NFL temperature and scoring

Across 3,474 outdoor games from 2006 to 2025 — 2,695 of them dry — temperature barely touches NFL scoring. From freezing to 85°F every band sits within about a point and a half of the mild 60–70°F control, and not one of those gaps is statistically significant. The slope across the whole dry sample is about 0.2 points per 10°F. Severe cold below 20°F is the only band that looks different, and it is too small to price.

NFL scoring is flat across almost the entire temperature range football is played in. A forecast of 34°F is not, by itself, a reason to expect a lower-scoring game, and nor is 86°F a reason to expect a higher one. Only severe cold under 20°F shows a drop, on 40 games — exactly the pre-set 40-game minimum, so directional evidence only. Wind, which we publish separately, is the outdoor condition that actually moves NFL totals.

The smaller figure in each cell is the season-adjusted gap against the mild control, with its t statistic. Anything under roughly 2 is noise. Bands under 40 games are flagged thin. The severe-cold row's n=40 merely clears that pre-existing minimum; it is shown for direction only, never as a priceable signal.

Points only tell you what happened; the line tells you what was expected. Where ESPN carried a closing total we compared actual scoring against it — 916 dry games in all, concentrated in the more recent seasons.

Bands with fewer than 40 games carrying a closing total are left blank rather than quoted — including severe cold, where only 2 games have a line. No band with an adequate sample lands more than about a point from the closing number, and no t statistic approaches significance. Under rates are shown for completeness, not as a record to bet.

Scoring drifts from year to year, and the share of cold games in a season is not constant either. That drift is exactly why every comparison above measures a game against its own season rather than against the twenty-year pool; we checked the adjusted and unadjusted band gaps against each other and the conclusions are the same.

• Ordinary cold is not a weather game. Freezing, cold and chilly games all score within about a point and a half of mild games, with t statistics well inside noise.
• Heat does nothing measurable either. The 80–85°F and 85°F-plus bands are flat against the control, unlike college football, where heat shows up.
• Severe cold is the only candidate. -5.2 points on 40 games (t = -2.14) — suggestive, at the pre-set minimum sample, and with only 2 games carrying a closing total, so no market claim is possible.
• The market is already there. Every adequately sampled band finished within about a point of its closing total. There is no temperature edge to take.
• Wind, not temperature, is the NFL signal. This study adds no temperature coefficient to our live model. That is a deliberate null, recorded on our accuracy page.

• Sample: 3,474 outdoor NFL games, 2006–2025, with resolved venue coordinates, kickoff temperature and a final score. Domes and closed roofs are excluded entirely rather than used as a control. The headline bands use the 2,695 dry games, defined as under 0.1mm of rainfall across the game window.
• Weather: hourly reanalysis at the stadium's own coordinates. Temperature and wind are read at kickoff; rainfall is summed across the game window.
• Control group: 60–70°F. Every figure is season-adjusted — each game against its own season's average — and the wet-inclusive column repeats the comparison on the full sample with its own control.
• Closing totals come from ESPN's odds feed and only cover part of the window, so the market table is a smaller and more recent sample than the scoring table. Bands below the 40-game minimum are not interpreted.
• Temperature is a kickoff reading, not a whole-game one; a late-afternoon game that cools sharply is recorded at its warmer start. Wind is reported as a covariate here and studied properly in the NFL wind and totals study.
• Not controlled: team quality, pass-run balance, wind chill as distinct from air temperature, snow, playoff intensity, and roof state at retractable venues where no per-game feed exists.

Data generated 2026-09-22. Numbers refresh as the archive tops up; corrections are dated and kept visible.

• NFL Wind & Totals Study — the weather variable that does move NFL scoring.
• NFL Climate Shock Study — dome and warm-weather teams playing outdoors in cold and wind.
• NFL Turnovers in Bad Weather — rain, cold and ball security over 20 seasons.
• Forecast accuracy — where this deliberate null is recorded against the live model.

• Open-Meteo ERA5 historical reanalysis — archive-api.open-meteo.com (hourly game-window weather at each venue’s own coordinates)
• ESPN NFL scores and odds — espn.com/nfl (schedules, final scores and consensus closing totals)
• Internal NFL historical game archive — venue type, kickoff weather and final scores.
• Internal closing-total archive derived from ESPN odds for the seasons it covers.
• This study has been corrected since first publication. Every dated change is listed on our corrections log.