Using plotly express, in just one line of code we can visualize all these variables together:None / Light Rain / Light Snow / Rain Storm / SnowThere’s a lot going on here: each dot or bubble represents one QB game played.
The temperature is represented by color, the number of fantasy points scored corresponds with the height on the y-axis, and the wind speed is represented on the x-axis.
Additionally, the size of the scatterpoint represents the maximum wind gust for that game, which is why the larger bubbles mostly appear on the right side of each graph in games where the avg wind speed is higher.
Finally, each subplot represents the precipitation category for that game.
The two major takeaways from this graph are the negative correlation between fantasy points and wind speed/wind gust, across all precipitation categories, as well as the general downward trend and lower overall fantasy points scored in games with precipitation.
The plot above shows the same data (minus temperature), but has all the games grouped together, and here the color of the bubbles represent the precipitation category.
Now that we have seen some graphical representation of the overall effects of wind, precipitation categories, and temperature, I really wanted to dive in to analyze the difference between QB statistics in games played in great weather vs bad weather.
For great weather, I wanted to look at just the games played in pretty perfect conditions, that should not suffer any adverse effects due to weather.
I took a subset of the overall data where the temperature was moderate (between 60 and 80 degrees), clear skies (or in a Dome) with no precipitation, and an average wind speed of under 5 mph.
For bad weather, I took a subset with rain/storm or snow (light snow and light rain were excluded), avg wind speed above 12 mph, or extreme polar temperatures (below 10 degrees or above 95 degrees).
The resulting subsets each comprised a little over 2000 QB games played, a decent sample size.
Lets look at some charts that show some overall comparisons between QB in the bad weather games vs the great weather games:Negative % difference for all categories indicates worse QB stats in bad vs great weather, as expectedSame trend here- most of the data lies below 0, indicating worse performance in bad weatherThe main takeaways from the overall bad vs great weather datasets are that the production metrics, although more volatile, generally have a larger percentage decrease in bad weather.
Efficiency metrics, while also generally worse, aren’t as pronounced.
This is likely due to coaches calling fewer pass attempts in bad weather games.
Perhaps in the future I will take a look at running back data and look for a corresponding increase in volume/rush attempts in bad weather.
Finally, I wanted to look at a few individual players and compare how they perform in great vs bad weather games.
I chose 3 players from teams who play in home fields which frequently have bad weather, Tom Brady of the New England Patriots, Ben Roethlisberger of the Pittsburgh Steelers, and Aaron Rodgers from the Green Bay Packers.
To represent teams whose home game are played in a Dome/good weather, I chose Peyton Manning from the Indianapolis Colts, and Kurt Warner from St Louis Rams / Arizona Cardinals.
Unsurprisingly, all 5 QBs averaged more fantasy points in great weather vs bad.
The bad weather acclimated QBs each score between 2 and 3 more fantasy points in great weather games.
Also unsurprising is the drastic negative downtick for Kurt Warner.
While the modest difference for Peyton Manning was a bit of a surprise at first glance, Manning was well known for his preparation and a meticulous work ethic, and it shows in the consistency of his performance, regardless of any challenges from adverse weather.
Lastly, as a lifelong die hard New Orleans Saints fan, I would be remiss if I didn’t take a peak at the stats for Drew Brees.
It is fairly widely claimed in the fantasy football world that Drew Brees has some of the more drastic home-road splits of anyone in the league, meaning that his performance drops off significantly when he leaves the friendly ‘Dome-field advantage’ of the Mercedes Benz Superdome.
Is this characterization accurate?.Lets see if the numbers back up that reputation:Say it ain’t so, Drew!Alas, it pains me to admit the pundits are right.
In summary, here are the major takeaways I gleaned from exploring the NFL weather data.
The biggest factors that affect a quarterbacks performance are the wind and precipitation, especially when they occur together, (as they usually do).
The effect of wind/wind gusts is the most significant weather variable affecting QB performance.
The effects are fairly slight for avg wind speeds from 0 mph up to about 15 mph, but above that threshold, the difference in fantasy points is about 12% less.
In games with winds over 20 mph, the negative effects are more significant, with ~17% fewer fantasy points per game.
Precipitation, on its own, is basically inconclusive.
Although the average fantasy points are lowest for games with light rain, the data for games with rain/storm and snow are not statistically significant.
Temperature, as an isolated weather variable, doesn’t have much of an affect on QB performance.
QBs across the board tend to score 8–10% more fantasy points in great weather than bad.
However, QBs who play their home games in a dome or great weather cities don’t always perform worse in bad weather than QBs who are more used to playing in bad weather.
Thanks for reading!.For a deeper dive into the data and analysis behind this blog post, here is the link to my github repo.
.