The Fantasy Football Mythbusters: Whether Weather Really Matters
We are beset with tidbits of ‘conventional wisdom’ in our everyday lives. For example, everyone knows that ‘a penny saved is a penny earned’ and ‘little strokes fell great oaks’. Fantasy football, in the final analysis, is no different. Every manager has internalized various adages over the years that are generally considered fact by the broader fantasy community.
In this series, we challenge the status quo and set out to discover if there is any truth to these long-held wisdoms. More succinctly, in the style of the classic show, we are Mythbusting. This first entry will focus on weather.
In general, fantasy managers are told to avoid games with bad weather. A recent example that comes to mind, if you listen to The Fantasy Footballers Podcast, is when Jason suggested pivoting away from the Week 10 matchup between the Houston Texans and Cleveland Browns last year. As it turned out, this was prudent: the Browns slogged out a 10-7 win among all sorts of precipitation and extreme gusts of wind.
Still, a lot of questions naturally arise. Does weather really play a significant role, or was this just a one-off? Is there a ‘breaking point’ at which the weather gets too severe? How are different facets of the game affected?
Let’s dive in. As usual, data comes from nflfastr, and the games in this article are from 2015 onward.
The simplest place to start is looking at a variety of key offensive metrics in games with either snow or rain. This is pretty straightforward since the breakdown is binary: it’s either raining (or snowing) or it’s not. We’ll consider the game averages of both teams combined, or the total game production. Let’s start with rain:Rain clearly has a negative effect on the passing game. When it’s raining, completion percentage goes down by nearly 3%, total game passing production falls by 45 yards, and .6 fewer passing touchdowns are thrown. In these conditions, teams attempt about two fewer passes per game (each).
Rushing production is slightly higher in the rain, but not by much: teams post 15 more combined yards on the ground and .1 more rushing touchdowns per game while averaging about 1.5 more rushes each.
These effects are, as you might guess, more extreme in the snow:
Passing attacks are significantly hampered when the white stuff is falling. Teams attempt nearly five fewer passes per game each, completion percentage falls by 7%, total passing production declines by 110 yards, and 1.1 fewer passing touchdowns are thrown. Air yards take a hit too, with 60 less per game (combined).
From a fantasy perspective, that’s a difference of -2.2 points from both yardage (55 fewer yards per team, times .04 points per yard) and touchdowns (.55 fewer touchdowns total per team, times 4 points per TD) for a total of -4.4 points expected per game for QBs (this would be more negative for WRs since reception yards and TDs are worth more).
Unlike the rain example, rushing attacks do significantly better in the snow. Teams post 45 more combined rushing yards per game and nearly one full rushing touchdown more; this yields +4.3 fantasy points expected per game for rushing attacks. This appears to be largely a product of play-calling: teams rush more in the snow, to the tune of about 2.5 more rushes per game. Note that the number of rushing attempts increases less than the number of passing attempts decreases: of course, in general, rushing takes more time off the clock, so you can fit fewer rushing attempts in a game than passing attempts!
Passing ‘heat maps’ (here, we plot completions) tell the same story through a different lens. This coordinate data is thanks to Ethan Douglas and Sarah Mallepalle et al. (you can find the data here), while the base code for the charts is from Thomas Mock.
In the snow, passing attacks are limited both vertically (far less deep throws) and horizontally (less heat around the edges of the field). We saw how well Baltimore defended from aerial attacks in the map from our Fantasy Historian series; it turns out snow is almost as good a defender as the Ravens!
Key takeaways: do your best to avoid QBs/WRs, and target RBs, in rainy and snowy games. Don’t hear what I’m not saying: it doesn’t make sense to sit Patrick Mahomes for a waiver-wire QB just because the Chiefs are playing in the rain. However, if you have relatively similar options, pick the one with the better weather forecast!
Temperature and Wind
Now that we’ve covered things falling from the sky, let’s take a look at other common manifestations of Mother Nature. To start, let’s consider these temperature charts. We round the temperature to the nearest 10 and take the average value (pass attempts, pass yards etc.) of games at that level of heat. This data is mostly for outdoor games since games in roofs/domes generally have sparse wind and temperature tracking.
We see a similar pattern: passing attacks are stronger in the heat, rushing attacks fare better in the cold. There are a few quirks of the data, but it’s clear that pass completion, passing yards, and touchdowns are all higher once the temperature rises, while the reverse is true for rushing.
Let’s consider wind:
Again, as is to be expected, passing is more difficult on windier days, and rushing attacks generally pick up the slack. What’s interesting is that passing attempts fall off significantly during windy games, which was not the case for cold games (teams pass just as much in cold games and still see less production).
Key takeaways: Similar to the rain and snow, try to target QBs/WRs with warmer weather and less wind, and vice versa for RBs. Do note, though, that the effects of wind and temperature are not as large as the effects of rain and snow.
Raise the Roof
Finally, many NFL franchises essentially control the weather with domed stadiums. Of course, this means milder conditions and thus, as one would expect, better passing attacks. Let’s look at the breakdown, which confirms that passing is more effective under roofs:
However, there’s something a bit more nuanced here; specifically, it might not be the roof that is causing this difference. Instead, it could be the quality of the teams that generally play in domed stadiums.
Over the past six years, the New Orleans Saints, Houston Texans, Dallas Cowboys, Detroit Lions, Arizona Cardinals, and Atlanta Falcons have led the league in ‘domed games’. These are all teams with tons of offensive talent (or, at least, in the Lions’ case, solid offensive talent). Meanwhile, traditionally anemic offenses like the Browns, Bengals, and Jets play under domes much less.
Therefore, it could be simply that the teams that happen to play under roofs are far better. A good way to check this is with regression analysis: we include predictive variables both for the existence of a roof and individual variables for each of the 32 NFL teams. After building the model, the additional passing yards expected from being under a roof (controlling for the team that is playing) are just two yards. What’s more, it’s not even statistically significant!
Key takeaways: The roof conclusion is nuanced. We’ve already seen how warm, mild weather is better for passing attacks, and roofs essentially guarantee this type of weather. Therefore, compared to a snowy, rainy, cold, or windy game, a roof game is certainly preferable; however, there is no significant difference between a mild weather outdoor game and a roof game. This means that, in a redraft league, it could be useful to target QBs/WRs that play in domes (vice versa for RBs) since you are guaranteed more mild weather games throughout the season. However, in a single week, it doesn’t make sense to pivot to a ‘roofed’ fantasy option vs. a player in a mild-weather outdoor game.
A quick note: I did the same test with the rain, snow, temperature, and wind, and found that, even after controlling for the teams playing, the weather effects were still significant.
This one is CONFIRMED! We see moderate, statistically significant deterioration in passing attacks in the rain, snow, high wind, and low temperature; rushing attacks improve in these tough conditions. This difference is the most significant in snowy conditions, but there is still a measurable change in rainy, cold, or windy games. It turns out that the old adage is true: be sure to check the weather before your Sunday slate!
Did I miss anything? Curious to test another miss? Message me on Twitter.
Why not do these calculations for the offense as a whole? Would help to determine when to start a defense.
Did you block the temperature plots for with/without precipitation? It would be interesting to see if cold is only worse because precipitation drives down temperatures, or if cold weather alone has an impact.