My last post looked at the effect of temperature on home field advantage. We saw that cold weather put dome and warm climate teams at a disadvantage. The post was titled How Does Temperature Affect Road Teams?, but I really didn't answer that question. I measured the size of the effect, but I didn't solve the riddle of actually how temperature makes a difference. This post will begin to look at just how weather makes a difference, starting with the passing game.
Here's how passing fares for home and visiting teams by temperature. The chart below shows Adjusted net Yards Per Attempt (AYPA), which accounts for sacks and interceptions, according to temperature. Keep in mind there are smaller sample sizes at the extremes.
Home teams have an advantage across the temperature spectrum, which appears to be exaggerated at extreme temperatures. This chart included playoff games. In the last post, commenter Matt pointed out that very cold games are disproportionately playoff games, which would bias the numbers in favor of home teams. This bias is due to the fact that better teams are usually seeded higher and therefore usually at home. I removed playoff teams and the difference between home and visitor AYPA below 30 degrees is actually slightly bigger. The visitor AYPA is virtually the same, but the home AYPA is about 0.2 or 0.3 yards per attempt higher than this chart above shows.
The big takeaway should be that temperature does in fact hurt the passing game. Does it hurt dome and warm climate teams more than others? The chart below tells the tale. It's noisier than we'd like, but one thing we can tell is that the colder the weather the lower the passing efficiency for all types of team climates. Keep in mind there are very few games in the extreme bins, only a total of 36 games in the 11-20 degree bin, and only 15 games in the 91-100 bin.
I wouldn't be too quick to dismiss the drop off in passing numbers in the 91+ degree bin. It's not statistically significant, but with more data it may appear to be a real effect.
I always thought that it was really wind that affected the passing game the most, and not temperature. They often go hand-in-hand in places like Chicago, Cleveland, and Buffalo, so it's difficult to parse them apart. Here is AYPA according to wind speed. It's actually the minimum wind speed reported in the official gamebook. Often, the gamebook will report winds as something like 10-15 mph, or 10 gusts to 15. In this case, I took '10' as the wind speed.
Wind certainly has an effect. Let's take a look at how it affects visiting teams by climate type. Are dome teams victimized as much as we'd expect?
Nope. It looks like the dome teams aren't affected any worse than the other teams. We can safely conclude that it's the temperature and not the wind that causes the dome at cold effect. Wind appears to be an equal-opportunity hazard.
So do teams adjust for the weather conditions. Do they bias their play selection toward the run in cold conditions, as they should?
Yes, they do tilt more toward the run in cold weather, but very slightly. The bigger effect is that there are fewer plays of both types. I suppose more runs means less plays overall because they tend to use more time. Perhaps there are more punts and field goal attempts because passing effectiveness drops.
How does wind affect play calling?
Teams are definitely leaning to the run on windy days, but the effect is only prominent in significantly windy conditions, greater than 15 mph. And even in winds of over 20 mph, offenses are trading only about 5 passes for runs in an entire game, or about 2-3 per team.
Weather Effects on Passing
By
Brian Burke
published on 1/13/2012
in
research,
run-pass balance,
weather
Subscribe to:
Post Comments (Atom)
One of the more thought provoking articles I have read in awhile. Thanks.
Is the weather data available?
Awesome stuff.
So what is it about the cold that negatively affects dome team's passing? Lack of attempts, more conservative playcalling, inability to throw/catch the ball, poor traction/visability, reduced cognative skills due to cold? Is there a distinction between just being cold versus cold AND snowy? What are the differences between snow and rain games?
I LOVE this article and need more of it.
Thanks!
I predict that one of the factors is that dome/warm teams overcompensate for non-windy snowy conditions. In these sorts of conditions, while it is harder to catch and throw the ball, it is also much harder to rush the QB and to cover wide receivers (it is also harder to run block!). The result is that the correct strategy in snow with little wind conditions is to pass the ball alot, and I don't think warm/dome teams do this (they assume snow = can't pass).
Could these results be skewed by the correlation between games played at a certain temperature and the passing ability of the teams who play more in colder temperatures?
For example, I imagine having Jacksonville/Tampa/Miami in the set of teams who play a lot of high-temperature games has a negative impact on the AYPA of hot-weather games.
Perhaps a more meaningful metric would be (Game AYPA) - (Average YPA for that team that season) to weed out some of the effects of that correlation.
Brian, Kam Chancellor leads all safeties in +WPA, +EPA, and +EPA/G. What does this tell me statistically, and what does it tell me about his impact on the field? Your help would be appreciated.
Thanks
When I did a similar study on wind speed earlier this year I found almost the same finding. Significant changes occur over the 15 MPH threshold - points, pass attempts, pass yards go way down and rushing attempts go way up. I wonder if this is out of necessity or relates more to groupthink by coaches. Great stuff.
hi. It's precipitation that affects the passing game.
Wind will have an effect, but only starting at 'quite breezy' and escalating quickly up to 'hurricane'.
If you entered all that weather information manually from game books, I'm impressed.
Love what was said about overcompensating for snow. On thing to add to the analysis: even many cold weather teams often move indoors or to a practice bubble when the weather gets cold, so it may be misleading to automatically categorize a team based on where they play. Don't know if that was taken into account. I would theorize that a team like the Patriots who practice outside all year would have an advantage in adverse conditions versus what you'd expect them to have (though they are recently 0-2 at home in the platoffs in recent years, that's a small sample).
So what is it about the cold that negatively affects dome team's passing? Lack of attempts, more conservative playcalling, inability to throw/catch the ball, poor traction/visability, reduced cognative skills due to cold?...
I don't know, but I do know what I once heard Jimmy Johnson say.
He was asked if playing in the cold really hurts the play of warm weather teams like his were. He said "It sure does." Then he was asked "Why?"
He said: "I remember being on the sideline in winter in the Meadowlands and it was so cold I didn't know what down it was, much less what play to use."
So I think he'd vote for reduced cognitive skills.
1) As someone else said, unless you adjust for team quality, there is a problem with confusing temp. effects with warm/cold-weather team quality effects. If cold weather teams have been abnormally good at pass defense and bad at pass offense historically, for example, you would not detect this bias.
2) Is there a correlation between temp. and precipitation? If so, temp. may not be a big factor.
3) The idea that your data shows that dome teams are abnormally affected by cold temps. is absurd. You have one outlier data point (11-20 degrees) based on a tiny sample. And within that tiny sample, you have a selection bias since only certain opponents have weather that is even capable of being sub-20 deg.
4) I think the most pressing/obvious remaining question is what happens in cold temps. with low wind, and in high temps. with high wind? I would be shocked if it was actually the temp. having anywhere near as strong an effect as the wind.
In summary, when you are dealing with small samples (as you always are in NFL), you cannot fail to adjust for opponent quality when trying to isolate environmental effects.
Even if team quality biases the numbers, you can easily compare how dome teams do when the weather is warmer compared to colder temps.
"3) The idea that your data shows that dome teams are abnormally affected by cold temps. is absurd."
When someone writes something like this, it's a dead give-away that he is a socially-retarded know-it-all. He hates not being the smartest guy in room, and loves reading his words on the Internet.
In rebuttal, the sample sizes are not enormous, it's true. However, the results are as robust as they get. There is a clear trend, not only across temperatures, but across team types as well. If you want to play ignoramus and claim there is no environmental effect on dome teams, that's your choice.
If you're an idiot, like this guy, here's some advice: Don't write obvious 'tells' like "the idea that X is absurd." Maybe it's not yet proven, or established, or more research is warranted. But it's certainly not "absurd." Only an idiot would claim that.
So is the weather data available?
2 questions:
(1) as previous commenters ask: rain?
(2) error bars?
Can you guys tell us how accurate will be passing in rain