Aggressiveness Index 2017
by Aaron Schatz
Usually we wait until later in the offseason to bring out the Aggressiveness Index numbers, but we wanted to publish them a little earlier this season since they matter for this Sunday's Super Bowl matchup.
Football Outsiders introduced the concept of Aggressiveness Index back in Pro Football Prospectus 2006. The goal was to find a way to rank coaches based on their tendencies on fourth downs in a manner that was easy to understand but accounted for the different rates at which the average coach will choose to "go for it" in different situations. Although no NFL coach is as aggressive as the data suggests he should be, we discovered there is quite a wide range of fourth-down tendencies among coaches.
Aggressiveness Index numbers center around 1.0 and generally describe how much more (or less) likely each coach is to go for it on fourth down compared to his peers; for example, a coach with 1.20 AI is roughly 20 percent more likely to go for it than an average coach in equivalent situations. The Aggressiveness Index excludes obvious catch-up situations: third quarter, trailing by 15 or more points; fourth quarter, trailing by nine or more points; and in the last five minutes of the game, trailing by any amount. It also excludes the last 10 seconds of the first half, and it adjusts for when a play doesn't actually record as fourth-and-short because of one of those bogus delay of game penalties that moves the punter back five yards. Only regular season is included.
You may have read elsewhere that Doug Pederson went for it on fourth-and-1 more often than any other head coach in the league this season. There are actually two reasons for that. One reason is that Pederson was, in fact, really aggressive on fourth-and-1. He went for it on 61 percent of qualifying fourth-and-1s, more than any head coach other than Sean Payton. But the second reason is that no team got into fourth-and-1 as often as the Philadelphia Eagles. The Eagles ended up with 23 different fourth-and-1 opportunities that qualified for Aggressiveness Index. No other team had more than 17. Tampa Bay only had two.
Pederson doesn't actually end up No. 1 in Aggressiveness Index, which helps demonstrate how a couple of very rare fourth-down opportunities can heavily influence these numbers when we're looking at the small sample of just one season. Adam Gase is No. 1 instead, specifically because he decided to just throw caution into the wind with David Fales at quarterback in Week 17. Gase had Fales throwing on fourth-and-7 and fourth-and-3, both in the first quarter, and then on fourth-and-8 in the third quarter (losing by 13 points, which counts for AI). Without Week 14, Gase would have an AI of 1.14, which is still above average but not No. 1. Gase's aggressiveness is a bit of a shock since he was the most conservative coach in the league in his first season, going for it on just a single qualifying fourth down in 2016.
Jack Del Rio is also ahead of Pederson. He was aggressive on fourth-and-short, though not quite as aggressive as Pederson. However, he passes Pederson in AI because the Raiders had Marquette King run a fake punt on fourth-and-11 from his own 30.
Here are the full Aggressiveness Index results. Each situation is listed with the number of times the coach went for it and the number of opportunities, the rate, and then the expected number of times an average coach would go for it given the down, distance, score, and location on the field.
|2017 Aggressiveness Index|
Some of these results further show how one season is just a small sample size of fourth downs. Gase is not the only head coach to make a big change from 2016 to 2017. John Harbaugh, usually among the most aggressive head coaches, comes out in the bottom half. Mike Zimmer drops from ninth to last. Jim Caldwell had also been very aggressive the previous couple of seasons.
With that in mind, we may look at including a three-year rolling average of AI in Football Outsiders Almanac 2018 instead of just the single-season number. We'll have more on Aggressiveness Index later in the offseason, including updated career numbers that will add in 2017 as well as 1986-1988.
21 comments, Last at 06 Feb 2018, 1:35pm
#4 by Pat // Feb 01, 2018 - 11:43am
It drives me nuts that Pederson goes by win probability so much, because sometimes it just doesn't work. Going for it on 4th and 8 near midfield close to the half, up 7, where your opponent gets the ball first in the second half? There's no way that's the right call. I mean, just as the obvious criticism, there's no way you know how often you're going to get the first down because *no one goes for it on 4th and 8 unless they have to*.
Of course, the Giants make a stupid decision too, and it ends up being fine. Grr.
#5 by Aaron Brooks G… // Feb 01, 2018 - 12:15pm
No one goes for it on 4th-2 from your own 28, with a lead, either.
Seriously. Since 2000, that play is the only time anyone has gone for it on 4th-2 on their own end of the field in the last 3 minutes of a game in which they were tied or had the lead. The three other attempts were either botched punts (2) or an intentional safety (1).
Only one other team (0.2% of opportunities) has gone for it in 4th-longer-than-1 in a similar situation, and they didn't do it from scrimmage. The Rams converted (the sole conversion) via a Hekker pass against the Seahawks in 2014. There are more botched punts, kneeling it outs, and intentional safeties than bona fide attempts.
Even 4th-1 is super rare, with three attempts in 62 opportunities. One was by the Patriots (it also didn't work).
Only 10% go for it in any situation.
Belichick made a naked call here; there was no precedent for going for it.
#14 by Anon Ymous // Feb 02, 2018 - 2:38pm
Given how poorly the defense was playing, it was undoubtedly the right call. Where NE screwed up was in passing the down beforehand and wasting a TO that could have been used to challenge the spot (or to mount a response afterward).
#19 by PatsFan // Feb 03, 2018 - 10:20am
Going for it was absolutely the right call. The NE defense was in even worse shape than the ATL defense was at the end of SB52. IND was winning that game if NE punted the ball there.
The problem is the whole sequence of plays. While BB claimed afterwards he was planning to go for it on 4th-and-short all along, the playcalls belied that. They clearly didn't decide to go for it on 4th down until it was 4th down and were not playing true 4-down football that series. He played to maximize his chance of getting a first down on or before 3rd down, not to maximize his chance of getting a first down at all.
#20 by doktarr // Feb 06, 2018 - 1:26pm
I find it somewhat heartening that Pats fans can see this was the right call. I still remember having a big argument with a Pats fan co-worker the next morning where I filled his white board with conditional probabilities.
The only way you can make it a bad call is if you argue that the Colts were dramatically less likely to score after a punt than after taking over on downs. Like, half as likely. And given the way the game had been going in that quarter, that really seems like a crazy argument to make.
#11 by Pat // Feb 02, 2018 - 1:34pm
Yes, I get the idea that it sounds hilarious, but I meant sometimes a win probability analysis doesn't reach the right conclusion.
For one thing, you're either neglecting information, or you're looking at too sparse a data set to draw conclusions. Whether or not the neglected information is significant is a question.
#15 by justanothersteve // Feb 02, 2018 - 3:38pm
I don't think he thought it was hilarious. The fact that it sometimes doesn't work it is what makes so many coaches timid. Here's an example.
You roll a multi-sided die in a game. In this scenario, there is a 80% chance something good happens and a 20% chance something bad happens. Most people would take that risk. However, there are some incredibly risk-averse people who would not risk because 20% is too high a risk.
Some NFL coaches are like those risk-averse people. Some have figured it out - going for it on fourth and two from your opponents 40 instead of punting, and in other situations where it's worth the risk. You can't cherry-pick the times it didn't work it and claim victory. Just because it didn't work out doesn't mean it wasn't the right thing to do in that situation.
#17 by nat // Feb 02, 2018 - 4:51pm
Okay, thanks for not taking offense. I was teasing a bit.
Yes, it's possible to do a win probability analysis wrong. If, for example, you use WP from league averages and don't consider the teams involved in the current game, you're wrong. If you aren't adjusting that based on what you've seen so far in the game, you're wrong. (This part is what a lot of stats-heads conveniently forget.)
As for the sparse data set problem, that applies to either decision you could make. You don't have the option to say "This decision is too hard. Let's pretend it's fourth and twenty instead." And no matter what you think the odds are at fourth and eight at the forty-five, you're saying "Punting (or going for it) gives me the best chance of winning, based on the limited information I have." Maybe not out loud. But with your actions, which speak still louder.
I think the data will always be sparse, too. The best a coach can do is a color-coded "go for it" chart, with colors ranging from "always punt" to "always go for it", with way stations at things like "punt unless the defense is showing weakness" - with entries for combinations of distance, margin, field position, time remaining, etc. This could be a real chart, or something a coach has in his head: a gut feeling, if you will.
#10 by jtr // Feb 01, 2018 - 9:41pm
>there's no way you know how often you're going to get the first down because *no one goes for it on 4th and 8 unless they have to*.
This is a really easy one. You just look at third and 8 success rate. That's it. There's your probability for how often a team gets 8 yards when they have to.
And yeah, like nat said above. The fundamental nature of win probability is that it dictates which call makes it more likely for you to win the game.
#16 by jtr // Feb 02, 2018 - 4:16pm
Defenders pick off passes that they shouldn't on 4th down pretty regularly. I seriously doubt any defense in the league treats 4th down any differently than 3rd down. In both cases, your goal is to get a stop short of the sticks on this specific play so you get the ball back.
#21 by doktarr // Feb 06, 2018 - 1:35pm
The thing is, it *did* work, because the Giants didn't score anyway. The possibility that your defense will hold is part of the reason you make that decision. Of course, in that case it was a goal line stand that allowed the defense to hold, but that's part of the realm of probability you are considering.
Everybody remembers the surprise onside kick in Super Bowl XLIV, but fewer people remember that the Saints got stuffed on a 4th and goal from the 1 with 2 minutes left in the half. They ended up getting the ball back and kicking the FG as time expired in the half, so they got the 3 points anyway. That's not just some random detail - it's *part of the reason you go for it in the first place*.
#6 by Badfinger // Feb 01, 2018 - 1:03pm
There isn't really a way to account for philosophy on this sort of thing is there? My understanding of the Eagles-style aggressiveness is you definitely know by 3rd down if you will go for it on 4th down, and you may even know at the start of a series, which changes playcalling philosophy. I'm certain they're not alone in that school of analytics, but the caveat that Pederson isn't as aggressive as he appears because the Eagles have more 4th and 1 opportunities reads to me more like they were comfortable putting themselves into a situation where it might be 4th and 1, rather than the fortuitous circumstance where they were willing to demonstrate they'd go for it if they got there.