2009 Adjusted Games Lost Totals
by Bill Barnwell
In the NFL, success can be as simple as staying healthy. There have been a million articles written about extra hours of practice and film study yielding surprise playoff appearances, but often those same teams do the same things a year later and lose three or four games more than they did in the previous campaign. Injuries are the NFL's version of Occam's Razor.
While 2009 might have thrown our preseason prediction system for a loop, on the whole, the injury status quo remained the same: The teams that saw a dramatic rise or fall in their health had a commensurate change in their record.
To compile injury data, we use a metric called Adjusted Games Lost (AGL), introduced in Pro Football Prospectus 2008. AGL uses the injury report, the league's various long-term injury lists, and historical rates of participation to measure the level of impact injuries have on a team's starters in a given season. A player listed as Probable, for example, produces a fraction of the AGL that a player listed as Out does, because the historical rate of participation for starters with such injury designations is dramatically different.
The result is a figure representing an estimate of how many "games" each team's starters missed in a given season. AGL is more accurate than simply counting "starter games lost" because, by counting appearances on the injury report, it also accounts for fact that playing hurt means playing at less than full strength. NFL teams recorded an average of 53.9 AGL this season, roughly the equivalent of losing three starters for the entire season in training camp.
These figures begin with the starting lineups as projected at the start of training camps. So a player like Ma'ake Kemoeatu would be included as a starter, even though he never actually started a game this season. We also include a number of "important situational players" for each team, like slot receivers from teams that often go three-wide or situational pass rushers. (Examples include Steve Breaston, Felix Jones, or Bertrand Berry.) Finally, we also adjust the numbers to account for the new starters replacing injured players, factoring in the impact of going from a second-string player to a third-string player. Kickers, punters, and long-snappers are not included.
(For those of you looking for HGL, the metric we introduced in the New York Times this September, you'll have to wait till Football Outsiders Almanac 2010, where we'll be delving even further into our injury research.)
Who was 2009's most injured team, according to AGL?
The Buffalo Bills were the league's most injured team by a wide margin in 2009, beating out the perennially-injured Colts for the top spot. 20 different Buffalo players we considered to be starters, if healthy, missed games during the season. That includes five defensive backs (Donte Whitner, Drayton Florence, Jairus Byrd, Terrence McGee, and Leodis McKelvin) and five offensive linemen (Brad Butler, Demetrius Bell, Eric Wood, Jonathan Scott, and the late-arriving Kendall Simmons).
Speaking of the Colts, it's pretty clear that the Polian Era consistently results in injury rates higher than the league average, likely owing to the team's strategy of fielding an undersized, quicker defense. Fortunately for Indy fans, the Polian Era also has given them a team that drafts better than virtually anyone else in football, so there's not the dropoff that often comes for other teams with injuries. Since 2000, the only time the Colts have had an AGL figure better than league-average is 2005, when they went 14-2 and had the best point differential of the Polian Era.
It's also extremely unusual for the two teams competing in the Super Bowl to rank among the league's least-healthy teams. The Saints aren't perceived as a team that's had to battle through many injuries this year, but they've lost a variety of starters for huge chunks of the season, including fullback Heath Evans (10 games), left tackle Jammal Brown (16), defensive tackles Kendrick Clancy (13 games) and Sedrick Ellis (six), and virtually their entire secondary; Darren Sharper, Jabari Greer, Randall Gay, and Tracy Porter combined to miss 17 starts this season.
On the other side of the coin sit the Denver Broncos, who beat our preseason expectations with some close wins and a remarkable string of health. Broncos starters combined to miss 15 games this season; the two Super Bowl teams each lost more starts than that from just one guy (Jammal Brown, Anthony Gonzalez). Ryan Harris was responsible for eight of those 15 missed starts by himself, thanks to a gruesome-sounding dislocated toe. It's a remarkable total, and it's one not likely to recur, which is why our 2010 projection for Denver is unlikely to be very hopeful.
Among the six healthiest teams were organizations that have made a habit of showing up at the bottom of these lists: Tennessee, Kansas City, and Dallas, whose injury history we detailed in Pro Football Prospectus 2008 and Football Outsiders Almanac 2009. While we've established that injury totals for a team tend to regress to the mean, it's becoming apparent that those organizations might be quantifiably better at keeping their players healthy than the average team. We'll be exploring that issue further in FOA 2010.
The impact of injuries on a team, though, really stands out when we compare the injuries a team had in 2009 to how healthy they were in the previous season. This table does just that, with a positive figure indicating that the team suffered more injuries in 2009 than they did in 2008.
Buffalo was about league-average in 2008, so it's not surprising to see them at the top of this list. While Philadelphia and San Diego were able to improve on their records despite their health issues, Carolina (#2), Miami (#4), Washington (#6), the New York Giants (#7), and Chicago (#9) all fell off and/or didn't meet expectations. In the cases of Carolina and especially Miami, this turnaround was built into their projections. It was also built into our pessimistic projection for the Jets, who managed to overcome an injury to defensive tackle Kris Jenkins and still make it into the playoffs.
The bottom of this list shows the impact staying healthy can have. Denver's second-half woes cost them a playoff spot, but the five next-healthiest teams all made it, including the surprising Bengals. Cincinnati suffered through the most AGL in the history of the metric last year, and while they were still the ninth-most injured team in the league this year, their dramatic improvement drove a significant increase in wins.
Other injury curiosities:
Washington had the most-injured offense in football, with 55.2 AGL on offense alone. They were 22nd on defense. For essentially the second half of the season, they were playing down four offensive starters every week: halfback Clinton Portis (and would-be starter Ladell Betts), tight end Chris Cooley, and offensive linemen Chris Samuels and Randy Thomas.
Jacksonville was the league's healthiest offense. Their only starter to miss time before Week 17 was tackle Maurice Williams, who missed four games. (Tra Thomas, included as a situational player, did miss Week 1.)
Indianapolis topped the defensive injury charts for the second time in three years, with five defensive backs, three linebackers, and both defensive ends missing at least two games.
Denver had the league's healthiest defense. Just behind them was Dallas, whose front seven and starting cornerbacks didn't miss a single start. Safeties Gerald Sensabaugh and Ken Hamlin combined to miss five games.
Please also note that these figures are preliminary, and may differ some from what you see in FOA 2010.
51 comments, Last at 01 Feb 2010, 9:13pm
#1 by Yuri (not verified) // Jan 29, 2010 - 11:53am
I am mildly curious about what happens when a starter is lost to injury and then his replacement is also lost--do the losses count twice in the games both missed?
CAPTCHA: vastness filled
#4 by ammek // Jan 29, 2010 - 12:03pm
I believe so, yes, to take account of the loss in expected value from starter to reserve to third-string. For example: Peyton to Sorgi to Painter. (Or, less dramatically, Anderson to Quinn to Ken Dorsey, in 2008.)
#6 by Neoplatonist B… (not verified) // Jan 29, 2010 - 12:13pm
Or Jamal Williams - Ryon Bingham - Travis Johnson - The guy from the rental car company - A bunch of rookies/KC rejects and guys who were cut in the preseason.
#2 by ammek // Jan 29, 2010 - 11:58am
I agree with you Bill, I think there isn't enough time spent looking at total injuries in relation to performance. Injuries to particular individuals tend to get a lot of press, especially when those individuals are highly paid and have their own shampoo commercials, for example.
This was mentioned last year, but my main concern about AGL is to do with horses and carts. It needs looking into, but anecdotally I'd say a lot of the injuries on bad teams happen with the season well in progress, ie after the team has already 'become' bad.
I realize that AGL is an approximate statistic, and I think it's very interesting. I just wonder whether it makes more sense to imply (as your article does slyly) that 'teams are bad because they are injured', rather than 'teams are injured because they are bad'.
#12 by Still Alive (not verified) // Jan 29, 2010 - 12:50pm
Except you know, a lot of the really bad teams were relatively healthy...
#20 by Chip // Jan 29, 2010 - 3:34pm
Umm, except you know, there is only one really bad team (KC) in the ten 10 most healthy list.
#22 by thendcomes (not verified) // Jan 29, 2010 - 4:17pm
There's two. Just saying.
#25 by thendcomes (not verified) // Jan 29, 2010 - 4:21pm
Nope, sorry, just one. Looked at the wrong chart.
#28 by ammek // Jan 29, 2010 - 4:52pm
And of the nine teams who finished 6-10 or worse, we have numbers 1, 3, 5, 6, 7 and 8 in AGL. I'd say this year's numbers strongly support the link between games lost to injury and poor performance. It remains to be determined in what direction it works. Probably both.
A first strategy could be to measure AGL on a month-by-month basis, over several seasons. Perhaps look at extreme peaks of new injuries. When did the 2009 Browns become abnormally injured? In September? Or November? How does that compare with other oft-injured, 5-11 teams? Is there a correlation between the moment at which injuries are suffered en masse, and performance (DVOA, wins, etc)? Is there a delay? Once FO has enough seasons of AGL, it ought to find patterns that will throw light on whether (and which) injuries hurt teams the most. It's a very interesting area for research — a valuable statistic.
#32 by jedmarshall // Jan 29, 2010 - 6:21pm
Intuitively it sort of makes sense bad teams and injuries go hand in hand. Sometimes injuries cause the team to lose. Sometimes the team is already losing and sits players out more often than someone with something to play for. If your star QB or other young promising player is a little beat up, you are more likely to sit him the last few games and make sure he comes back completely healthy next year if you are 4-9 rather than 7-6 and fighting for a playoff spot.
#40 by ammek // Jan 30, 2010 - 4:58am
And I think players may get hurt more when they're playing from behind, just as they fumble more — the desperate folly of trying to make an extra yard when you need one.
#46 by Jerrytron // Jan 30, 2010 - 3:57pm
How about a simple correlation between AGL and win %? The same way FO got its start questioning the link between "establishing the run" and winning.
The correlation between AGL and win % is -.1876. Which is weaker than the ~.23 correlation I believe was found between rushing yards and win %.
#3 by Big Woody (not verified) // Jan 29, 2010 - 12:02pm
Does the way different teams seem to use the injury classifications (eg. probable vs questionable) affect AGL?
#34 by HostileGospel // Jan 29, 2010 - 9:38pm
Er, I suck at reading, disregard.
Overall, I'd be kind of embarrassed to critique something when I didn't know what the hell I was talking about, but then, oh yeah, my NAME is on what I write, isn't it?
#5 by Erithtotl (not verified) // Jan 29, 2010 - 12:08pm
I think this article shows a lot of insight into why the Team Efficiency Ratings seem to rate the Colts low. The Colts have always been very conservative about injuries and this year even more so. In addition to havingt lost several key players to near season long injuries, they also sat started in many games that perhaps a more desperate team, one that didn't get off to such a great start, might not have.
The fact that the Colts pulled out many games where they were at less than full strength is a testament to Manning and the other key stars. But it also means that a lot of those games the Colts played with one hand tied behind their back, explaining their closer results against weak teams. In a year when many of the perenial AFC powers had off years, the Colts could afford to take the risk of dropping one of these gimmie games.
The end result is that the Colts team going into the playoffs is likely better than any team they fielded in any game over the course of the regular season, and it's whon in their first two playoff victoies.
#35 by Bobman // Jan 30, 2010 - 3:10am
While that works in theory and one week ago I'd have seconded that line of reasoning, the current Indy injury list looks rather lengthy. Now part of it might be Polian's temper tantrum aimed at keeping his boys home from the Sham Bowl this weekend (if they're injured they don't have to travel just to wave to the crowd and then head home). The other two parts are probably actual injuries and gamesmanship before the big game.
So they might come out of the tunnel as healthy as you say. But even if 21 starters are as healthy as they have been all season, but Freeney is not, that's a bad place to be.
Well, they've lived on the edge all year; why should their last game be any different.
#7 by DGL // Jan 29, 2010 - 12:14pm
So it would appear the obvious question would be, have you looked at the correlation between AGL and how much your preseason projection was off by?
#8 by Joe T. // Jan 29, 2010 - 12:29pm
Interesting data set.
If you were to use it to argue that certain teams are better at player conditioning, I'd think you'd want to exlude at least training camp and potentially all pre-season injuries.
#9 by Anonymous2 (not verified) // Jan 29, 2010 - 12:34pm
A point on the Colts: IMO they are extremely conservative when it comes to injuries. The 'Bill Polian era' of injuries is probably related more to that philosophy than it is to smaller players. These guys are blowing out knees and twisting ankles and pulling hamstrings in practice, not getting run over on gameday.
#36 by Bobman // Jan 30, 2010 - 3:23am
I was going to ask what a 270 lb DT has to do with Gonzalez blowing out his knee without being touched? Edge blowing out his knee in 02? Rhodes the same in 02? Stokley always being hurt except 04? Even Freeney's Lis Franc break would be hard to pin on his measurables--he planted, spun, and broke. If he was bigger and heavier, it might have been worse! The CBs who were hurt this year are the bigger ones--Jackson (IR) and Hayden (8 games?) And while Bob Sanders is short, he's not exactly 160 lbs dripping wet. Marvin Harrison is 6-0 and about 185 and until he blew out his knee in 07, he had gone about 8 seasons with hardly any time at all missed.
In fact, I'd like to see their D injuries plotted by weight against other injuries at the same positions. They had a couple tiny DTs last year who managed to play all or most of the year. That's the one spot where they have been relatively healthy this year and--shazam! if you exclude the final two rollover games, their Run D improved a good amount from last year despite facing Chris Johnson (x2), MJD (x2), Steven Jackson, Ray Rice (x2), and Frank Gore.
Where havig smaller D players might result in more injuries is the "basketball" type injuries, because they're fast guys they sprint, stop, cut, shift, sprint again etc and that type of use might lead to more knee and ankle injuries than lumbering guys who are pretty much straight-line bulldozers. Just a thought.
#10 by nat // Jan 29, 2010 - 12:40pm
Has any attempt been made to calculate an AGL for each team's opponents?
#11 by Sid // Jan 29, 2010 - 12:43pm
It's really difficult to make an accurate compilation due to the differences in how the injuries are reported.
Remember when the Patriots would list half their roster as questionable?
#13 by DZ (not verified) // Jan 29, 2010 - 1:05pm
Good point. Couldn't this say more about how the Colts approach injury reporting? Granted, they've lost a lot of players (Gonzalez, Vinatieri, Jackson, Sanders, Hagler), but still it makes you wonder if there isn't a methodological flaw in the way this report is compiled that exaggerates the Colts totals..
#14 by DZ (not verified) // Jan 29, 2010 - 1:06pm
I'd be interested to see if the Colts averages for guys who play when marked as probable, questionable and doubtful sync up with the averages for the rest of the league.
#18 by Tim (not verified) // Jan 29, 2010 - 2:03pm
This was my thought as well. When in doubt, indy seems to put guys on the injury list, whereas some teams won't reveal as much about borderline cases.
I doesn't help that indy listed a large percentage of their roster as injured for those forfeit games they had at the end of the season. Hopefully this was taken into account and adjusted for as well.
And does missing Bob Sanders count as losing a starter? I mean, come on. Who really expects him to start more than half of the games in any one season? ;)
#17 by jedmarshall // Jan 29, 2010 - 2:00pm
Seems pretty accurate to me. If I understand the methodology correctly it doesn't count Vinatieri in the report because he is a kicker.
They lost their #2 WR, #2 CB and All-Pro SS for the entire year in addition to nagging injuries to their LT, #2 RB and a lot of the defense. They luckily (and strategically) had the depth necessary to cover for the 3 big injuries with very little dropoff in performance.
#15 by Temo // Jan 29, 2010 - 1:26pm
Just quick reaction to the article:
Indy's defense is fairly small, as we all know, while Dallas and Denver had similar types of defenses (fairly big guys, all around, and a 3-4). It's probably just coincidence, though. Just thought I'd spit that out there for now.
#16 by Temo // Jan 29, 2010 - 1:28pm
Just realized you'd already mentioned as much in the article.
#19 by 4tuna (not verified) // Jan 29, 2010 - 3:09pm
Interesting tidbit: Schwartz moves from Tenn to Det and the Lions have a 28% reduction in injuries, yet still finish 5th. Last year must of beeen brutal for them.
#26 by DrewTS (not verified) // Jan 29, 2010 - 4:29pm
0-16? Yeah, I'd call that brutal.
#37 by Bobman // Jan 30, 2010 - 3:26am
Yeah, about half of their injuries were guys stepping into moving traffic and sticking their heads in (electric) ovens. Or maybe that was just the fans.
#41 by jimmy oz (not verified) // Jan 30, 2010 - 8:15am
Cincy ranked fairly high on the list, but also improved by a fair margin.
#21 by Jim A. (not verified) // Jan 29, 2010 - 3:51pm
Dallas had one of the most talented and healthy teams in the league this year and they still managed to flame out in the playoffs?
#50 by Temo // Jan 31, 2010 - 3:34pm
Healthy? Yes. The Vikings were more talented overall, however. Better pass rush, better QB, better WRs, better D-Line.
#23 by Hank (not verified) // Jan 29, 2010 - 4:18pm
I feel that the consistently better teams, have better depth. So that if a starter goes down, the replacement drop off isn't as terrible. Some teams field players that have no business in the league, others have potential stars locked behind entrenched players.
And certainly different positions are easier/more difficult for a new guy to suddenly take a starting roll.
The Oline comes to mind as a big beneficiary of time spent together as a healthy unit.
Kickers don't seem to matter as much (though punters and longsnappers can kill a team)
#24 by Stan (not verified) // Jan 29, 2010 - 4:20pm
I don't think it's a coincidence that a team like Dallas stayed healthy. Parcells' theory has always been that bigger teams stay healthy and peak at the end of the year. That is why he always drafts big people first with Dallas and Miami.
But when a "smaller" team has Freeney, Mathis, Brackett, Session, and Bethea as their "undersized players", it doesn't matter as much bc those guys are such good players. As well as when you have Manning on offense. All of that covers up a lot of injuries and everything else.
But I really do think that is the untold story of why the Colts didn't try to go 16-0. It has more to do with the fact that they couldn't afford any more injured players on defense or in their receiving corps having already lost Gonzalez, than any chance that Manning would get hurt.
#27 by DrewTS (not verified) // Jan 29, 2010 - 4:32pm
"But I really do think that is the untold story of why the Colts didn't try to go 16-0."
I get your point, but it's hardly the "untold" story. That was the explicitly stated reason they didn't go for 19-0.
#38 by Bobman // Jan 30, 2010 - 3:32am
I think maybe Stan means the obvious subtext that a lot of irate fans, many in the media, and the "they disrespected us" opponents overlooked. It's a subtle difference.
If Polian had come out and said "look, we're the most injured team in the league, waah waah" he'd have been abused even more. By saying the same thing in a more forward-looking, optimistic way (we're doing what we feel is best to succeed in the playoffs), he kind of hid the fact they they were very injured to anyone who didn't care to do some digging. He put the focus on the future and not the past. So the people who complained about the move had an easy out to not look at the injury history, only the record.
#29 by roguerouge // Jan 29, 2010 - 4:55pm
I believe that they did a NYT article on injury reporting classifications and rate of actual games played.
#30 by Danish Denver-Fan // Jan 29, 2010 - 5:04pm
#31 by Stan (not verified) // Jan 29, 2010 - 5:14pm
Drew, just saying that everyone thinks that it's about Manning, but to me it's more about the other starters on defense and WRs, who have a much better chance of getting hurt than Manning does. Once you take all of those players out, there is no sense in keeping Manning in so that he has to put up 40 points to beat the Jets or Bills.
#39 by Bobman // Jan 30, 2010 - 3:36am
Ah well, Stan, I misinterpreted you above. But you're answer #31 is good, too. ;-)
With CJ being hurt at LT quite a bit near the end, it makes even more sense to keep Manning out of harm's way. If he has to go for 40 each game, that means a lot of drop-backs, and more chances to visit a hospital.
#33 by speedegg // Jan 29, 2010 - 7:22pm
I betcha FOA '10 could have a whole CHAPTER on this topic. Depends what superstar is injured, what position is injured, for how long, what GM's are better drafting for depth, what coaches are better adapting to injuries, when is the tipping point to crash (or rise), how does this affect their DVOA ranking, and factor that into next year's projections.
I bet Colts and Cowboys would be some of the better teams. Colts for replacing injured starters (except Manning) and Cowboys for keeping their players healthy.
Can't wait to read it in July......
#42 by loneweasel (not verified) // Jan 30, 2010 - 1:13pm
More concrete proof of Josh McDaniels' smoke and mirror run.
That franchise is going to crash hard.
#43 by NBOYER (not verified) // Jan 30, 2010 - 1:33pm
I think what these stats don't reflect, but you should really try to consider, is 'presenteeism' or the opposite of absenteeism. In other words, determining which teams started a high number of 'probables', because in all likelihood, these were the teams whose injured players started despite having to play at much less than 100% due to injury.
As a Denver Broncos fan, I think this metric would be particularly important in understanding the Broncos performance in these areas. While they appear to have had one of the healthiest teams all season, and it certainly showed in the first half of the season w/ the 6-0 start, the Broncos faded down the stretch. Many of Denver's Key Players down the stretch like Kyle Orton, Brian Dawkins, C. Buckhalter, Ryan Clady, Kris Kuper (and even K. Moreno to a degree), were all starting games late in the season while battling serious injuries (rightly or wrongly, this also happens to be why McDaniels took issue with Brandon Marshall refusing to play due to injury in the last week of the season).
#44 by Thomas_beardown // Jan 30, 2010 - 1:49pm
Whoops read the list wrong, nevermind.
#45 by Bill Barnwell // Jan 30, 2010 - 2:39pm
1) BUF 57
2) STL 47
3) WAS 46
4) HOU 43
T5) CHI/CIN/PHI 42
Of course, some teams put players on IR when they're hopeless and know they can't contribute. Good teams also put players on IR too, though, when they're competitive and need a roster spot.
Denver players played 94% of the time when they were listed as Probable, the tenth-most frequently in the league. The league-average was 92.3%; San Diego led the league, as their Probable players suited up 98.6% of the time.
#47 by Kibbles // Jan 30, 2010 - 10:50pm
Everyone raise your hands if you would have thought before the season that the oldest secondary in NFL history would have resulted in the healthiest defense of the season. Everyone with your hands up, you're a shameless liar.
I'm terrified at the prospect of how terrible Denver is going to be next season. On the one hand, I'm hoping they're bad enough to get McDaniels fired... but on the other hand, I know that if they do fire McDaniels, it'll basically just be an admission that they chased away a borderline HoF HC, dismantled the most promising young offense in the league, sided with a capologist over an ON FIRE scouting department in a power struggle, *AND* sent a miracle-working DC on his way, all for nothing. I mean, I think McDaniels is probably a terrible head coach, but I'd almost rather just keep him around than admit that all of that happened for absolutely no reason whatsoever.
#48 by loneweasel (not verified) // Jan 31, 2010 - 10:51am
Considering Pat Bowlen is a drunk senile fool, Denver fans probably should have prepared for Raider/Bills-level of owner-FAIL the moment Shanahan was gone.
#49 by Anymouse (not verified) // Jan 31, 2010 - 11:33am
All the comments about the Colts being "conservative" with injuries (i.e. sitting starters who were recovering from injury but could have played) also apply to the Saints. Just about every injured Saints player this year got a week or two longer to recover than they might have. In the case of the Saints, that philosophy was enabled by two key factors: first, for most of the season they had a multigame lead in the division race; and second, they have good depth at almost every position.
CAPTCHA: farmyard student
#51 by Steph // Feb 01, 2010 - 9:13pm
I think the 2009 Curtis Painter experiment may demonstrate that the Colts do not have any special ability to draft players but rather that a healthy and active Peyton Manning is like spackle fixing holes. Even a competent quarterback may not be able to pull out wins in situations where he does. Offense can't get on the field against Miami? No problem, Indy has The Peyton. Houston gets out to a big lead? No problem, Indy has The Peyton.
Much easier to put a team together when you have one guy who can make the rest of the fungible parts go.