's choice:** Let Quintana face leadoff hitter Jackson Chourio a third time in the 5th inning, then kept him for the 6th. - **Reasoning:** Pitching coach Jeremy Hefner: "Q had his good stuff... so it was one of those things where you kind of go against the..." [plan] - **Outcome:** Quintana performed well. - **Assessment:** NOT A CLEAR OVERRIDE. The plan was industry standard (limit starters to two times through the order in playoffs), not necessarily a live analytics recommendation. This is a pre-game plan being adjusted, not a model being overruled. - **Source:** https://www.mlb.com/news/mets-nl-wild-card-series-game-3-pitching-plan-pays-off ## Key Finding After searching NFL, NBA, soccer, MLB, and NHL, no fully verified case exists where a coach/manager overrode a documented analytics recommendation and was demonstrably correct (not lucky). The Mendoza case is the closest match but does not meet the full bar. ## Possible Explanations 1. **Model superiority:** Analytics models may genuinely be better at these decisions than human intuition 2. **Media bias:** Coverage favors reporting conservative mistakes and failed overrides over successful ones 3. **Under-reporting:** Clubs rarely publish internal model disagreements; successful overrides may happen but not be documented 4. **Bar too high:** "Demonstrably correct" may be impossible to establish from single-game outcomes ## Sources - MLB.com (Mendoza gut feel) - MLB.com (Mets Wild Card pitching plan) - New York Post (Mendoza pitching intuition)