Writers of Pro Football Prospectus 2008

Most Recent FO Features

TrubiskyMit17.jpg

» Audibles: 2017 NFL Draft Day 1

The surprises started early in this year's NFL Draft. Your Audibles crew discusses some of the major winners and losers from the first round.

17 Apr 2017

NFL Head Coaches: Adjusted Games Lost

by Scott Kacsmar

Last week, we posted our annual study of adjusted games lost (AGL), which is Football Outsiders' metric for the impact of injuries on a team during the regular season. None of the five teams that accumulated at least 100 AGL made the playoffs in 2016.

We also made note of how Mike McCoy, who was fired after four seasons as San Diego's head coach, had AGL ranks of 28th, 31st, 26th, and 31st during his tenure. Any coach would find his job that much more difficult to deal with when there are that many injuries to starters and key reserves on a yearly basis.

This led us to wondering which coaches have benefitted from generally healthy seasons, and which coaches were dealt a tough hand like McCoy had in San Diego. That data is compiled below, and as it turns out, McCoy has had the worst injury luck of the 70 coaches since 2002 with at least three seasons of experience. McCoy's teams averaged the most AGL (104.8), the worst average AGL rank (29.0), and the worst average z-score (minus-1.3) in AGL. So it has been an exceptionally bad run of injuries for the Chargers since 2013.

This is not to say that AGL alone is a great indicator of a team's win-loss record. We have found the correlation between AGL and wins to only be in the 0.2 to 0.3 range. However, when we chart the last 15 years of AGL rankings vs. winning percentage, there is an obvious disadvantage to having a banged-up squad.

We only have 15 data points per ranking to go on, but there is a pretty clear trend line to that chart. For teams ranked one to 19 in AGL, only two groups (teams ranked sixth, and teams ranked 17th) have a combined record that is under .500. For teams ranked 20 to 32 in AGL, every group is below .500, and the teams finishing last (32nd) have the worst winning percentage (.310). In 2016, that bottom-ranked team was Chicago, which finished 3-13 after having the highest AGL (155.1) of any team in our database.

Twelve of the 15 teams to finish first in AGL (the healthiest teams each season) won at least eight games. Only the 2016 Rams (4-12) were worse than 6-10, and we detailed last week how they had the healthiest offense, but the worst performance, as coaching and talent still outweigh health on the importance scale. Only one of the 15 teams to finish 32nd in AGL (most injured) finished without a losing record, and the 2012 Packers (11-5) had an all-time great quarterback in his prime in Aaron Rodgers to combat that. When Rodgers hurt his collarbone and missed half a season in 2013, the Packers barely made the playoffs at 8-7-1. This is similar to how the Colts typically won 10-plus games despite ranking in the bottom eight in AGL for years, but any injury to Peyton Manning or Andrew Luck and the team failed to make the playoffs in those seasons.

We are not looking to credit or blame any specific coach for his team's injury luck, though things such as practice habits and the application of sports science can certainly have an impact on these numbers. We have seen Chip Kelly post great AGL numbers in Philadelphia, but that was not the case in San Francisco this past year (ranked 25th) while the Eagles (ranked fourth) still did well without him. Sometimes the hiring of a new strength and conditioning coach can also help, but football is a violent game and sometimes bones are just going to break in September.

Most coaches will tell us that injuries are not an excuse for a decline in performance. They are in fact much more of an explanation than an excuse, as a lost star player is practically impossible to replace with a run-of-the-mill backup. That unit is going to operate differently without its star. Even if some of the statistics look similar, the means of getting to those numbers will likely change when a Rob Gronkowski or J.J. Watt goes down, as they both did in the 2016 season. Was Houston still a good defense without Watt? Yes, the Texans ranked seventh in DVOA and were still able to hold their own against New England in the playoffs. Did the Patriots still score at a high rate and go undefeated in their last 10 games without Gronk? Obviously, but this meant other players had to step up in that star's absence.

A great coach will be able to overcome injuries and prepare the "next man up," as we often hear in discussions of the NFL. We just needed to add some data to see which coaches have been able to do this.

As for any questions about that data, we are only looking at the regular season for the eight-division era (2002-2016). In the 26 cases of an interim coach taking over during the season, we credited the Week 1 head coach for that entire season. For example, Wade Phillips gets credit for the 2010 Cowboys even though he was fired after a 1-7 start. Jason Garrett, who took over for the fired Phillips and has coached Dallas ever since, only gets credit for the 2011-16 Cowboys. Garrett is one of seven interim coaches to become the full-time coach since 2002, and he has easily been the most successful so far. The other interim hires are Mike Singletary (2008-09 49ers), Tom Cable (2008-09 Raiders), Leslie Frazier (2010-11 Vikings), Romeo Crennel (2011-12 Chiefs), Mike Mularkey (2015-16 Titans), and Doug Marrone (2016-17 Jaguars).

AGL by Head Coach (2002-2016)

The following table ranks all coaches with at least three seasons by AGL from healthiest to most injured. Since the league average AGL has gone up so much over this period of time with changes to injury reporting, it makes more sense to focus on the z-score ranking, which is the column the coaches are ranked by. This way the coach is being compared to how many standard deviations he is above or below the league average AGL for each season. But we did also include where his average rank is in total AGL and his average rank per season. For anyone who was replaced by an interim coach, their record listed below includes all of the games from that season. Active coaches are in bold.

Rk Head Coach Avg. AGL Z-Score Seasons Avg. AGL HC Rk Avg. Rk HC Rk Record Pct. HC Rk
1 Bill Parcells 1.2 4 19.1 1 6.3 1 34-30 0.531 26
2 Dom Capers 1.0 4 21.7 2 7.8 3 18-46 0.281 65
3 Todd Haley 0.9 3 34.6 9 9.0 4 21-27 0.438 48
4 Leslie Frazier 0.9 3 43.8 22 7.7 2 18-29-1 0.385 57
5 Tony Sparano 0.7 4 39.1 15 10.0 8 31-33 0.484 38
6 Dick Vermeil 0.7 4 26.9 3 9.5 7 38-26 0.594 12
7 Mike Munchak 0.6 3 48.7 28 9.3 6 22-26 0.458 44
8 Chip Kelly 0.6 4 57.6 46 9.3 5 29-35 0.453 45
9 Jim Haslett 0.6 4 28.9 4 12.3 14 28-36 0.438 49
10 Wade Phillips 0.6 4 37.9 13 11.0 11 39-25 0.609 11
11 Jeff Fisher 0.5 14 40.5 18 11.4 12 108-115-1 0.484 39
12 Bill Cowher 0.5 5 29.6 5 10.8 9 50-29-1 0.631 7
13 Mike Mularkey 0.5 4 45.8 26 11.5 13 25-39 0.391 55
14 Joe Gibbs 0.5 4 34.1 8 11.0 10 30-34 0.469 43
15 Jim Mora Jr. 0.5 4 38.1 14 13.0 17 31-33 0.484 40
16 Pete Carroll 0.4 7 54.8 41 13.3 20 70-41-1 0.629 8
17 Rex Ryan 0.4 8 54.8 42 12.8 16 61-67 0.477 42
18 Mike Tice 0.4 4 31.8 7 13.8 22 32-32 0.500 37
Rk Head Coach Avg. AGL Z-Score Seasons Avg. AGL HC Rk Avg. Rk HC Rk Record Pct. HC Rk
19 Brad Childress 0.4 5 40.6 19 14.4 25 42-38 0.525 27
20 Gary Kubiak 0.4 10 49.3 29 13.0 18 82-78 0.513 31
21 Joe Philbin 0.4 4 60.1 51 14.0 24 29-35 0.453 46
22 John Harbaugh 0.4 9 54.2 38 13.3 21 85-59 0.590 13
23 Jim Harbaugh 0.3 4 59.4 50 14.8 27 44-19-1 0.695 3
24 Mike Sherman 0.3 4 31.4 6 14.5 26 36-28 0.563 17
25 Dennis Green 0.3 3 35.0 10 12.3 15 16-32 0.333 62
26 Mike Smith 0.3 7 55.4 43 14.0 23 66-46 0.589 14
27 Herm Edwards 0.2 7 37.3 12 13.1 19 44-68 0.393 54
28 Mike Tomlin 0.1 10 58.4 49 15.8 35 103-57 0.644 6
29 Sean Payton 0.1 10 56.0 45 15.6 31 94-66 0.588 15
30 Marty Schottenheimer 0.1 5 35.4 11 17.0 43 47-33 0.588 16
31 Bruce Arians 0.1 4 69.0 59 16.5 40 41-22-1 0.648 5
32 Marvin Lewis 0.1 14 51.7 35 15.8 33 118-103-3 0.533 24
33 Jack Del Rio 0.1 11 49.5 30 15.4 30 89-87 0.506 35
34 Bill O'Brien 0.1 3 72.1 61 16.3 37 27-21 0.563 18
35 Lovie Smith 0.1 11 51.2 33 16.5 39 89-87 0.506 36
36 John Fox 0.0 15 55.7 44 15.8 34 128-112 0.533 25
Rk Head Coach Avg. AGL Z-Score Seasons Avg. AGL HC Rk Avg. Rk HC Rk Record Pct. HC Rk
37 Mike Nolan 0.0 4 41.0 20 16.3 36 23-41 0.359 60
38 Raheem Morris 0.0 3 54.5 39 17.3 47 17-31 0.354 61
39 Ken Whisenhunt 0.0 8 58.3 48 16.9 42 50-78 0.391 56
40 Mike Shanahan 0.0 11 49.9 31 16.4 38 90-86 0.511 32
41 Eric Mangini 0.0 5 50.3 32 15.2 29 33-47 0.413 51
42 Dave Wannstedt 0.0 3 40.0 17 17.0 44 23-25 0.479 41
43 Jason Garrett 0.0 6 69.0 58 17.5 49 53-43 0.552 21
44 Norv Turner 0.0 8 51.2 34 17.1 45 65-63 0.508 34
45 Brian Billick 0.0 6 39.7 16 17.7 50 50-46 0.521 28
46 Ron Rivera -0.1 6 69.1 60 14.8 28 53-42-1 0.557 19
47 Gus Bradley -0.1 4 73.3 63 19.5 56 15-49 0.234 67
48 Andy Reid -0.1 15 54.8 40 17.5 48 146-93-1 0.610 10
49 Mike Zimmer -0.1 3 78.6 65 16.7 41 26-22 0.542 23
50 Mike McCarthy -0.1 11 62.7 52 17.2 46 114-61-1 0.651 4
51 Jim Schwartz -0.2 5 62.9 53 18.2 51 29-51 0.363 59
52 Rod Marinelli -0.2 3 53.4 37 15.7 32 10-38 0.208 69
53 Jon Gruden -0.2 7 45.1 23 19.3 54 57-55 0.509 33
54 Chan Gailey -0.2 3 64.1 55 18.3 52 16-32 0.333 63
Rk Head Coach Avg. AGL Z-Score Seasons Avg. AGL HC Rk Avg. Rk HC Rk Record Pct. HC Rk
55 Mike Holmgren -0.2 7 45.3 25 18.6 53 62-50 0.554 20
56 Mike Martz -0.3 4 42.2 21 19.8 58 33-31 0.516 30
57 Tony Dungy -0.3 7 47.4 27 19.3 55 85-27 0.759 2
58 Steve Mariucci -0.5 4 45.2 24 22.8 61 26-38 0.406 52
59 Bill Belichick -0.5 15 64.3 56 21.0 59 185-55 0.771 1
60 Dennis Allen -0.6 3 82.4 67 22.3 60 11-37 0.229 68
61 Jim Caldwell -0.7 6 78.4 64 23.7 64 53-43 0.552 22
62 Chuck Pagano -0.7 5 87.3 68 25.4 66 49-31 0.613 9
63 Butch Davis -0.8 3 51.9 36 25.7 67 18-30 0.375 58
64 Dick Jauron -0.8 6 57.9 47 19.7 57 38-58 0.396 53
65 Tom Coughlin -0.9 13 73.3 62 23.3 62 108-100 0.519 29
66 Scott Linehan -0.9 3 63.7 54 25.0 65 13-35 0.271 66
67 Romeo Crennel -1.0 5 66.7 57 23.4 63 26-54 0.325 64
68 Steve Spagnuolo -1.1 3 80.5 66 26.0 68 10-38 0.208 70
69 Jay Gruden -1.2 3 101.7 69 26.3 69 21-26-1 0.448 47
70 Mike McCoy -1.3 4 104.8 70 29.0 70 27-37 0.422 50

We talked about McCoy, but Jay Gruden has not had much injury luck either in his three years with Washington. He ranks next to last in z-score at minus-1.2. Gruden is an offensive-minded coach, which is good since he has had the most injured defense by far in our database. Gruden's average defensive AGL is 70.5. The next closest coach is Dennis Allen at 46.6 AGL in Oakland. McCoy has had the most injured offense (61.5 AGL), but defensive guru Mike Zimmer (55.8 AGL) is awfully close after having the most injured offense we have ever tracked with the 2016 Vikings. Again, these numbers are slanted towards the last few seasons since AGL in general has gone up, but they are still interesting to see for context. Gruden has done very good things with Kirk Cousins and the offense, and maybe would have won more games if he had healthier bodies on defense. The same can be said of Zimmer's Minnesota teams, which need more of an offensive boost after clearly improving on defense under his leadership.

Despite ranking 57th and 59th in z-score, Bill Belichick and Tony Dungy are the only coaches in this study to win more than 70 percent of their games. Of course, they also had Tom Brady and Peyton Manning at quarterback for most of those games. Belichick still looks very impressive here, and he did manage an 11-5 season in 2008 with Matt Cassel at quarterback after Brady tore his ACL in Week 1. The Patriots were 28th in AGL that season, but still had Belichick calling the shots. Since 2002, 15 teams won at least 11 games while ranking 28th or worse in AGL. Belichick coached six of those 15 teams.

Tom Coughlin, 65th in z-score, had the worst injury luck for any coach with a winning record, including two Super Bowl-winning seasons. Of course, Eli Manning has never missed a start since 2004, and the Giants really padded Coughlin's injury counts by leading the league in AGL from 2013 to 2015 when Coughlin went 19-29 (.396). The Giants finally had a healthy year under Ben McAdoo last year, and not coincidentally returned to the playoffs. Jim Caldwell, 61st in z-score, also has managed a winning record despite bad health luck, but he had two years of prime Peyton Manning and has gotten the healthy version of Matthew Stafford in Detroit. When Caldwell had a miserable quarterback situation with the 2011 Colts, he went 2-14 and was fired.

So the quarterback's health obviously has a huge role in winning games, but we did not want to get heavily into AGL by position for these coaches. That feels better suited for a team-level analysis rather than tying things to the head coach.

Only 11 of these coaches won at least 60 percent of their games, and Wade Phillips is the only one of those 11 to rank in the top 10 for healthiest teams. He was 34-22 (.607) in Dallas; the Cowboys went 5-3 with Garrett down the stretch in 2010 after Phillips was fired, and Phillips would still be above 60 percent if we included those eight games. The Cowboys were certainly healthier in 2003-2010 when Bill Parcells and Phillips were the coaches rather than in recent years with Garrett (ranked 43rd). Parcells had the healthiest rosters of any coach here, with a plus-1.2 z-score. The average AGL was only 38.9 from 2003-2006 compared to 71.6 for 2013-2016, which again is why we are looking at z-scores more than the raw AGL averages. Still, Parcells only averaged 19.1 AGL per season and had the lowest average rank at 6.3. Leslie Frazier (2011-13 Vikings) and Dom Capers (2002-05 Texans) were the only other coaches to rank in the top quarter of the league on average in AGL. It is hard to win with an expansion team like Capers tried to do, or to win when your offense relies so much on Adrian Peterson like Frazier's Vikings did.

Few of these coaches lasted more than five years on the job, but among the 28 coaches with at least six years of experience, Jeff Fisher had the healthiest teams on average, ranked 11th overall in z-score. Yet only four of those 28 coaches had a worse winning percentage than Fisher's .484. Fisher only made the playoffs four times in his final 14 seasons, and he can't blame his failures on his team's health.

Pete Carroll (17th in z-score) has the second-healthiest rosters among active coaches (Mike Mularkey is 13th), but that will just prompt jokes about what he is hiding from the injury reports in light of his admission about Richard Sherman's undisclosed MCL issue this past season. Seattle has been a top-five AGL team in three of the last five seasons.

Rather than finish with a data overload of offensive and defensive splits, I figured it made more sense to look at the cases where a coach spent multiple seasons with multiple teams. There were 14 coaches to do so for 15 cases, because John Fox has actually stuck around that long with three teams now. These cases are sorted from the most drastic injury increases to the biggest health improvements.

Head Coach AGL: Multiple Seasons with Multiple Teams (2002-2016)
Head Coach Team Seasons Years Avg. AGL Z-Score Avg. AGL Avg. AGL Rk Record Pct.
Eric Mangini NYJ 3 2006-2008 1.0 25.5 4.7 23-25 0.479
CLE 2 2009-2010 -1.6 87.6 31.0 10-22 0.313
DIFF -2.7 62.1 26.3 - -0.167
John Fox DEN 4 2011-2014 0.4 56.3 12.5 46-18 0.719
CHI 2 2015-2016 -1.9 123.9 30.0 9-23 0.281
DIFF -2.3 67.7 17.5 - -0.438
Rex Ryan NYJ 6 2009-2014 0.8 43.4 8.5 46-50 0.479
BUF 2 2015-2016 -0.6 89.1 25.5 15-17 0.469
DIFF -1.4 45.7 17.0 - -0.010
Lovie Smith CHI 9 2004-2012 0.2 44.5 15.1 81-63 0.563
TB 2 2014-2015 -0.4 81.1 22.5 8-24 0.250
DIFF -0.6 36.5 7.4 - -0.313
Mike Shanahan DEN 7 2002-2008 0.1 40.9 14.6 66-46 0.589
WAS 4 2010-2013 -0.2 65.5 19.5 24-40 0.375
DIFF -0.4 24.5 4.9 - -0.214
Ken Whisenhunt ARI 6 2007-2012 0.0 53.6 15.8 45-51 0.469
TEN 2 2014-2015 0.0 72.4 20.0 5-27 0.156
DIFF 0.0 18.9 4.2 - -0.313
Head Coach Team Seasons Years Avg. AGL Z-Score Avg. AGL Avg. AGL Rk Record Pct.
Andy Reid PHI 11 2002-2012 -0.1 48.1 18.1 103-72-1 0.588
KC 4 2013-2016 0.0 73.2 15.8 43-21 0.672
DIFF 0.1 25.2 -2.3 - 0.084
John Fox CAR 9 2002-2010 0.2 40.3 14.1 73-71 0.507
DEN 4 2011-2014 0.4 56.3 12.5 46-18 0.719
DIFF 0.2 15.9 -1.6 - 0.212
Gary Kubiak HOU 8 2006-2013 0.3 47.1 13.8 61-67 0.477
DEN 2 2015-2016 0.5 58.5 10.0 21-11 0.656
DIFF 0.2 11.4 -3.8 - 0.180
Jeff Fisher TEN 9 2002-2010 0.4 34.4 11.7 77-67 0.535
LARM 5 2012-2016 0.7 51.4 10.8 31-48-1 0.394
DIFF 0.3 17.0 -0.9 - -0.141
Dick Jauron CHI 2 2002-2003 -1.0 45.4 20.5 11-21 0.344
BUF 4 2006-2009 -0.7 64.1 19.3 27-37 0.422
DIFF 0.3 18.7 -1.3 - 0.078
Jack Del Rio JAC 9 2003-2011 0.0 46.9 16.1 70-74 0.486
OAK 2 2015-2016 0.4 61.3 12.0 19-13 0.594
DIFF 0.4 14.3 -4.1 - 0.108
Herm Edwards NYJ 4 2002-2005 -0.1 40.1 16.3 29-35 0.453
KC 3 2006-2008 0.7 33.6 9.0 15-33 0.313
DIFF 0.8 -6.5 -7.3 - -0.141
Norv Turner OAK 2 2004-2005 -1.1 59.8 28.0 9-23 0.281
SD 6 2007-2012 0.3 48.3 13.5 56-40 0.583
DIFF 1.4 -11.5 -14.5 - 0.302
Jim Caldwell IND 3 2009-2011 -1.4 85.6 30.0 26-22 0.542
DET 3 2014-2016 0.0 71.2 17.3 27-21 0.563
DIFF 1.4 -14.4 -12.7 - 0.021

The first six coaches listed saw injury rates shoot up at their second jobs while their winning percentages all dropped. Eric Mangini left some healthy New York rosters for a myriad of injuries and lack of talent in two short-lived years in Cleveland. John Fox and Rex Ryan have had loaded injury reports for the last two seasons, but at least Fox has still kept his job for 2017 while Ryan is out in Buffalo. Fox and Kubiak both saw a health increase and win boost by going to Denver, and both benefited from the Manning effect. Fisher and Herm Edwards were the only coaches to get better health in their second jobs, but saw their winning percentages drop by roughly 14 points. Norv Turner moved from Oakland into a much nicer situation with a loaded San Diego team that was coming off a 14-2 season in 2006. Caldwell eventually swapped the annual injury festival in Indianapolis for a more mediocre (in a good way) health record in Detroit.

All in all, it does not appear that these coaches had any voodoo magic or forces working against them that carried their injury luck from one city to another. Like with any person dealing with one of life's ailments, we just hope to avoid the serious ones and try to heal as best we can. A coach can't heal his players, but he can always get the backup ready to play as best he can.

Posted by: Scott Kacsmar on 17 Apr 2017

18 comments, Last at 23 Apr 2017, 2:22am by LionInAZ

Comments

1
by Theo :: Mon, 04/17/2017 - 4:46pm

"We also made note of how Mike McCoy, who was fired after four seasons as San Diego's head coach, had AGL ranks of 28th, 31st, 26th, and 31st during his tenure... This led us to wondering which coaches have benefitted from generally healthy seasons, and which coaches were dealt a tough hand like McCoy had in San Diego."

It seems as though you're excluding an influence of a head coach on the health on his players.
I surely think there's an amount of luck. But luck = chance * opportunity.
Some coaches must me doing things smarter than others.

4
by Alternator :: Tue, 04/18/2017 - 12:53am

There's also the consideration that some coaches are more willing to accept injury risks, whether because they need to swing for the fences to keep their jobs, or because they already have a solid team and an average mid-to-late round pick isn't much of an improvement. Older players also tend to accumulate more injuries, which is likewise not a negative to hold against a coach.

11
by TopherDoll :: Tue, 04/18/2017 - 2:43pm

This is an interesting thing to me to see someone say (though you've said it in a few articles already, though never responding to people who question you) because in many cases the team's trainers stick around much longer than the coach. In Denver for example their head trainer has been through multiple coaches and GM's after serving over two decades.

Injuries are largely luck, unless you know something that we don't about how NFL practices work that the rest of us don't. From everything I've heard, read and watched if you wanted to blame a person, rather than the luck it is, I'd look at the trainers or position coaches who run practice.

13
by Scott C :: Tue, 04/18/2017 - 6:50pm

LMGTFY:

http://bleacherreport.com/articles/2555624-preseason-acl-injures-can-be-...

Soft tissue injuries, especially non-contact ones, can be reduced as they are in large part due to over-work.

17
by Tomlin_Is_Infallible :: Wed, 04/19/2017 - 2:21pm

I wonder also how much of it is square-peg/round-holing. Well, we drafted this guy with terrible hips/knee/ankle flexion, and he's undersized. But damn it all, we're going to turn him into the next Patrick Willis.
--------------------------------------
The standard is the standard!

2
by RickD :: Mon, 04/17/2017 - 4:55pm

I see Parcells at the top and am reminded of that oft-seen video of him shouting "This is why you lift all them weights".

7
by Lebo :: Tue, 04/18/2017 - 9:27am

I also remember, when he joined the Dolphin, Parcells cutting players who got injured during camp (this didn't apply to star players, obviously). That probably reduced the proportion of his roster that was injury-prone.

10
by Independent George :: Tue, 04/18/2017 - 11:05am

You're cut too, shushy.

3
by RobotBoy :: Mon, 04/17/2017 - 7:42pm

Really interesting way to approach the elusive issue of how head coaches affect team performance.

5
by jtr :: Tue, 04/18/2017 - 9:20am

That graph hits a big pet peeve of mine. It's absolutely silly to connect the dots like that. There is no reason whatsoever to believe that curve has any value. Do we really think it's of any predictive value that win percentage climbs between 25th and 28th in AGL? I think it's pretty obvious that it's just random noise.
This should be a scatter plot with a best-fit curve, probably just a simple least-square linear regression since we don't have any particular reason to assume there are higher-order effects at play. Throwing in a spline to connect each point is visually distracting at best and misleading at worst.

8
by Independent George :: Tue, 04/18/2017 - 11:04am

I would expect a sigmoid curve, with the biggest impact at right around the median AGL scores. Intuitively, it makes sense that steadily increasing injuries would have larger and larger impact until you reached the point of completely screwed (PCS), and it just doesn't much matter anymore - especially since you're bound by a maximum 16 losses.

12
by RickD :: Tue, 04/18/2017 - 5:35pm

These things are all true.

14
by Scott C :: Tue, 04/18/2017 - 6:53pm

based on the looks, I'd say an order-3 polynomial fit would look decent -- it looks like there are two high and low domains, with a transition between them.

This seems plausible, in that a small number of inuries are fairly easy to overcome, and then there is a breaking point and eventual state where an extra couple injuries won't really make it much worse.

16
by Tomlin_Is_Infallible :: Wed, 04/19/2017 - 2:15pm

"order-3 polynomial fit " why should that cross the x-axis 3 times or have a redundant root?

--------------------------------------
The standard is the standard!

18
by LionInAZ :: Sun, 04/23/2017 - 2:22am

The graph is simply the result of the author using a simple-minded graphing program. I don't think it was meant to be misleading. I agree that a simple linear regression would have been more suitable, given that the standard deviation in win pct is probably pretty large at each AGL rank. Show the std devs, or just plot all the points!

6
by jtr :: Tue, 04/18/2017 - 9:26am

>When Rodgers hurt his collarbone and missed half a season in 2013, the Packers barely made the playoffs at 8-7-1. This is similar to how the Colts typically won 10-plus games despite ranking in the bottom eight in AGL for years, but any injury to Peyton Manning or Andrew Luck and the team failed to make the playoffs in those seasons.

Are there currently no position adjustments for AGL? I think the easiest way to come up with rough positional values would be based on average salary at the position. For instance, per OTC, the 16th highest-paid QB (roughly median) is on an $18M/yr contract, while the median fullback is on a contract under $1M/yr. That could be interpreted as the NFL signalling that a QB is roughly 18 times as valuable as a fullback, which doesn't sound unreasonable to me.

15
by Scott C :: Tue, 04/18/2017 - 6:58pm

Yeah, this same data would be very interesting to see if adjusted for position.

Unfortunately, it is not the same thing to lose a top end QB for the season as it is losing a replacement level QB.

Nor losing a pro-bowl left tackle vs losing a glorified turnstyle.

That said, there is probably a way to adjust the values that would lead to better correlation with win/loss that incorporates position and player quailty. Its just not going to be easy.

9
by Independent George :: Tue, 04/18/2017 - 11:05am

I'd be very curious to see a plot of AGL vs Win% for individual coaches rather than just the averages in that final table.