It truly is beginning to look a lot like Christmas. Decorations are in every store (ok, so that’s not new), you can’t escape hearing songs about Santa, MIT students are all furiously cramming for their finals, and NFL games are taking place in weather like this:
It’s obvious that this kind of weather affects every aspect of the game (except LeSean McCoy, apparently). But to us, the most interesting stat line from Sunday’s Detroit-Philadelphia game to us isn’t McCoy’s 217 rushing yards or Detroit’s 7 fumbles. It’s the fact that the football never made it through the uprights. There were no field goal attempts, and 7 two-point conversions after 8 touchdowns. The only kick attempt of any kind was an extra point that the Eagles blocked.
This game is certainly an abnormal and extreme example of weather’s impact of the kicking game. But what about games that aren’t being played in the middle of a blizzard? How does the weather affect the field goal attempts in those games? This was the focus of a paper that we wrote for the 2013 Sloan Sports Analytics Conference. You can watch Aaron give a bow-tied presentation on our full methodology and results on the SSAC website. But here are the highlights, followed by some new statistics on the 2012 season.
Modeling the Likelihood of Field Goal Success
In our paper, we used logistic regression to look at the factors that make a field goal attempt more difficult, and then used these findings to quantify the difficulty of every field goal attempt in the 2000-2011 NFL seasons. The largest effect was, of course, the distance of the attempt. But beyond that, we confirmed a lot of the conventional wisdom about environmental factors. Kicking at altitude in Denver is easier, and kicking on grass, during precipitation, in cold temperatures, or on windy days is harder. We also looked at situational factors – the regular season vs. postseason, home vs. away, icing the kicker, and situational pressure (when the game was on the line). And this is where the conventional wisdom isn’t so wise after all. We found that none of these situational factors had a significant effect. Now, we’re not saying that no kicker has ever let a high-pressure situation, a rowdy away crowd, or a timeout get into his head. It’s just that on average, across all kickers, the potential effect of these factors was so small that it couldn’t be differentiated from random chance.
Our logistic regression allows us to build a model that predicts the likelihood of success from 0–100% of any given attempt. This is a representation of that attempt’s difficulty. All you do is plug in the current conditions, and it tells you what the probability of success is. We unfortunately don’t have a robot to calculate this for us, but we’re working on an app that will help you with this process. Interested NFL front offices should feel free to contact us sooner. (Or they can just read our paper, as the equation’s written in there.)
We used our equation to calculate the difficulty of every NFL field goal attempt from the 2000-2011 seasons. (There are 11,896 of them, in case you were wondering.) Armed with this information, kickers can be evaluated based on the difficulty of attempts that they make and miss. We don’t need to consider all field goals to be created equal, like the traditional statistic of make percentage does. To assess kickers, we created a metric called “added points.” This metric subtracts our model’s predicted likelihood of success from the actual outcome of the attempt (1 for a make, and 0 for a miss), and multiples this difference by the 3 points awarded for a make.
Prater’s Longest Field Goal isn’t the Most Difficult Field Goal
As an example, look at two field goal attempts from Sunday’s games. The Broncos’ Matt Prater attempted a 64-yard field goal at home at the end of the first half. According to the model, the kick had a 31.2% likelihood of success. Prater nailed it, setting an NFL record and earning himself (1–0.312)*3 = 2.06 added points. On the other end of the spectrum, Jay Feely missed a 25-yard field goal indoors in Phoenix. This kick had a 95.3% chance of success, which gives Feely (0–0.953)*3 = -2.86 added points. As these examples demonstrate, kickers are penalized more for easy misses and rewarded more for difficult makes.
While Prater’s kick sets the NFL record for distance, it is not the most difficult field goal ever made. In fact, it is only the 9th most difficult successfully converted between 2000-2012. The single most difficult field goal made in our database is Rob Bironas’ 60-yard game-winning field goal in the wind against the Indianapolis Colts in week 13 of 2006. This attempt had a likelihood of success of only 22.4%.
2012 Kicker Rankings
In our paper we ranked kickers based on their added points / attempt. However, those rankings were from the 2000-2011 seasons only. Since SSAC 2013, we have applied the logistic regression to 2012 NFL data obtained from Armchair Analysis.com. The results are presented in this section.
The table below shows the top and bottom 5 kickers that had at least 20 attempts in 2012:
|5||Nick Novak||San Diego||0.364||90.0|
|26||Garrett Hartley||New Orleans||-0.075||81.8|
|27||Nick Folk||New York Jets||-0.135||77.8|
|29||Mason Crosby||Green Bay||-0.265||65.7|
|30||David Akers||San Francisco||-0.340||70.2.|
*Indicates Pro Bowl selection
It is interesting to note that of the five worst kickers – whose negative added points/attempt indicate that they are actually taking away points from their team, compared to an average kicker in the same situation – four are still with their current team. The only exception is David Akers, who was release by the 49ers in the offseason and picked up by the Detroit Lions. And whom did the 49ers sign to replace Akers? None other than Phil Dawson.
You can also see that the two kickers selected for the 2013 Pro Bowl – Phil Dawson and Blair Walsh – were ranked in the top 5. However, this hasn’t been the case every year. Looking back through our year-by-year rankings from 2000-2011, there was only one year when the top-ranked AFC and NFC kickers made the Pro Bowl. That was 2004, when Adam Viniateri and David Akers were selected. On average, the kicker selected to the Pro Bowl is ranked #3-4 in his conference. The worst year? 2008, featuring Stephen Gostkowski (#5 in the AFC, 0.15 added points/attempt behind #1 Jeff Reed) and John Carney (#11 in the NFC, 0.57 added points/attempt behind #1 Jason Hanson).
Another comparison we make in our paper is the rating differential between kickers’ rankings based on make percentage and added points/attempt. Since make percentage is difficulty-ignorant, it underrates kickers who attempt proportionately more difficult kicks, and overrates kickers who make proportionately easier kicks. Here’s are our top and bottom 5 over- and underrated kickers for 2012:
|Rank||Kicker||Team||Degree Underrated /
Overrated by Make Percentage
|1||Greg Zuerlein||St. Louis||15|
|3||Connor Barth||Tampa Bay||6|
|27||Stephen Gostkowski||New England||-5|
|29||Lawrence Tynes||New York Giants||-9|
Greg Zuerlein and Lawrence Tynes’ 2012 performances truly show the problems with only evaluating kickers based on their make percentages. Zuerlein’s make percentage was only 74.2%, which was bad enough to rank him 28/30. Tynes’ make percentage was more respectable, 84.6% (16/30). However, when you consider added points/attempt, their rankings essentially flip-flop. Zuerlein gained 0.181 added points/attempt (13/30) and Tynes earned -0.009 added points/attempt (25/30).
If you want to take a look at our year-by-year data, feel free to check it out here. There’s no guarantee that these analyses will finally help you to pick a consistent fantasy kicker, or that they’ll indicate whether your team should keep its current kicker or let him go. But, it will give a quantitative justification for the nervousness you feel when your kicker trots out onto the field to make a critical kick and the weather looks like this:
Aaron Johnson and Alexander Stimpson are Ph.D. students in the Department of Aeronautics and Astronautics at MIT. Torin Clark, Ph.D. is a postdoctoral research fellow at the Massachusetts Eye and Ear Infirmary.