College football temperature and scoring

Across 6,593 dry outdoor games from 2016 to 2025, temperature moves scoring in the direction everyone assumes — but only the cold end moves it much. Games below freezing score -5.2 points against the mild control (t = -3.08). Games at 85°F and above score +1.8 points (t = 2.40). Against the closing total, neither end shows an edge.

Freezing kickoff temperatures lower college football scoring by about -5.2 points compared with mild weather, a result that clears our pre-set significance threshold on 106 games. Heat of 85°F or above raises scoring only modestly (+1.8 points). Against the closing total, no temperature band produced a reliable edge — the market appears to price the thermometer correctly.

The smaller figure in each cell is the season-adjusted gap against the mild control, with its t statistic. Roughly 2 is our threshold. Only the freezing and hot bands clear it on points scored, and no band clears it against the closing total. Bands under 40 games are greyed out and flagged.

Scoring and closing totals drift year to year, and so does the weather mix of a schedule, which is why every comparison above measures a game against its own season rather than against the pooled sample.

That drift is large here and worth stating plainly: dry-game scoring falls steadily from about 57.9 points in 2016 to about 52.8 in 2025, a decline of roughly five points across the span. Because each game is measured against its own season's average, that is a level shift the adjustment removes rather than a signal. We checked it directly: run the same band comparisons with no season adjustment at all and the gaps move by no more than 0.2 points (freezing reads −5.2 adjusted against −5.3 unadjusted). The reason is that the temperature mix of the schedule is stable — the share of games in each band varies only a few percentage points year to year, with no trend that lines up with the scoring decline, so no band is quietly loaded into the high- or low-scoring end of the era.

One thing the split does show: the freezing effect is stronger in the recent half of the sample. Across 2016–2020 (51 freezing games) it reads −3.6 points with t = −1.44; across 2021–2025 (55 games) it reads −6.7 points with t = −2.94. Both halves point the same way and neither is large enough on its own to settle whether that is drift or noise, so we quote the pooled figure and will re-check the split as more freezing games land.

• Freezing is the finding. Under 32°F, games score -5.2 points against the mild control with t = -3.08, on 106 games — past our 40-game minimum, though still the smallest band here.
• Merely cold is nothing. The 32–49°F band (908 games) reads -0.8 points with t = -1.19. A jacket-weather forecast is not information.
• Heat helps a little. 85°F and above adds +1.8 points (t = 2.40), consistent with the overall +0.6 points per 10°F gradient.
• The market prices it. No band beats the closing total on our test, which is the opposite of what we found for heavy rain in college football. Temperature is priced; heavy rain is not fully.

• Sample: 6,593 dry outdoor college football games, 2016–2025, each with a final score, a closing total and kickoff-hour temperature, averaging 66.9°F. Indoor venues and closed roofs are excluded entirely.
• Wet games (0.25 mm or more of game-window rain) are excluded so the temperature result is not carrying our separately measured rain effect.
• Weather: hourly reanalysis at the venue's own coordinates, read at the kickoff hour. Archived model data, not a gauge reading at the stadium.
• Control group: 50–69°F. Every figure is season-adjusted — each game against its own season's average, for both points and margin against the closing total.
• Odds: closing lines come from a sports data feed rather than a single sportsbook. Games with no recorded line in any source are simply absent. Live in-play prices are excluded.
• The 40-game minimum and the significance threshold are house rules applied to every study before results are seen.
• Not controlled: team quality, pace and pass-run balance, wind, altitude, kickoff time of day (night games in September are cooler than the same venue at noon), and snow.

Data generated 2026-09-20. Numbers refresh as the archive tops up; corrections are dated and kept visible.

• Open-Meteo ERA5 historical reanalysis — archive-api.open-meteo.com (hourly game-window weather at each venue’s own coordinates)
• Open-Meteo geocoding API — open-meteo.com/en/docs/geocoding-api (venue coordinates and elevation)
• ESPN college football scoreboard and odds — espn.com/college-football (schedules, final scores and consensus closing totals)
• Internal college football game archive — venue coordinates, final scores and closing totals.
• Every dated change we have made to a published study is listed on our corrections log.