The Factors: Do you want to play a snow game?

On paper, the Colts-Bills game last week looked pretty unappealing. The Colts entered with just three wins and seemed to be treading water just waiting for next season and the hopeful return of Andrew Luck. The Bills were still in the postseason mix, but their flip-flop from Tyrod Taylor to Nathan Peterman and back seemed to take the wind out of the sails of their out-of-nowhere run to playoff contention. At least, that was the case until Mother Nature intervened.

Snow games are great because snow so obviously affects the way football is played. It’s difficult for quarterbacks to throw the ball, and a lot of coaches don’t even bother trying. When backs try to run, they frequently slip and fall, and so do the defenders who try to cut to them to make the tackle. In the Bills game, Adam Vinatieri made an extra point — the most routine play in a typical NFL game — and I practically lost my mind.

Oh my goodness, @adamvinatieri. Got it! ❄️❄️❄️ #Colts pic.twitter.com/mQAhS5jjBe

— NFL (@NFL) December 10, 2017

But for as different as a snow game looks from a regular game, I was worried I would have a difficult time researching their impact because they so rarely happen. I could only think of a couple of snow games from recent memory, and whenever one happens, all I see are video clips from the 1967 Ice Bowl and the 2001 Tuck Rule game.

I still wanted to try, and so I started by looking up the ISD station weather readings I pulled during the hours of the Bills game. There are only 6 ISD stations near New Era Field in Buffalo, it being a non-major market, but the closest one (11.9 miles) at Buffalo Niagara International Airport showed some telltale signs of likely snow. First, the temperate stayed below 32 degrees Fahrenheit. Second, there were several recordings of accumulated precipitation.

Precipitation readings are much more hit-and-miss than temperature readings, so I opted to try to find other snow games using the maximum precipitation reading of any nearby station during the hours of each game. That created some false positives, but I watched video for every game that met my criteria and was surprised to find that snow games were much more common than I remembered. Since 2009, I found 17 snow games, and in 10 of them, the snow was pretty heavy.

That still isn’t a ton of games to work with, but the weighted average approach that I have typically taken with this type of research is built to elegantly handle small samples. I ran sets of comparisons that looked at both a team’s willingness to pass and run in snow and in normal weather as well as their pass and run production in snow and in normal weather. My first thought was to split light and heavy snow games from each other, since I believe they would have noticeably different impacts. But I decided to also combine them into a test of all snow games in case the small sample sizes made the data too fine to split.

Just based on what I had seen in the handful of snow games I’ve watched, I was confident that passing attempts would be down and rushing attempts would be up when there was snow. That proved to be true. In both light and heavy snow, teams threw between 5 and 6 fewer passes per game.

Interestingly, the rush attempt increase was higher in light snow (4.4 per game) than it was in heavy snow (1.9 per game). The sample sizes could contribute to that disparity, and there is evidence of at least some of that in the fact that the pass ratios decline a more similar 7.0 and 5.3 percent in light and heavy snow. I also expect all offense is hindered more by heavy snow, which could contribute to an overall decrease in offensive plays per game.

Even with their counterintuitive differences, it’s clear that both light and heavy snow causes teams to run the ball more than they normally would. What is interesting is that performance is completely uneven depending on the intensity of snow fall. When snow is light, quarterbacks have completed the ball a bit more often and thrown for a bit more yards per attempt than they normally would. When the snow becomes heavy, however, completion percentage and yards per attempt take major hits.

I actually don’t think that’s a small-sample quirk because passing efficiency is related to pass percentage. An extreme metaphor would be if a team ran the ball on 99 percent of their plays, even in good weather, then that 1 time in 100 when they threw the ball, chances are good their receivers would be wide open and the play would go for a lot of yards. No doubt defenses choose to play more heavily for the run in snow because they know runs are more likely to happen. That difference could create a subtle uptick in passing efficiency even if light snow failed to help or even hurt the defense-independent difficulty of passing the ball.

In heavier snow, the elements start to take their toll. Heavy snow makes the ball wet, makes receivers and defenders more difficult to see, and makes it more difficult for receivers to run precise pass routes. All of those factors seem to conspire to hurt the quarterback. And it turns out they hurt backs a bit, too, although nowhere near to the same extent.

In heavy snow, backs lose a very small 0.035 yards per carry, although, from a fantasy perspective, they more than offset that with increases in their total carries and their touchdown rate. In light snow, things are even better for them. They tend to carry the ball more often, run for more yards per attempt, and score more touchdowns per attempt.

Sadly, there is only a three percent chance of precipitation in Buffalo this weekend. At his recent rate of production, opposing back Kenyan Drake might run for 200 yards if he played in a blizzard. Fortunately, this research shows that snow is a bit more common than you might have expected, and with three northern, outdoor teams — the Eagles, Patriots, and Steelers — in the driver’s seats for home-field advantage, we may well have a white football Christmas this year.