Staley Leads NFL in Re-Calibrated Aggressiveness Index

Brandon Staley, Los Angeles Chargers
Brandon Staley, Los Angeles Chargers
Photo: USA Today Sports Images

NFL Offseason - The debate over fourth downs is over, and the analytics have won.

Oh, it may not quite seem that way when you watch commentary on TV. There are still old school football people attacking coaches for being too aggressive, for "chasing points" or alternately "leaving points on the board," for not "playing the percentages" (when they are in fact doing just that). You'll still hear stuff like this on sports radio, attacking "the freaking nerds."

It doesn't matter. Head coaches still might not be quite as aggressive as the analytics would suggest, but they are much more aggressive on fourth downs compared to just four seasons ago. The analytics won the argument, and things have changed.

And because we've won the argument, this year Football Outsiders has to change our Aggressiveness Index metric which measures how often coaches go for it on fourth down compared to the league average. What constitutes "league average" has changed so dramatically that the old numbers don't quite make sense now.

Football Outsiders introduced the concept of Aggressiveness Index way 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. There are other methodologies now for measuring fourth-down aggressiveness, mostly based on win probability analysis: for example, the EdjSports head coach rankings or results spit out by Ben Baldwin's fourth-down simulator. Each methodology will have small differences in how it ranks the coaches, but Aggressiveness Index differs from the others by measuring coaches not against what they should do but against the actual decisions made by coaches themselves.

Aggressiveness Index numbers were designed to 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 problem we're now having is that when we compare current coaches to historical averages, AI numbers no longer center around 1.0. The leaguewide Aggressiveness Index in 2021 was up to 1.90. In 2020, only two head coaches came in with an AI below 1.0: Vic Fangio and Brian Flores. This season, not a single head coach in the league came in below 1.0. Using the original AI baselines, Joe Judge was the lowest head coach in the league at 1.03.

NFL (Old) Aggressiveness Index, 2014-2021
Year Old
AI
Overall
Att.
Expected
Att.
Overall
Opps.
Rate
2014 0.98 197 201 3217 6.1%
2015 1.00 211 212 3265 6.5%
2016 1.06 233 220 3193 7.3%
2017 1.03 215 209 3259 6.6%
2018 1.37 272 199 3049 8.9%
2019 1.46 307 210 3086 9.9%
2020 1.60 386 242 2884 13.4%
2021 1.90 450 237 3136 14.3%

The Aggressiveness Index, by the way, 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 the regular season is included.

Obviously, there's going to be a problem with a metric that is designed to average around 1.0 if every single coach is coming out above 1.0. The average decision-making on fourth downs has changed dramatically since 2018. Our baselines have to change also.

Before we introduce the new numbers with the new baselines, let's take a step back and look at how much things have changed since 2018. These charts will include all fourth downs, including those in obvious catch-up situations.

First, raw go-for-it rates show how things clearly changed in 2018. The league go-for-it rate, including in late-game catch-up situations, hovered around 12% for three decades. Then it started going up four years ago. It's now above 20%.

AI graphic

Go-for-it rates have changed at pretty much every distance, but the difference is clearest in the shortest yardage: fourth-and-1 and fourth-and-2.

AI graphic

Where are coaches going for it more often? Not all over the field. There are two places where coaches are going for it more often. First, they go for it more often within 10 yards of the goal line. Second, they go for it more often at midfield, roughly between the 35-yard lines. The rest of the territory shows go-for-it rates not too different from those we saw historically from 1983 through 2017.

AI graphic

Put these charts together, and you can see where these three situations are driving the change in head coach aggression. In short yardage, near the goal line, or between the 35s, attempts on fourth down are way up. The rest of the time, they're up very slightly.

AI graphic

And so, that's where we've changed our baselines. Coaches are now expected to go for it more often on fourth-and-short, more often near the goal line, and more often near midfield. The average Aggressiveness Index for the period 2018-2021 is back to 1.0.

Here are the full Aggressiveness Index results for 2021, based on the new AI baselines representing 2018-2021. The identity of the most aggressive head coach in the league will not surprise you.

Aggressiveness Index, 2021
Rank Coach Team AI Go Opp Rate Exp
1 Staley LAC 2.08 21 92 22.8% 10.1
2 Kingsbury ARI 1.74 18 90 20.0% 10.3
3 Stefanski CLE 1.59 16 91 17.6% 10.1
4 Campbell DET 1.48 23 99 23.2% 15.6
5 LaFleur GB 1.46 18 98 18.4% 12.3
6 Sirianni PHI 1.38 16 97 16.5% 11.6
7 McCarthy DAL 1.37 18 99 18.2% 13.2
8 Nagy CHI 1.35 20 95 21.1% 14.8
9 Rivera WAS 1.34 16 97 16.5% 12.0
10 Payton NO 1.26 15 99 15.2% 11.9
-- Gruden LV-1 1.24 5 35 14.3% 4.0
11 Reich IND 1.21 23 109 21.1% 19.0
12 Meyer JAX-1 1.13 10 67 14.9% 8.9
13 McDermott BUF 1.09 16 93 17.2% 14.7
14 Reid KC 1.04 12 77 15.6% 11.5
15 Fangio DEN 1.03 14 100 14.0% 13.6
16 Harbaugh BAL 1.00 14 106 13.2% 14.0
17 Zimmer MIN 1.00 12 116 10.3% 12.0
18 Vrabel TEN 0.99 12 102 11.8% 12.1
19 Taylor CIN 0.99 16 102 15.7% 16.2
20 Saleh NYJ 0.98 9 87 10.3% 9.2
21 Shanahan SF 0.93 15 94 16.0% 16.2
22 Arians TB 0.93 11 95 11.6% 11.9
23 Tomlin PIT 0.91 12 108 11.1% 13.2
24 Rhule CAR 0.91 11 106 10.4% 12.1
-- Bevell JAX-2 0.89 4 23 17.4% 4.5
25 Smith ATL 0.89 12 90 13.3% 13.6
26 Bisaccia LV-2 0.84 8 71 11.3% 9.5
27 Flores MIA 0.84 10 108 9.3% 12.0
28 McVay LAR 0.83 13 91 14.3% 15.7
29 Belichick NE 0.76 8 88 9.1% 10.5
30 Culley HOU 0.69 7 103 6.8% 10.2
31 Judge NYG 0.66 10 100 10.0% 15.2
32 Carroll SEA 0.62 5 108 4.6% 8.0

With the old baselines, Brandon Staley had an AI of 3.21. This would not be a record; John Harbaugh was at an astonishing 3.95 in 2019. But still, that's pretty high. With the new baselines, Staley leads the league at 2.08. He's the only coach to go for it more than twice as often as the expectation, but that's the new, higher set of expectations based solely on the last four seasons.

Some additional notes on this year's Aggressiveness Index:

  • Frank Reich ranking only 11th is a bit of a surprise because he went for it so often on fourth down, but he went for it so often because he was in fourth-and-short so often. The Colts had the most qualifying fourth-and-1 opportunities, going for it on 16 out of 20.
  • We're still not at the point where any coach goes for it on every qualifying fourth-and-1, but we got close. Nick Sirianni (6 of 7), Matt LaFleur (7 of 8), and the combined Jacksonville coaches Urban Meyer and Darrell Bevell (6 of 7) all went for it on every qualifying fourth-and-1 but one.
  • Pete Carroll was at the bottom of the league overall and specifically on fourth-and-1, going for it on just three of 10 qualifying fourth-and-1 opportunities.
  • Staley and Bruce Arians led the league with 12 qualifying fourth-and-2s. Staley went on seven on those, Arians on only two.

Here's a look at the top 10 coaches since 2018 in the new Aggressiveness Index:

Highest Aggressiveness Index, 2018-21
Rank Coach Team AI Go Opp Rate Exp
1 Staley LAC 2.08 21 92 22.8% 10.1
2 Campbell DET 1.48 23 99 23.2% 15.6
3 Stefanski CLE 1.47 30 171 17.5% 20.4
4 Kingsbury ARI 1.41 52 286 18.2% 37.0
5 Sirianni PHI 1.38 16 97 16.5% 11.6
6 Pederson PHI 1.32 47 306 15.4% 35.6
7 Harbaugh BAL 1.29 63 385 16.4% 48.9
8 LaFleur GB 1.29 46 278 16.5% 35.8
9 McCarthy GB/DAL 1.25 39 265 14.7% 31.2
10 Williams CLE 1.20 6 38 15.8% 5.0

And a look at the bottom 10 coaches since 2018:

Lowest Aggressiveness Index, 2018-21
Rank Coach Team AI Go Opp Rate Exp
55 Philbin GB 0.00 0 19 0.0% 0.6
54 Jay Gruden WAS 0.45 4 125 3.2% 8.8
53 Joseph DEN 0.52 5 103 4.9% 9.6
52 Wilks ARI 0.52 3 90 3.3% 5.7
51 Patricia DET 0.54 15 257 5.8% 27.8
50 Carroll SEA 0.61 26 396 6.6% 42.9
49 Fangio DEN 0.64 22 289 7.6% 34.6
48 Lynn LAC 0.65 21 259 8.1% 32.3
47 Judge NYG 0.66 20 195 10.3% 30.2
46 Gase MIA/NYJ 0.68 21 291 7.2% 30.7

Note that we'll still track both versions of Aggressiveness Index so that we can compare current head coaches to those of the past. But when we're just comparing current head coaches to each other, we'll use the new Aggressiveness Index so that we don't live in some Lake Wobegon NFL where every head coach is above average.

Thanks to Jim Armstrong, who came up with the idea of Aggressiveness Index and computes it for every season.

Comments

26 comments, Last at 04 Mar 2022, 12:59pm

1 Carroll's not a surprise …

Carroll's not a surprise (lot of defensive coaches on the low-offense-aggression list). What is remarkable is how low the Belichick Tree is on this measure. Is Vrabel the highest? And he's only tree-affiliated. But Judge, Patricia, Flores, even Belichick are low.

2 Not a surprise

Any Pats fan (myself included) who has watched a lot of their games over the last few years knows that Belichick has gotten very conservative the last few years. Maybe it's due to personnel changes, or because the team has become more defense-oriented over the last few years, but he's definitely not nearly as aggressive as he was when he had great offensive teams (or maybe it's better said as when he's had weak defensive teams).

12 I actually looked up…

In reply to by crw78

I actually looked up Belichick's history during the season due to a conversation I was having, and according to the old AI calculation, Belichick actually didn't change as much as you think. There's some year to year variation, but he was never crazy-aggressive outside of 2010 (and that team played incredibly quickly; they'd go for it on fourth downs because opponents wouldn't be ready). It was more that the rest of the league got more aggressive, and he didn't increase his aggressiveness over time. Illustrative example: in 2019, the last Brady season, his AI was 1.39, which was good for 17th. A decade earlier, his AI was 1.33, and that was 3rd.

3 There is a clear meanshift…

There is a clear meanshift in the graph that occurred in 2017 that has kicked off an ever increasing trend?

Anyone want to haphazard a guess as to what changed in 2017(or 2016)? I might have assumed the infamous 4th and 2 call would be a seminal moment in the aggressiveness index, but it appears that was more muted.

 

7 But that happened literally…

But that happened literally in the last game of 2017.

 

The mean shift occured in 2017( would have to look at the yearly splits to see when it happened).

Something either happened in 2016 or early in 2017 to warp the entire league.

8 2018

If you look at the table of original AI per year, you can see the shift happened in 2018. It was after the Eagles won the Super Bowl.

4 I know Carroll was a great…

I know Carroll was a great coach once upon a time, but observing him the last few years, I just get this sense that the game might be passing him by. To see him on a list with a bunch of other coaches who were fired within a few years makes me wonder at what point do fans/media start calling for Carroll to be pushed out in Seattle. 

16 Rare NFL Stat

I find the aggressiveness index interesting because it doesn't assign value to its rankings. This is the rare NFL stat that doesn't assign 'goodness'. It's just a comparison between other coaches.

Maybe it would be best to be in the middle? That would indicate consensus among coaches.

But is the consensus flawed?

But there are myriad reasons why a coach may deviate from the consensus. And I love it. This list gives all of us a baseline to speak from. 

19 25% is still not that much…

25% is still not that much. I was kind of surprised he didn't leave this year. He's a good coach in many respects, but his philosophy seems so stale by this point it reeks. A fresh start somewhere else would work wonders for his career. And now what about Russell Wilson? He's got to be thinking the same thing for himself now that Carroll is staying. Hope he gets his wish, he's wasting away there.

21 I get what you are saying…

I get what you are saying but I did find the wording funny when discussing a 70 year old man. You can always have a fresh start and you can always improve your career those terms tend to be applied to younger people. I would think when he leaves Seattle he leaves football completely and fully retires. Age is much less of a factor in modern times, especially if you have the kind of money NFL head coaches have, but still there have only been 6 coaches in the history of the NFL to head coach a game at 70 or older. That will jump to 8 if Arians and Belichick coach next season.

For the curious who don't want to filter though google results
Romeo Crenel was 73 when he named interim in Houston in 2020 and broke the record of the legend
George Halas coached till he was 72
Marv Levy also coached till he was 72 (he was like 6 months younger than Halas for his last game)
Al Saunders coached till he was 71
Gunther Cunningham made it to 70
Pete Carrol as made it to 70 (a few months younger currently than Cunningham was for his last game)

 

24 Your post made me wonder…

Your post made me wonder about my own career. At first blush, software engineering ( or data science) feels like a young man's profession. The technology constantly changes and it requires a constant need to keep up with the new technologies. Some of them have pretty steep learning curves. And it absolutely is a field that can cannibalize a lot of your free time. At certain points, I would look around at my team and say no wonder no one here has kids because how would they find the time?

Obviously the NFL coaching profession takes all of that and amplifies it 10 x. And yet, seeing older men doing it and being successful either suggests age confers benefits I am not able to perceive yet or they are just rare humans.

 

23 Thinking of the Stafford trade

Made by a team that lost in the divisional round but might need a "slight" bump in play with a now 33 year old QB...I think the Titans should think heavily of upgrading from Tannehill with Russ with a similar trade.

And Russ already knows how to throw to a 2019 2nd round Ole Miss WR. 

That's of course if Seattle is foolish enough to do such a thing.

20 Casually looking at it

The top 7 coaches this year kept their jobs. 11 of the bottom 27 are out (10 if you exclude Payton). 

The top 9 from '18-'21 are HCs currently. The bottom 5 are not. A total of 9 of the bottom 10 are not.

Lafleur and Mccarthy being right next to each other is funny too.