Research
Heat, humidity and travelling teams in college football
College football opens in the hottest fortnight of the year, and Northern programs regularly fly into 90F, high-dew-point Southern venues. We scored every August and September road game of the last ten seasons for how far outside the visitor's own climate the conditions were, then graded the results against closing spreads and totals.
The headline
Shocked visitors ATS
48.0%
cover rate vs 53.2% for control games
319 graded games with a closing spread
Shock games under rate
55.2%
vs 51.6% in control games
270 graded games with a closing total
Sample
2,020
outdoor Aug–Sept road games, 2016–2025
473 of them registered a heat shock
Results by shock bucket
| Bucket | Games | Visitor ATS | Cover rate | Under rate | Avg points |
|---|---|---|---|---|---|
| No shock (control) | 1547 | 523–461 | 53.2% | 51.6% | 54.8 |
| Mild shock | 282 | 89–110 | 44.7% | 53.0% | 56.7 |
| Moderate shock | 94 | 38–28 | 57.6% | 52.5% | 54.3 |
| Strong shock | 97 | 26–28 | 48.1% | 66.0% | 53.9 |
Cover rate is the visiting (travelling) side against the closing spread. Pushes are excluded from every rate but counted in the game column.
What the numbers say
The travelling side covers less often when it walks into heat well outside its own climate — 48.0% across 319 graded shock games against 53.2% for the 984 control games in the same window. That is a real gap in the right direction, but on a sample this size it is a lean rather than a finding you should lever up on.
Totals lean under, mildly: 55.2% across 270 graded shock games against 51.6% in control games, with average scoring a shade lower. The strongest bucket is where the effect concentrates, but it carries the fewest graded games of any row in the table, so treat it as a hint rather than a result. Unlike NFL cold — where the under edge was large and stable across ten seasons — late-summer heat produces a soft signal at best, and we publish it exactly as it came out.
Method
- Completed FBS regular-season games in August and September, 2016–2025, taken from the ESPN scoreboard.
- Indoor and roof-closed venues are excluded outright, as are neutral-site games where neither side is genuinely at home.
- Kickoff-hour temperature, dew point, apparent temperature and humidity come from the Open-Meteo ERA5 reanalysis archive at each venue's coordinates.
- Each program's climate baseline is the ten-year mean temperature and dew point at its own home venue across the same mid-August to end-September window and the same evening kickoff hours.
- Shock score = heat gap above 8F plus dew-point gap above 5F, and only when the venue itself feels 85F or hotter. Buckets: mild under 1.5, moderate under 3, strong at 3 and above.
- Spreads and totals are the ESPN consensus closing numbers. Games without a closing line are counted in the sample but not graded.
References & data sources
- Open-Meteo ERA5 historical reanalysis — archive-api.open-meteo.com
- Open-Meteo geocoding API for venue coordinates and elevation
- ESPN college football scoreboard, game summary and pickcenter endpoints
- Companion research: NFL Climate Shock study (cold, one-way, same design)
Model cfb-heat-shock-v1
Frequently asked
What counts as a heat shock?
Three things have to line up. The venue must be genuinely hot at kickoff — an apparent temperature of at least 85F. The visiting team's late-summer home climate must be materially cooler, more than 8F below the venue reading. Humidity adds on top when the venue dew point is more than 5F above what the visitor is used to. A Southern team playing in Southern heat scores zero, because it is acclimatised.
Why only the visiting team?
The home side is by definition adapted to its own climate, so there is no shock to measure. The model is deliberately one-way, the same design as our NFL cold-weather Climate Shock study.
Where do the climate baselines come from?
Each program's baseline is computed automatically from ten years of Open-Meteo ERA5 reanalysis at its own home venue, restricted to the mid-August to end-September game window. Nothing is hand-entered, so no team is classified by reputation.
Is this a betting system?
No. It is a conditions study on a limited sample and it is published exactly as the numbers came out. Check the weather before you bet — these tables describe how heat and humidity travel has historically lined up with results, not what to wager.