NFL Analytics Counter-Examples Search: Coaches Who Overrode the Model and Were Right

## Search Objective Find a recent NFL decision (2024–2025) where a coach overrode an analytics recommendation and was demonstrably correct—not lucky, but right because he saw something the model couldn't. ## Result: None Found No source documented a case meeting all three criteria: (1) coach overrode a clear analytics recommendation, (2) the override produced a better outcome, (3) the reasoning was sound and identifiable before the play. ## Cases Examined ### 1. Dan Quinn, Commanders vs. Eagles — Nov. 14, 2024 - **Situation:** Trailing 12-10, 8:01 left, 4th-and-2 at Eagles' 26 - **Model:** Ben Baldwin's model said field goal improves win probability by 2.1% - **Coach's choice:** Went for it - **Outcome:** Stopped, lost 26-18 - **Assessment:** Failed override. No sound reasoning cited. - **Source:** https://www.nbcsports.com/nfl/profootballtalk/rumor-mill/news/dan-quinn-went-against-the-analytics-which-said-commanders-should-kick-the-field-goal ### 2. Mike Macdonald, Seahawks vs. 49ers — Sept. 7, 2025 - **Situation:** Tied 10-10, 3:28 left, 4th-and-1 at 49ers' 19 - **Model:** Seahawks' internal model favored going for it - **Coach's choice:** Kicked field goal for 13-10 lead - **Stated reasoning:** Believed Seahawks needed more than a yard; preferred putting 49ers in three-down mode; cited own estimate that drive from that spot scores 40–60% of the time - **Outcome:** 49ers scored TD, won 17-13 - **Assessment:** Closest to "sound reasoning the model missed," but reasoning only partly verifiable. Outcome turned on a single play (Purdy's 3rd-down TD pass), indicating luck-dependence. Seattle Times noted: "Had Purdy's third-down TD pass on the following drive fallen incomplete, the decision might have looked like a good one." - **Sources:** https://www.geekwire.com/2025/data-be-damned-seahawks-coach-recounts-real-time-call-to-ignore-analytics-model-in-close-loss/, https://www.seattletimes.com/sports/seahawks/why-the-seahawks-didnt-go-for-it-on-fourth-down-vs-49ers-four-downs/ ### 3. Nick Sirianni, Eagles vs. Falcons — Sept. 16, 2024 - **Situation:** Leading 18-15, ~1:46 left, 4th-and-3 at Falcons' 13 - **Coach's choice:** Kicked field goal - **Stated reasoning:** Relied on "scenario chart"; analytics department said kick from opponent's 34 or closer while up 1–5 points with under 4 minutes left - **Outcome:** Atlanta drove for TD, won 22-21 - **Assessment:** Not a verified override—Sirianni claimed analytics supported kicking. Next Gen Stats showed win probability drop from 94% to 85%, but that's a model value, not evidence the kick was better. - **Sources:** https://www.inquirer.com/eagles/eagles-nick-sirianni-defends-late-game-calls-loss-falcons-20240918.html, https://www.cbssports.com/nfl/news/why-nick-sirianni-feels-even-more-confident-about-eagles-kicking-fg-instead-of-going-for-it-in-loss-to-falcons/ ## Key Finding The absence of verified counter-examples is itself noteworthy. It suggests either: - Analytics models are genuinely better at these decisions than coaches - Media coverage is biased toward reporting conservative mistakes over aggressive ones - Coaches who override analytics successfully don't get written about because it doesn't fit the narrative - The bar for "demonstrably correct" is too high for single-game decisions ## Sources - NBC Sports (Dan Quinn) - GeekWire, Seattle Times (Mike Macdonald) - Philadelphia Inquirer, CBS Sports (Nick Sirianni)