{
  "slug": "when-wind-changes-nfl-offense",
  "title": "Outdoor and open-roof NFL games at 15+ mph wind scored 3.8 fewer combined points than sub-10 mph games, 2018–2025",
  "dek": "Across 1,323 outdoor and open-roof NFL games from 2018 through 2025, those with reported wind at 15+ mph had lower neutral-down passing, completion rate, and pass efficiency than games below 10 mph. Above 20 mph, only 31 games exist.",
  "answer": null,
  "email_subject": "Outdoor NFL games at 15+ mph wind: 3.8 fewer combined points vs sub-10 mph, 2018–2025",
  "email_preview": "Across 1,323 outdoor/open NFL games from 2018–2025, 15+ mph wind came with lower dropback rate, completion rate, and pass EPA than sub-10 mph games. 20+ mph is still thin.",
  "series": "Fantasy Football Weather",
  "article_kind": "finding",
  "lead": "Fantasy managers tend to downgrade an entire passing game the moment a wind icon shows up on the forecast. A preregistered test at a single wind threshold shows the drop is measurable. Across 1,323 outdoor and open-roof NFL regular-season games from 2018 through 2025, games with reported wind of 15 mph or more averaged 42.3 combined points, against 46.1 in games below 10 mph. That 3.8-point gap carries a 95% interval from -6.2 to -1.3 points, and it arrived alongside less neutral-down passing, fewer completions, and pass efficiency that slipped from slightly positive to slightly negative. The 15 mph cut was locked before the results were run; it is a tested comparison point, not a breakpoint the data went looking for.",
  "blocks": [
    {
      "type": "table_ref",
      "text": "Games at 15+ mph (153 games, 143 for dropback rate) against games below 10 mph (871 games, 822 for dropback rate). All four registered outcomes were lower in the windy group, and none of the 95% bootstrap intervals include zero. These are unadjusted game-level differences.",
      "label": null,
      "artifact_id": "wind-primary-effects",
      "items": [],
      "metrics": [],
      "players": [],
      "callout_style": "note"
    },
    {
      "type": "heading",
      "text": "Efficiency moved the most, volume the least",
      "label": null,
      "artifact_id": null,
      "items": [],
      "metrics": [],
      "players": [],
      "callout_style": "note"
    },
    {
      "type": "paragraph",
      "text": "Passing efficiency shows the largest gap. Pass EPA per dropback averaged 0.056 in games below 10 mph and -0.011 at 15+ mph, a -0.067 difference with a 95% interval from -0.104 to -0.031. Completion rate went from 60.3% to 58.0%, an interval of -3.4 to -1.1 percentage points. Play-calling moved too, but less: on neutral early downs, offenses dropped back on 52.1% of snaps in 15+ mph games versus 54.0% in calm ones, a -1.8 point gap with an interval from -3.4 to -0.4 points. The volume difference is supported, but it is the smallest of the four.",
      "label": null,
      "artifact_id": null,
      "items": [],
      "metrics": [],
      "players": [],
      "callout_style": "note"
    },
    {
      "type": "paragraph",
      "text": "The registered adjusted model, which treats wind as a continuous variable and controls for temperature, week, and season, points the same direction on all four primaries. Combined points fell an adjusted 1.39 per 5 mph of reported wind (95% interval -2.26 to -0.52), and pass EPA per dropback fell 0.020 per 5 mph (-0.033 to -0.008). Venue, opponent, and offensive quality remain uncontrolled in both versions, so some of the gap may still reflect who was playing rather than the wind.",
      "label": null,
      "artifact_id": null,
      "items": [],
      "metrics": [],
      "players": [],
      "callout_style": "note"
    },
    {
      "type": "paragraph",
      "text": "One expected pattern did not appear. Average target depth was computed in the locked analysis as a secondary outcome, not one of the four registered primaries, and it showed no supported difference at the 15 mph threshold. Offenses in windy games completed fewer throws and got less out of each dropback, but this sample offers no evidence that they systematically shortened their passes to compensate.",
      "label": null,
      "artifact_id": null,
      "items": [],
      "metrics": [],
      "players": [],
      "callout_style": "note"
    },
    {
      "type": "figure_ref",
      "text": "Neutral early-down dropback rate and pass EPA per dropback by reported wind bin. The 20+ mph bin rests on only 31 games and should not be read on its own.",
      "label": null,
      "artifact_id": "wind-offense-by-bin",
      "items": [],
      "metrics": [],
      "players": [],
      "callout_style": "note"
    },
    {
      "type": "heading",
      "text": "What this means for a lineup, and what it does not",
      "label": null,
      "artifact_id": null,
      "items": [],
      "metrics": [],
      "players": [],
      "callout_style": "note"
    },
    {
      "type": "paragraph",
      "text": "Treat this as a population-level nudge rather than a lineup command. On average, a 15+ mph game came with roughly four fewer combined points, a modest cut in neutral-down passing, and a real dip in completion rate and efficiency. That supports shading a passing game's expectations down when reported wind reaches that range; it does not justify benching a starter on wind alone. Three scope notes matter here. The wind figure is a single reported game value, not gusts or in-game variation, and a forecast checked on a phone app before kickoff may not match the value recorded for the game. Games at 10 mph or more but below 15 mph sit in neither group of the headline comparison, so that comparison says nothing about the middle band fantasy players face most often. And the headline table reports only the 15 mph cut against calm games, so on its own it cannot separate a step change at 15 mph from a gradual slide; the registered 10 and 20 mph thresholds and the piecewise-linear wind spline were run as sensitivity checks but are not summarized in this article.",
      "label": null,
      "artifact_id": null,
      "items": [],
      "metrics": [],
      "players": [],
      "callout_style": "note"
    },
    {
      "type": "callout",
      "text": "Only 31 games in the sample reached 20+ mph, so the extreme-wind estimate is underpowered and is not a standalone threshold. These are retrospective associations from reported game-day conditions, with no adjustment for venue, home team, or offensive quality in the headline comparison; windy games cluster at particular stadiums and late-season dates. They do not establish that wind caused the drop, and they do not show that a pregame wind forecast improves projections, since observed wind is only known after kickoff.",
      "label": "Where the evidence stops",
      "artifact_id": null,
      "items": [],
      "metrics": [],
      "players": [],
      "callout_style": "caveat"
    }
  ],
  "methods": [
    "Population: NFL regular-season games at venues recorded as outdoors or open with reported wind and temperature, 2018 through 2025; 1,323 of 1,487 outdoor/open schedule games had complete weather (89% coverage). Outcomes with too few plays in a game were set to missing, which leaves 1,251 games for neutral early-down dropback rate.",
    "Primary comparison, locked before results: games at or above 15 mph versus games below 10 mph on four registered outcomes: neutral early-down dropback rate, completion rate, pass EPA per dropback, and combined points. Games at 10 mph or more but below 15 mph are in neither group.",
    "Neutral early-down plays were first or second down with win probability between 20% and 80%, score within 14 points, and more than two minutes remaining; kneels, spikes, aborted, and deleted plays were excluded.",
    "Uncertainty from 4,000 game-level bootstrap draws. Holm adjustment was applied across the five threshold outcomes computed in the locked analysis (the four registered primaries plus average target depth, a secondary outcome computed in the locked code but not among the registered primaries), which is conservative for the four primaries.",
    "Registered baseline: a continuous per-5-mph model adjusted for temperature (linear and quadratic), week, and season fixed effects with HC3 robust intervals. The reported table figures are unadjusted; the adjusted per-5-mph estimates are cited in the body.",
    "Sensitivity checks included 10 and 20 mph thresholds, a piecewise-linear wind spline with knots at 10, 15, and 20 mph, and leave-one-season-out direction stability, which passed. The alternate-threshold and spline estimates are not summarized in this article.",
    "Data made available by nflverse contributors: schedules and play-by-play, CC BY 4.0."
  ],
  "limitations": [
    "Associations from observed game conditions, not causal effects and not an evaluation of pregame wind forecasts.",
    "Neither the headline comparison nor the adjusted model controls for venue, home team, opponent, or offensive quality; windy games are concentrated at particular stadiums and late-season dates, so some of the gap may reflect who was playing rather than the wind.",
    "Only 31 games reached 20+ mph, so extreme-wind estimates are underpowered and not publishable as a standalone threshold.",
    "Reported weather coverage was below 85% in the 2022 and 2023 seasons, which limits how representative the complete-case sample is for those years.",
    "Average target depth was a secondary outcome computed in the locked code but not one of the four registered primaries.",
    "The headline threshold comparison alone does not distinguish a step change at 15 mph from a gradual decline, and games at 10 mph or more but below 15 mph were excluded from it.",
    "Wind is a single reported game value, not gusts or in-game variation."
  ],
  "data_through": "2025 NFL season",
  "reading_minutes": 4
}