Research
Crosswind, kicking and totals in college football
Wind is the most talked-about weather factor in football and the most sloppily measured. We resolved every game's wind into the component blowing across the field using each venue's true field bearing, then measured field-goal accuracy and final scoring against a calm-conditions baseline built from the same sample.
Correction
Updated 3 September 2026
Our original Strong crosswind headline (+5.3 pts) was calculated from distance bands too small to be reliable (8–26 attempts each, below our own 40-attempt minimum). Recalculated — see below.
The headline
At sample sizes we trust, crosswind doesn't move field-goal accuracy in our CFB data — the popular “wind kills kicking” narrative doesn't show up here (yet).
Every crosswind band with an adequate sample lands within about a point of the calm expectation: Calm (0–5 mph cross) -0.1 pts (n=6,862), Light (5–10) -0.2 pts (n=2,959), Moderate (10–15) 0.0 pts (n=559). We'll keep grading Strong crosswind games as they accumulate.
Strong crosswind kicking
Insufficient data
81 attempts across four distance bands, none individually clearing our 40-attempt minimum sample threshold.
We'll report a headline figure here once enough games have accumulated.
Strong crosswind totals
+0.6 pts
average final score vs closing total
Under hit 50.0% of graded games
Sample
10,817
field-goal attempts graded
across 3,643 completed games
Field-goal accuracy vs the calm baseline
Each cell shows the make rate in that crosswind band, and how far it sits from what the same distance band produces in calm conditions. Cells below the minimum sample are greyed — we publish them for transparency, not as findings.
| Crosswind | Under 30 yds | 30-39 yds | 40-49 yds | 50+ yds | All (mix-adjusted) |
|---|---|---|---|---|---|
| Calm (0–5 mph cross) | 92.2%-0.2 · n=1860 | 81.2%-0.3 · n=2117 | 65.0%0.0 · n=2140 | 48.2%+0.2 · n=745 | 75.5%-0.1 · n=6862 |
| Light (5–10) | 90.5%-1.9 · n=829 | 80.7%-0.8 · n=918 | 66.0%+1.0 · n=899 | 50.5%+2.5 · n=313 | 75.8%-0.2 · n=2959 |
| Moderate (10–15) | 90.8%-1.6 · n=163 | 78.3%-3.2 · n=180 | 67.7%+2.7 · n=158 | 55.2%+7.2 · n=58 | 76.6%0.0 · n=559 |
| Strong (15+) | 96.2%+3.8 · n=26 | 73.1%-8.4 · n=26 | 81.0%+16.0 · n=21 | 75.0%+27.0 · n=8 | 82.7%+5.3 · n=81 |
Scoring vs the closing total by crosswind
| Conditions | Games | O–U–P | Under rate | Avg vs total |
|---|---|---|---|---|
| Indoor | 41 | 17–23–1 | 57.5% | -1.5 pts |
| Calm (0–5 mph cross) | 970 | 477–482–11 | 50.3% | -0.2 pts |
| Light (5–10) | 409 | 194–209–6 | 51.9% | -0.2 pts |
| Moderate (10–15) | 87 | 41–45–1 | 52.3% | -1.6 pts |
| Strong (15+) | 16 | 8–8–0 | 50.0% | +0.6 pts |
Method
- Completed FBS regular-season games are pulled from the ESPN scoreboard and matched to our venue profiles for coordinates, roof type and field bearing.
- Kickoff-hour temperature, wind speed, wind direction and precipitation come from the Open-Meteo ERA5 reanalysis archive — the same source behind our NFL research.
- Crosswind is the wind vector resolved perpendicular to the field's long axis, so a pure headwind or tailwind registers as zero crosswind.
- Indoor games and games under 5 mph crosswind form the calm baseline. Each wind band's expectation is that baseline reweighted to the band's own distance mix, which removes the bias created by coaches declining long kicks in bad weather.
- Totals are graded against the ESPN consensus closing over/under. Pushes are excluded from the under rate but counted in the game total.
References & data sources
- Open-Meteo ERA5 historical reanalysis — archive-api.open-meteo.com
- ESPN college football scoreboard, game summary and pickcenter endpoints
- OverUnderWeather venue profiles — field bearings, roof type and coordinates
Model cfb-wind-v1 · last recomputed 2026-08-31
Frequently asked
Why crosswind instead of wind speed?
A 20 mph wind straight down the field is a very different kick to a 20 mph wind blowing across it. Every venue in our database carries a field bearing — the compass direction of the field's long axis — so we resolve the reported wind into the component that actually pushes the ball sideways. Raw wind speed blurs the two together.
How do you handle coaches avoiding long kicks in bad weather?
That selection bias is exactly why a raw make-rate-by-wind table is misleading: windy samples skew short, which flatters the make rate. We build an expected make rate per distance band from the calm sample, then report how far each wind band lands from that expectation given its own distance mix.
Is the wind measured at the moment of the kick?
No. Historical play-level wind is not available, so every attempt inherits the kickoff-hour ERA5 reanalysis reading for its venue. That adds noise but no directional bias, and it is the same weather source used across the rest of the site.
Is this a betting system?
No. It is a conditions study. Buckets below the minimum sample are marked as such and we do not draw conclusions from them. Check the weather before you bet — the numbers here describe how conditions have historically bent scoring and kicking, not what to wager.