How does wind affect a college football game (and its over/under)?

What effect does wind have on the game? It only truly affects two things — the ball in the air (passes and kicks) and decision making (play calling, fair catches, etc). Sometimes it can really affect these things, like this kick, or deciding to choose which side of the field to defend instead of receiving the ball in sudden death overtime (thanks, Marty!).

Save for those and some other extreme examples, though, the effect on any single play is small, even imperceptible. But after adding all those little things up the impact can be much more than one might think.

Many bettors will hear all the touts saying bet under on windy games. This article contained an extremely brief analysis (if you can even call it that) claiming there was some value in unders on windy NFL games. Is that really true for College Football? I set out to find out. There have been almost 10,000 college football games since 2005, and using Weather Underground’s free API, I was able to scrape weather data for 7,323 of them.

Let’s start simple with a table showing over/under record for all games and games with or without much wind. A lot of analyses tend to use 10mph as the cutoff for value; I tend to think it’s closer to 12mph, but whatever (note, whenever I cite a wind value, I’m using the average wind speed for the entirety of that game at the nearest weather station to the stadium site; thankfully most stadiums have their own stations).

A rate of 54.7% is certainly high enough to net you some cash, but there are some caveats.

• Is the result statistically significant? Turnovers are unpredictable and can create huge swings in games and points — that typically makes things like this pretty “noisy”.
• Also, how much do you trust weather predictions? I’ve had all my apps and local TV weather people saying it would rain all week, and nary a drop. Whatever weather prediction you distrust the least will have some variability. If a prediction calls for 12mph winds, ending up with 8mph winds or something lower is not rare, and with it your play goes from valuable to a coin toss.

As wind increases over 10mph, do we expect to have better and better chances at hitting an under? Yes! The red line below represents the percentage of games that were under when then wind is greater than the x value.

Here’s what that means: for ALL games greater than 10mph, about 55% of them are under, for all games over 15mph about 58% are under, all games above 17 mph will have 60% under the total. Beyond 17mph you start to lose reliable sample size (only 250 games).

Let’s look at this another way. Instead of just using was the game over or under, let’s examine the predicted total points (the Vegas line … soon to be the “anywhere” line) versus the actual points scored.

Do we expect the total points in a game to get further and further away from the prediction (the actual line or modeled value) as the amount of wind increases? The answer is yes, but here is where you’re going to see all that noise I mentioned above:

Modeling the deviation of actual points scored versus the Vegas total line with average wind is statistically significant but noisy.

If you were to use this as your sole betting system, you would expect to win over the long haul, with 1.5 points of value per 10mph of wind. But on a smaller sample size of, say, 10 games, you’re nearly as likely to go 5-5 as you are to go 6-4. Put another way, as wind increases, we can generally expect totals to go further and further under the Vegas line. Just don’t be surprised when the prediction for a single game is off by 20 points. Football is a finicky game.

So yes, wind matters. But why? What aspect of the game does it most affect? Are teams with passing offenses more susceptible to the wind than a run-heavy team like Navy? Do teams pass less often? Are teams less successful when passing in the wind? Are there more interceptions thrown? Do teams take less chances downfield and throw shorter passes in the wind? Are field goals less successful or attempted less?

To spare you scrolling through eight charts that all look the same, I’ll just tell you that the answer is… all of the above.

Essentially all of those questions are answered how you think they would be, just with very very small effects that slightly increase as you increase the wind; call it death by a thousand paper cuts.

• Teams on average call two percent fewer pass plays in the wind.
• Teams are very slightly less successful throwing in the wind. Interception rate is slightly higher.
• Teams on average throw shorter passes.
• Fewer field goals are attempted and the success rate decreases (this was more noticeable than the passing stuff).
• More fair catches are called on punts.
• And so on, and so on.

One surprising thing is that high winds didn’t seem to affect passing teams scoring more than rushing teams, nor did windy games involving two passing teams have a better chance of going under the total line. I think this is explained by the fact that there are more incompletions in these windy games and more clock stoppages, which in turn leads to more plays and possessions and scoring chances. So that surprising result kind of cancels itself out.

In the end, all these small effects accumulate to generate a slightly depressed scoring environment in games with wind above 10ish mph. That depressed environment creates some value on betting unders — just watch those forecasts and look out for late moves against the under.

And don’t blame me when you get moosed with some noise — I warned you.