2011 Adjusted Games Lost

2011 Adjusted Games Lost
2011 Adjusted Games Lost
Photo: USA Today Sports Images

by Danny Tuccitto

Peyton Manning's degenerating cervical disc and subsequent surgery dominated headlines and had far-reaching implications during the 2011 season -- and subsequent offseason -- like no other injury situation in recent NFL history. We've all had the details drilled into our heads over the past year, so I won't recap them here, but suffice it to say that enough dominos fell to have a decent-sized double-sixes tournament. Looking back, not even Tom Brady's knee injury in 2008 ended up meaning so much to so many cities.

Seemingly lost in all the hoopla surrounding Manning, though, was that plenty of other teams saw their fates determined in part by player health (or lack thereof). For instance, just ask Texans fans what they think might have been if Matt Schaub's foot injury -- and let's not forget Matt Leinart's noncompliant collarbone -- didn't force fifth-round rookie T.J. Yates into the starting lineup during their playoff run. In contrast, the 49ers' unexpected ascension into the NFL elite probably would not have happened if their "shallow chart" on defense didn't benefit from an almost pristine injury record.

Of course, every fan base plays the what-if game after their team's season is over, and injuries usually play a central role in the game. Sometimes blaming injuries is justified, and other times it isn't, so let's put some hard data to use here. Using Football Outsiders' adjusted games lost (AGL) metric, let's find out which teams were most (and least) infected with the injury bug during the 2011 regular season.

To refresh memories, the key ideas underlying AGL are that all players don't affect winning and losing equally, and missing a game isn't the only way a player injury affects winning and losing. Injuries to starters, important situational reserves (e.g., nickel cornerbacks), and injury replacements (i.e., new permanent starters) count towards AGL, whereas injuries to benchwarmers don't. Similarly, injuries that land a player on injured reserve affect AGL more than injuries that force a player to be listed as "questionable," which in turn affect AGL more than injuries that lead to a "probable" game status.

Before we get into last season's AGL totals for specific teams, let's first look at a few leaguewide trends that have emerged over the past few years, the most glaring of which has to do with the sheer amount of AGL that teams have been experiencing over the course of the past decade. Below is a graph showing the trajectory of aggregated AGL over time across the entire NFL.

Essentially, injuries cost players nearly double the number of AGL in 2011 as they did in 2002. Converting the numbers in the graph to team averages, we're talking about the average team losing 60.5 adjusted games due to injury last season, as opposed to 33.6 a decade ago. That's the equivalent of having an additional two important players miss an entire season.

The obvious question here is, "Why?" Certainly, if we focus only on the one-year increase from 2010 to 2011, there's a pretty obvious culprit: the lockout. Of course, it stands to reason that the league's increased emphasis on player safety should have at least resulted in less steep of an increase. There's also the possibility -- close to a statistician's heart -- that 2011 was just a random fluctuation.

Whatever the reason for last season's uptick, the increasing trend over a 10-year period -- or even a seven-year period if you want to look at it that way -- is unmistakable. For the long-term trend, we have several hypotheses that we're in the process of fleshing out, which you'll likely see addressed in Football Outsiders Almanac 2012 (available in July). They include, but of course are not limited to, the following: (1) players are actually "more injured" in the general sense, (2) teams are reporting more injuries, (3) injury information is more available now than it was a decade ago, and (4) the AGL method needs an update similar to how Aaron revamps DVOA every so often due to changes in the nature of the game.

One thing we can say right now with confidence is that, as far as weekly injury reports go, teams do appear to be listing more players, and the increase can be seen among both starters and reserves. Specifically, if we look at the total number of injury listings in our database, we find that 2011 (4,357) had over 500 more than in 2008 (3,755) and over 1,000 more than in 2002 (3,317).

What's more, the past three seasons suggest that teams might be engaging in a bit of what I'll call "probable dumping:" After averaging about 1,300 probable listings from 2002 to 2008, the league has averaged a shade under 2,000 since 2009. Whether you think that's due to being conservative with injuries (a la Bill Polian) or playing shenanigans with reports (a la Bill Belichick) depends on your level of dispositional cynicism. We'll remain agnostic for the time being.

A second important trend is that year-to-year AGL correlations for each team have been much higher over the past three seasons than in the prior seven. Take the Seahawks, for instance. Their AGL has gone from 78.4 (ranked 28th) in 2008 to 75.2 (27th) in 2009 to 61.5 (24th) in 2010 to 78.6 (27th) in 2011. To be sure, we're still not seeing this sunrise-esque level of consistency across all teams, so the year-to-year correlation across the league isn't massive (in the 0.30 range). Nevertheless, that's a difference from 2000-2008, when the year-to-year correlation was .09.

The final leaguewide trend, and perhaps the one that's most definitive, is that NFL teams have officially completed their 10-year plan to eliminate "doubtful" as an injury listing worthy of public consumption. Recall that the way injury reports are supposed to work is that "probable" means a 75 percent chance of that player participating in the upcoming game, "questionable" means a 50 percent chance of playing, and "doubtful" means a 25 percent chance of playing.

It's been known for years that those guidelines aren't anywhere near what teams actually follow. That said, take a look at the graph below, which shows game participation rates for both starters (complete lines) and reserves (dotted lines) based on whether the players were listed as probable (green), questionable (amber), or doubtful (red).

There appear to be two clear ways doubtful is different from probable and questionable. First, whereas reserves play far less often than starters when they're listed as probable or questionable, there's basically no difference between the two groups in playing frequency when doubtful.

Second, and more importantly, the rate at which probable and questionable players actually play has remained level over the past decade. In contrast, doubtful players have gone from playing about 15 percent of the time in 2002 to playing only about 1 percent of the time in 2011. Furthermore, it's basically been that way since 2007.

The important point here isn't so much that the doubtful rate diverges from the league guideline. We can say the same about the probable rate, and -- again -- everyone knows this already. Rather, it's that shenanigans in the context of doubtful have increased over time, and to such an extent that the doubtful listing is completely meaningless.

So, to recap, what our injury data shows is that coaches have been "probable dumping" the past three years, and have been undermining the value of doubtful with increasing fervor over the past 10.

With that out of the way, here are the 2011 AGL totals for every team:

Team 2011 AGL 2011 Rank 2010 AGL 2010 Rank
BAL 18.8 1 50.9 15
PHI 21.9 2 70.1 27
NO 24.6 3 45.0 13
DET 28.0 4 51.0 16
NYJ 30.4 5 41.9 12
MIA 31.7 6 51.0 17
TEN 37.7 7 31.6 6
SF 38.4 8 27.4 4
ATL 48.6 9 16.1 3
MIN 48.8 10 55.4 20
HOU 50.1 11 55.2 19
TB 51.4 12 60.8 23
CIN 51.8 13 62.4 25
CHI 54.5 14 12.3 2
DEN 55.4 15 51.7 18
GB 58.7 16 86.3 30
Team 2011 AGL 2011 Rank 2010 AGL 2010 Rank
PIT 60.5 17 49.9 14
DAL 62.5 18 31.8 7
KC 65.5 19 10.0 1
WAS 67.5 20 69.9 26
BUF 71.0 21 41.7 11
CLE 71.8 22 94.7 32
SD 72.6 23 39.6 10
JAC 73.4 24 37.6 8
OAK 78.0 25 30.5 5
NYG 78.3 26 58.1 22
SEA 78.6 27 61.5 24
IND 84.7 28 90.5 31
ARI 86.8 29 38.3 9
NE 97.5 30 72.5 28
CAR 109.2 31 75.0 29
STL 110.0 32 56.9 21

Let's start with a few teams -- other than Indianapolis -- that had their seasons short-circuited by injury.

With their 110.0 AGL, the 2011 St. Louis Rams became the second-most injured team since 2002. Only the 2009 Buffalo Bills had it worse off. And like those Bills, the main contribution to the 2011 Rams AGL came from their defensive backfield, where their top three cornerbacks -- four if you count early-camp casualty Jerome Murphy -- were already out for the year by Week 10.

[ad placeholder 3]

Also not helping matters were another year of carnage at wide receiver (29.0 AGL after 26.4 in 2010) and 5.7 AGL for Sam Bradford. With injuries affecting the (offensive and defensive) passing game this much, it's no wonder St. Louis went from the cusp of the playoffs to winning only two games.

The team that saw the largest absolute increase in AGL (from 10.0 to 65.5) was Kansas City, so it's no surprise they lost three more games last season than in 2010. The interesting thing about the Chiefs is that about 60 percent of their AGL came from only four (albeit vital) players: Matt Cassel, Jamaal Charles, Tony Moeaki, and Eric Berry. Otherwise, they were an incredibly healthy squad: If not for those four significant losses, they would have ranked third overall.

Moving on to teams that saw their win totals clearly benefit from increased health in 2011, we have the two playoff participants from the NFC North. Detroit went from a team ranked in the middle of the AGL pack in 2010 to one of the healthiest teams in the league last season. Most of that good fortune came at quarterback, where Matthew Stafford's newfound durability resulted in a full season's worth of improved AGL at the position.

Classifying Green Bay as a team helped by increased health seems a bit odd given that they were coming off a Super Bowl championship in 2010. But remember that they only won 10 games that year. The same way improved health can be the catalyst for a rise from mediocrity (as in Detroit), it can also play a role in a 10-6 team improving to 15-1. Healthy seasons from Jermichael Finley and Ryan Grant dropped AGL at tight end and running back from 26.6 to 1.3, which likely contributed to a 24.5% improvement in offensive DVOA.

There were also a couple of teams that enjoyed relatively healthy seasons, but failed to capitalize on their good fortune. The New York Jets finished 2011 as one of the five least injured teams in the league, ranking seven spots higher than they did in 2010, but they lost three more games and dropped from sixth to ninth in total DVOA.

[ad placeholder 4]

The Jets have nothing on the 2011 Philadelphia Eagles, though. Despite having 48.2 fewer AGL than they did in 2010, ranking 25 spots higher in team health, the Eagles lost two more games, and missed the playoffs. It's true that Michael Vick had an assortment of injuries in 2011, but it turned out that he only contributed 0.2 AGL more last season than he did in 2010 (mostly because of differences in game status). Elsewhere, Philadelphia's offensive line AGL dropped by a full season's worth of games in 2011, as did their total AGL on defense. Taking their lack of injuries into account, calling the Eagles a disappointment last season might be understating it.

Finally, two teams in 2011 were what you might call the Bizarro Eagles: teams that found a ton of success despite ranking among the most injured in the league. You might also call them "last year's Super Bowl participants." Both the New York Giants and New England Patriots ranked in the bottom quartile of total AGL, and both saw their defenses decimated by injury over the course of the season.

For New England, almost half of their 57.6 AGL on defense came in the secondary (27.6), which is bad news for a team that has five or more defensive backs on the field over 60 percent of the time. For New York, last year's injury headlines were about the secondary and defensive line, but the largest AGL contribution on defense was from linebackers (23.4 out of 53.2). By the time Super Bowl XLVI came around, the Giants were down to their fourth-string middle linebacker, Chase Blackburn, who ironically ended up making the signature defensive play of the game.

Although the Patriots and Giants certainly deserve credit for successfully overcoming injuries last season, these weren't unprecedented feats by any means. In fact, last season marked the third year in a row -- and sixth time in the last nine years -- that the Super Bowl champion ranked in the bottom quartile of total AGL. It was also the third time in six years that both participants ranked 25th or lower.

Clearly then, as much as we talk about the importance of staying healthy, last year's Giants and Patriots showed us that doing so isn't necessary for a Super Bowl run. I guess you could say that last year's Eagles showed us that too, but not in a good way.

Next week, I'll profile the most- and least-injured team units from last season. Hint: The two most-injured units weren't even mentioned in today's article.


85 comments, Last at 15 Jun 2012, 3:45pm

#1 by Podge (not verified) // May 11, 2012 - 2:31pm

How many years in the last 5 have the Rams been in the top 3 or 4 for most AGL? It seems to happen every year. Different players, same results.

Points: 0

#4 by Pottsville Mar… // May 11, 2012 - 2:57pm

I wonder if this is due to the relative skill of teams' training staffs? I know that in the NBA, the Phoenix Suns are widely rumored to have trainers that do an excellent job at keeping players healthy, while the Portland Trailblazers' trainers have mishandled injuries on a number of occasions.

I sort of doubt that this is the case, simply because NFL teams have enough money to hire the best of the best, but it might be worth exploring.

Points: 0

#2 by The Voice (not verified) // May 11, 2012 - 2:34pm

I imagine if NE's AGL dropped from 97.5 to 84.7, but they lost Brady for the entire season, they would not have done quite as well as they did last year.

Points: 0

#22 by MJK // May 11, 2012 - 10:19pm

True, but I would bet that they would have done better than 2-14.

In 2008, they lost Brady for the entire season, and had the third most AGL in the league at 79.1, and yet finished 11-5, missing the playoffs only on tiebreakers. Not quite your scenario, but almost.

The impact of Manning's loss was primarly because the Colts had no plan B at all, and not a lot of talent beyond Manning.

Points: 0

#32 by Rocco // May 12, 2012 - 2:34pm

The Pats also got to play a weak schedule that year (28th according to DVOA). Indy's schedule this year was 8th in DVOA. The Colts would have been bad this year no matter the schedule but probably not as dreadful. And the 2008 Pats benefitted from playing the NFC West (average DVOA: -21.63%) and not the NFC East (average DVOA: 18.83%).

Points: 0

#50 by The Voice (not verified) // May 14, 2012 - 9:38am

True, but my point was that there are injuries and there are injuries. Given that the odds of the Colts making the Superbowl or even the playoffs without Manning were slim and none, I would argue it was a rational and good approach to put all of their eggs in that one basket without a backup plan. If the goal is to win and you realize you possess the one thing (Manning) which is special, it makes little sense to devote tons of resources to what you will do if that resource fails - you will lose if that resource fails no matter what!

Anyway my problem with the approach as I understand it above is a simple counting stat like "time starters miss" does not remotely account for the total impact of who missed the time. The point that the Eagles were really healthy as a team is valid, but it misses the more important fact that the single most important part of the team (Vick) not only missed 4 critical games but also played injured in others, and the dropoff between him and his backup isn't captured in the methodology above.

In a similar way, you can argue from the presentation above that the Pats were unhealthy, but the fact is the single most important part of their team was healthy and productive the whole year, and thus invalidates the basic thrust of the article.

Points: 0

#52 by RickD // May 14, 2012 - 11:38am

"In a similar way, you can argue from the presentation above that the Pats were unhealthy, but the fact is the single most important part of their team was healthy and productive the whole year, and thus invalidates the basic thrust of the article."

No, really. Football is a team game.

Brady had a great year, but he wasn't even considered in the top two among QBs last season.

The Patriots had a lot of injuries, but almost all of them were on defense. They lost Koppen early and missed a lot of games from Vollmer, but the offense was essentially intact all season long.

In contrast, the defense was in shambles, esp. the secondary. Andre Carter played well enough for half the season to make the Pro Bowl, and then went on IR.

So the story isn't that the presence or absence of one player made all the difference (this isn't the NBA). Rather, the Pats maintained an elite offense that was good enough to overcome a defense that started poor and was devastated by injuries.

Points: 0

#55 by The Voice (not verified) // May 14, 2012 - 4:27pm

A Quarterback who, along with Brees, broke a 25+ year old record for yards, had 39 TDs to only 12 INTs, and ranked #3 in DVOA and EYards wasn't all that important to the teams' success. OOOOOOO-kay then.

Points: 0

#66 by TacticalSledgehammer // May 15, 2012 - 1:24am

You're totally misreading his point
"the Pats maintained an elite offense"
I would say that allows for the fact that Brady was extremely important.

My take is that it's impossible to have a metric that properly identifies the differences in "value" between a Manning and a Brady and their backups; it's inherently subjective. This is the next best thing, and it's still a useful statistic.


“Treat a man as he is, and he will remain as he is. Treat a man as he could be, and he will become what he should be.”

Points: 0

#58 by Jerry // May 14, 2012 - 6:55pm

One correction: "played injured in others" is why probable/questionable/doubtful are included (proportionally) in the statistic.

To your larger point: we don't have good enough individual metrics to evaluate just how devastating the loss of a given right guard is. An injury to Peyton Manning obviously has a bigger effect than an injury to Rex Grossman, but quantifying that is a large undertaking, and it's harder at other positions. AGL is a good start at describing how injured a team was in any given year. Adjusting for the individuals involved would be an improvement, but if FO can come up with ratings for those individuals, AGL will be a relatively unimportant use.

Points: 0

#68 by The Voice (not verified) // May 15, 2012 - 8:37am

I get annoyed when someone presents a statistic, and when you point out the obvious flaws in that statistic for any meaningful analysis, you get the whole "this is just a start." The writer makes statements drawing conclusions about how great a season team A has despite a lot of injuries, and how dissapointing team B had even though they were healthy, when you consider the AGL measure. However not considering the other factors which make up the quality of games lost totally ignores the fact that there are games lost and there are games lost.

A more honest article would have at least admitted that this kind of counting statistic is so flawed as to be almost without real value. If you are presenting a statistic so flawed it will lead to the belief that just because you can measure something, it MUST be meaningful. That is an extremely common fallacy of the information age, and when I see such an obvious abuse of it I sometimes speak up. There's a lot of data but little insight in such work.

Points: 0

#73 by Jerry // May 15, 2012 - 10:01pm

"The best is the enemy of the better."

I get annoyed when someone complains that a new statistic doesn't include every possible adjustment, usually to help confirm the complainer's preconceived notion. The work that people here and elsewhere are doing to help us understand football better isn't easy. Some of the metrics they come up with aren't useful, and many can be (and are being) improved, but it's all part of the process of coming up with better stats. Constructive criticism is a useful part of the process.

AGL is an attempt to simply describe how injured teams were. Fans of every team can tell you how their team was hurt by injuries, but even a relatively simple counting stat can give us an idea of which teams were hurt more. Yeah, it would be nice if there was a way to compare the value of Aaron Smith's injury to Andre Johnson's (and if you can do it, please go ahead), but to claim that AGL is "almost without real value" because it doesn't know just how much losing any given player hurts his team presumes a level of information that we don't have. Not every statistic is going to provide a great deal of insight; sometimes collecting data is useful in and of itself, and it's certainly a step along the way to more insightful metrics.

Points: 0

#76 by Mr Shush // May 16, 2012 - 6:35am

Maybe a weighted average of the player's AV over the previous three seasons, with some kind of draft slot based adjustment for players in their first two or three years?

Points: 0

#74 by LionInAZ // May 16, 2012 - 12:24am

I'd say that the difference between the 2008 Patriots and the 2011 Colts is more about the relative value of Matt Cassel and Curtis Painter than it is about Manning vs. Brady.

Points: 0

#77 by Thomas_beardown // May 16, 2012 - 10:39am

I'd say it was more about the other 20 starters myself.

Points: 0

#78 by AJ (not verified) // May 16, 2012 - 1:21pm

seconded, the colts sucked no matter who was at qb, whether it was painter or collins or orlovsky. Outside of one season when he had a good td to int ratio(see alex smith), matt cassel has been nothing short of putrid-whats more likely, he was great with ne then hit his head and became mediocre - or that ne made him look much better than he ever was.

Points: 0

#79 by Marko // May 16, 2012 - 1:31pm

If the Colts only had 20 other starters besides the QB, then that provides an additional reason why they were so bad last year.

Points: 0

#3 by zlionsfan // May 11, 2012 - 2:49pm

The Lions not only got a full season out of Stafford, but also got 79 out of 80 possible starts from their OL. (Corey Hilliard started one game for Gosder Cherilus, who ended up playing in the game anyway.) Sure, the line isn't great, but at least they played virtually the entire season as a unit. That continuity has to make a difference.

Their receiving corps was also healthy for the majority of the season: Scheffler missed one game (although I can't recall if it was due to injury or something else), but everyone else was available the entire season. That really helped to offset the injury problems they had at RB.

Points: 0

#5 by akn // May 11, 2012 - 3:28pm

Regression to the mean + CJ on the Madden cover... Look out next season, I guess.

Points: 0

#64 by LionInAZ // May 15, 2012 - 12:21am

On the other hand, the Lions' secondary suffered from time lost to injuries, especially to Chris Houston and Louis Delmas down the stretch, as well as losing Erik Coleman most of the season.

Points: 0

#6 by akn // May 11, 2012 - 3:34pm

Not to be a statistical snob, but that AGL over time graph should be individual points for each year (not a connected spline) with a linear regression line. It would illustrate your point more clearly, and makes 2011 look less like a fluke (I don't believe it is).

Points: 0

#7 by Thomas_beardown // May 11, 2012 - 3:42pm

One thing that this study misses is how many injuries a team has at one time.

I think 40-50 AGL is manageable for a team if it's spread out over the year, but when it all hits at once in the last 5 games, they have no chance.

Points: 0

#8 by Ferguson1015 // May 11, 2012 - 3:45pm

Or if injuries pop up at the beginning of the season, only to have the starters return at the end of the season just before the playoffs.

Points: 0

#12 by Bjorn Nittmo // May 11, 2012 - 4:34pm

That was somewhat the case with the Giants. While they did lose a frightening number of defensive players in the preseason to the IR, several key defensive players got healthy by the end of the year, enough to turn them into a pretty wretched unit to a good one in the postseason.

(and I believe the Jets lost only 1 more game in 2011 than in 2010)

Points: 0

#15 by Marko // May 11, 2012 - 6:22pm

"One thing that this study misses is how many injuries a team has at one time."

Or if one of the injured players (who is knocked out for the last 6 games of the season) is named Jay Cutler and his backup is named Caleb Hanie. Or if two weeks after that injury, another player (Matt Forte) on the same team is injured and knocked out for the season, replaced by a backup (Marion Barber) who directly costs his team two games by (1) failing to line up properly on the line of scrimmage, resulting in a TD pass to him being nullified due to the penalty for illegal formation, and (2) in another game, allowing the defense to push him out of bounds to stop the clock inside the two minute warning when all the team had to do was run out the clock, leading to the opponent gettting the ball back, tying the game on a long field goal, and then winning the game in overtime when Barber again makes a critical mistake, losing a fumble while the team is in field goal range, letting the opponent get the ball in decent field position and drive for the game-winning field goal.

Points: 0

#29 by Jerry F. (not verified) // May 12, 2012 - 10:47am

All that said, I like what Barber offered last year. It seemed like the first time in a long time the Bears had been able to run in short yardage effectively. I'm actually sort of sad to see him go. Maybe I'm an idiot.

Points: 0

#46 by BigCheese // May 13, 2012 - 11:43pm

You'd rather have Barber than Bush? Yes, you're insane... :)

- Alvaro

Phil Simms is to analysts what Ryan Leaf is to NFL QBs

Points: 0

#9 by Ryan D. // May 11, 2012 - 3:52pm

"Next week, I'll profile the most- and least-injured team units from last season. Hint: The two most-injured units weren't even mentioned in today's article."

I'll take the Carolina Panthers Defense for 2000, Alex. If they get Jon Beason (missed 15 games) and Ron Edwards (16 games) back and close to 100% healthy, and can find a role for Thomas Davis (14 games) behind Luke Kuechly and James Anderson, the overall run defense should be greatly improved. Add a little more development and refinement for their dynamic offense, better production from their kicker and punter, and throw in a little luck, and this could be a surprise playoff contender for 2012.

Points: 0

#30 by Joseph // May 12, 2012 - 2:12pm

I think he means a unit of the D and O--in this instance, it's IMO Carolina's LB's. I'm going to pick CHI's O-line for the offense.

Points: 0

#44 by Joseph // May 13, 2012 - 7:23pm

Danny, if you are insinuating that I got both guesses right, do I win a prize or something? Or just the ability to boast that I was right?

Points: 0

#10 by cfn_ms // May 11, 2012 - 3:53pm

if you guys could put out this data for 1-A college football as well for the last couple years. Would be really useful, and no one else is really doing it (Phil Steele posts just the top and bottom injuries lost, not the large middle list).

Points: 0

#11 by chemical burn // May 11, 2012 - 4:17pm

Wow, I knew they Eagles had great injury luck last year (I can only recall Vick and Young missing games) but I had no idea it was that dramatic. Also, wondering how AGL accounts for Young missing the first few games of the season where Vick went out and they had to bring in Kafka. It probably cost them the Falcons game (well, that and terrible, terrible defense) and played a role in the (probably lost cause) Giants game. How do bench-warmers count in that sort of context?

Anyway, I think injury luck for the Eagles didn't cause under-performance - they had a lot of depth and could have weathered injuries at just about every position without much drop-off. If anything, Babin and Nate Allen missing some games might have helped them. Mainly kidding, but I think they're a team designed with a good amount of depth, especially on o-line, d-line, CB and WR, so I'm not hugely worried - now, if Mr. Frequently Injured Bell goes down after Peters, I won't be thrilled...

Points: 0

#35 by Pat (filler) (not verified) // May 12, 2012 - 11:53pm

Anyway, I think injury luck for the Eagles didn't cause under-performance - they had a lot of depth and could have weathered injuries at just about every position without much drop-off.

It should be noted that while this is true, it's not really how you normally think of as "depth" in the defensive sense. I agree I don't think injuries would've really done anything at safety or linebacker... because they were all pretty uniformly mediocre anyway.

Points: 0

#56 by chemical burn // May 14, 2012 - 4:45pm

Oh agreed - but that's mainly a result of a fairly consistent policy on their part: always make sure the o-line, d-line, CB and QB are as good as you can get them no matter the cost, let the chips at S, LB, RB and WR fall where they may. There's almost never a significant difference between their low-upside, mediocre back-up LB's and their mediocre starting SAM and WILL backers but they also have all 6-8 starter caliber guys ready to go on d-line, year-in, year-out. (And they also seem to have a higher than normal amount of guys specifically trained to be swing guys on the lines, so if a T or a G goes down, their best back-up can jump right in with little fall-off.)

Points: 0

#57 by Thomas_beardown // May 14, 2012 - 4:57pm

Under Reid the Eagles have drafted more 1st and 2nd round receivers than any other position except DT (tied at 5). Plus the whole TO experiment. They put a premium on receiver, they just kind of suck at evaluating it.

Points: 0

#59 by AJ (not verified) // May 14, 2012 - 8:32pm

Yeah I don't understand how this myth got started that the eagles build solely through the lines and ignore skill positions. Outside of linebacker, they've not behaved any differently than any other team. They went after Asante samuel, they drafted nate allen in the 2nd round, they signed namdi and traded for drc. Tulus mentioned the receivers.

I actually think its a myth in general that teams draft based on scheme. Maybe the raiders do, overemphasizing speed, but everyone else drafts based on talent.

The 49ers are a defensive team and yet went offense with 4 out of their last 5 first round picks. Does anyone really know what the patriots look for when they draft players? In fact, I suspect they didn't intentionally build a two tight end offense, they just happened to draft the best players available who happened to be tight ends, similar to how they drafted vareen and ridley with their 2nd and 3rd round picks last year despite having woodhead, law firm, and kevin faulk.

Points: 0

#60 by Thomas_beardown // May 14, 2012 - 8:41pm

Bellichick has drafted a lot of tightends and has always valued players who can do multiple things over players really gifted at one skill. So I don't think it's an accident their offense is built around tight ends.

Points: 0

#61 by Pat (filler) (not verified) // May 14, 2012 - 10:34pm

Yeah I don't understand how this myth got started that the eagles build solely through the lines and ignore skill positions. Outside of linebacker, they've not behaved any differently than any other team.

They ignore RBs in the first round, but this is common for a lot of good teams. RBs, LBs, and safeties are round 2+. Period. The end. They've basically stated this in interviews, and it's pretty clear in the results of the draft. QB's the #1 priority. Then OL/DL. Then WR/CB. Then you fill in LB, TE, RB, and safeties in round 2+.

Also, they don't really suck at evaluating receivers. They used to. Since 2004, they've been pretty good at it. Every receiver past the 4th round has stuck for quite a long time: the worst of the 4 is Reggie Brown, and that's a pretty good record.

Points: 0

#69 by Dean // May 15, 2012 - 9:25am

Even Safety, they've expressed a willingness to use a first round pick if they're convinced about "their guy."

Reid stated in interviews that he wanted to pick Brandon Merriweather a few years back. When "his guy" was gone, they traded down and picked Kevin Kolb instead, but the original target was Merriweather.

Points: 0

#82 by Pat (filler) (not verified) // May 17, 2012 - 11:27am

Yes, but that still would've been a relatively late pick. They didn't trade up to get him, for instance, because you just don't do that with safeties. Late first round is probably the highest they would ever go. It's not worth it. Instead they said "meh, whatever", traded down, and drafted Kolb.

Points: 0

#70 by chemical burn // May 15, 2012 - 11:15am

First round picks and investing in depth are two very different things. In fact, they are pretty close to being unrelated things. They've invested decently at WR#1 and then very little behind it (other than hitting on Maclin and Jackson in consecutive years.) But you might have noticed how hesitant they were to give Jackson (their 2nd biggest star) a new contract and how they never hesitate to pull the trigger on contracts for LT, QB, DE or DT. Vick, McNabb, Thomas, Runyan, Justice, Cole, Patterson, Sheldon Brown, Herremans - all these guys got wrapped up in big, long-term deals without hesitation. And on top of that, they're constantly bolstering depth at those positions while treating the depth chart at WR as an after-thought (such as when they let a good prospect in Brandon Gibson go for pennies on the dollar - they'd never do the same thing with an o-lineman who showed the same amount of promise.)

I'm sure you're aware "first and second rounders" is an arbitrary cut-off point that misses what I'm saying - is the success with Jason Kelce and Todd Herremans somehow indicative that they don't have depth at those positions? Also, I do not give a shit how they found the players they found, how they developed them, whether they are draft picks or free agents - the fact is they are willing to go into the season with little depth at RB, LB and S but always have tons of depth at o-line an d-line. It's too consistent to be a mistake.

If you look at their entire drafts, they average under 1 WR per draft under Reid (they literally didn't draft a WR several years under Reid despite their frequent use of 3 receiver sets in his offensive system.) At OL, DE, DT and CB, they go bananas in every draft, throwing sheer numbers at the positions: in 2010, for instance: 4 picks for d-line, 5 DB's (Clayton was drafted to play S) and a single WR. That's one draft, but that's their MO - tons of picks at DL, DT, T, G and CB, a few picks at WR or RB, surprising QB's taken earlier than expected. Then on top of that, they bring in the absolute free agents at d-line, o-line and CB with regularity: Cullen Jenkins, Jason Peters, Javon Kearse, Jason Babin, Asomugha, Samuel, DRC. Name me their last big fee agent signing at LB, S or WR. TO. Nine seaons ago. Other than him, it's a list of bargain basement cut-rate deals: Trotter 2.0, Will Witherspoon, Takeo Spikes, Jarrod Page, Sean Jones, Kevin Curtis, Steve Smith, even question-marked DeMeco Ryans - a bunch of dubious players that had few takers.

In short, you just don't know what you're talking about. (If you want to argue that the Eagles' aren't great at developing LB's, that I would accept that. They have spent a few picks on them and yielded few results. But there's no arguing their attempts to address the position have gone with quick fixes, stop-gap solutions, projects and players nobody wants.)

Points: 0

#71 by Thomas_beardown // May 15, 2012 - 11:39am

There are only 2 drafts when the Eagles didn't take a receiver. 2004, when they traded for some guy named Terrell Owens, and 2007.

Other than that, you're arguing a bunch of things I didn't actually say.

(this is also the same number of drafts they choose not to take an offensive lineman)

Points: 0

#72 by AJ (not verified) // May 15, 2012 - 2:48pm

No one denies they shortchange their linebacking core, but everywhere else it feels like they are no different than any other team. I think its purely a myth that the eagles believe in prioritizing strong lives vs other areas. One, theres no evidence of that, and 2 you could argue thats just regular old fashion coach speech like we need to establish the run and play physical.

Points: 0

#81 by Pat (filler) (not verified) // May 17, 2012 - 11:25am

Not... entirely. If the Eagles drafted in the top 5, they'd never draft a running back. Ever. Period. I actually think a lot of teams are like that, but those teams tend to be good (because drafting a running back high is dumb) and so they don't get the opportunity to draft high.

For the most part, I think it's safe to say the Eagles devalue linebackers, running backs, and safeties, but, well... most good teams devalue those players anyway.

Points: 0

#83 by Thomas_beardown // May 17, 2012 - 11:36am

Right, it's their treatment of linebackers that's at odds with other good teams. However, not receiver as was asserted above.

Points: 0

#13 by mschuttke // May 11, 2012 - 5:14pm

Now I think people can have some answers to "Why didn't Belichick pick a defensive back with one of those two first-round selections?!"

Points: 0

#53 by RickD // May 14, 2012 - 11:43am

We knew Dowling was coming back.

There was a lot of talk about the need for a safety, but it's hard to find fault with the picks he did make. I'd be reluctant to take a safety in the first round unless it was truly a pick based on talent, not need.

Points: 0

#14 by mschuttke // May 11, 2012 - 5:14pm

Now I think people can have some answers to "Why didn't Belichick pick a defensive back with one of those two first-round selections?!"

Points: 0

#16 by CX (not verified) // May 11, 2012 - 7:09pm

It is also worth noting that it is important how critical the decimated positions are to that team's success. Take the Giants for example, having your LBs wiped out is far preferable to having your D Line crushed by injuries. Likewise, GB in 2010 it did not make a huge difference whether Jones, Walden, or Zombo was the OLB opposite Matthews, so even huge injury numbers at that position in regards to games lost matterd little

Points: 0

#20 by Danny Tuccitto // May 11, 2012 - 9:55pm

Just an FYI: The plan for the book is to incorporate this idea into an updated version of AGL.

Points: 0

#48 by Mr Shush // May 14, 2012 - 9:10am

How much detail do you hope to go into? Positional value only, or quality of starter lost as well?

Points: 0

#17 by AJ (not verified) // May 11, 2012 - 7:57pm

The problem also is compounded by the fact that certain schemes and players are more intricate to their respective team's win loss than others. The most obvious example is peyton manning but how does one really weigh a qbs loss? I remember reading several stat sites attempt to predict just how much a hall of fame qb means to a team. Most often said it was 4 wins and yet the colts collapsed. Similar story seemed true with cutler as well who many probably felt at the time was only moderately valuable.

On the flip side, there are countless examples of qbs being injured and teams surviving regardless-big ben being one.

Again, it boils down to the fact that not all injuries are created equal because not all players are created equal and since we have no real way of measuring individual player's contributions, we can't expect more substance out of agl other than an outline.

Points: 0

#18 by Thomas_beardown // May 11, 2012 - 8:48pm

The problem with saying that Peyton Manning is worth 4 wins is that football doesn't work that.

Manning might indeed be worth 4 wins over an average QB, but that doesn't mean a team would 4 for games with him. It means they'll have a quarter of a win in every game. What I mean by this is a game that isn't decided until the last seconds or overtime is essentially a toss up. Each team really deserves about 1/2 a win for a such a game. So if you replace a QB who is average on one of those teams, with Manning it will turn into a blow out.

If you think about this, you realize 4 wins is HUGE. Try adding 25% to your win probability by adding a single other person to your team.

As for the Colts, this is what they lost, and they were barely winning games with Manning. So they went from ~55% win probability in each game to a 30% chance. That would be with an average QB, since Painter was even worse you can knock of another 10-20% and boom, there's your 2-14.

Points: 0

#23 by Led // May 11, 2012 - 10:21pm

That's a very insightful way of looking at it. Thanks.

Points: 0

#19 by BaronFoobarstein // May 11, 2012 - 9:00pm

Did the Steelers really survive though? They limped into the playoffs (rimshot) and then went one and done.

Points: 0

#21 by Danny Tuccitto // May 11, 2012 - 10:07pm

I think the underlying issue here is the difficulty we have modeling interactions between players. For instance, you can take Manning's individual value, production, whatever, and say, "If team X loses that amount of QB value, then history says they'll lose Y more games." What we can't say -- but what we actually want to say -- is, "If team X loses that amount of QB value, then it also loses Z amount of WR value, T amount of RB value, W amount of OL pass protection value (since QBs are big contributors to sacks), and not to mention S defensive efficiency (because of having to defend shorter fields, etc.), which history says will lead to Y*Y more losses."

IND was so dependent on Manning that there were consequences all over the field, leading to a kind of snowball effect on team losses.

Points: 0

#27 by MTR (not verified) // May 11, 2012 - 11:00pm

I've compared the Indy offensive to a formula 1 car: it has been built to exactly fit Manning and nobody else. Even if you have another good QB to put in you won't get the same results, since the new guy will have different strengths and weaknesses.

Imagine if the top three or four QBs were randomly switched between teams. Everybody would still have roughly as good a QB, but I bet everybody's results would drop.

Points: 0

#38 by RichC (not verified) // May 13, 2012 - 12:15pm

The Colts were 10-6 the year before last.

Their Schedule went from to 18th to 8th, and they were a completely average team by DVOA the year before.

So yeah, between the differences in schedule, etc, the colts were probably only an 8-8 team on their new schedule anyways, losing Peyton puts them at 4-12 (if hes worth 4 wins).

Which is close enough to 2-14 to be believeable, especially when Curtis Painter was significantly below replacement level.

Points: 0

#24 by MJK // May 11, 2012 - 10:25pm

One thing I always wonder is whether it's worse to lose a bunch of players all at the same position, or all on the same side of the ball but different positions. For instance, it is rougher to lose four CB's, or a CB, a S, a LB, and a DE?

I think getting hit with two injuries at the same position probably hurts worse than two injuries at different positions, but after that, it probably shifts towards being worse to lose starters at multiple positions. In other words, losing two CB's is worse than losing a CB and a LB, but losing a CB, a LB, and a DE is worse than losing three CB's...

I would also wager that barring losing a defensive player or an OL is worse than losing an offensive skill player barring QB. You can playcall around a given skill position, but the other team decides where they attack your defense or your O-line.

Points: 0

#26 by Jerry // May 11, 2012 - 11:00pm

If we're talking about losing guys during the season, I think losing multiple players at one position is worse. Once a team starts having to play street free agents (the definition of replacement level), they're in trouble.

OTOH, speaking generally, the dropoff from the third-best left guard to the fourth is probably less than from the first to the second.

Points: 0

#31 by Rocco // May 12, 2012 - 2:27pm

I think it depends on the team. For example, the Steelers defense can survive injuries to everyone not named Polamalu. He goes down and the defense goes to hell in a handbasket. The Bears D could probably survive waves of injuries to their secondary but would look awful if Urlacher and Briggs went down.

Generally though I'd think you would prefer the injuries spread out to various units. It's really hard to survive when an entire unit gets carpet-bombed with injuries. Teams usually have a 2nd string guy that can do a reasonable impersonation of a starter, but when you get to the 4th string/street FA players it can get ugly unless the rest of the team is really good.

Points: 0

#34 by Theo // May 12, 2012 - 7:57pm

Losing D-line, linebacker, offensive line, TE... you can game plan around that. Play another front, slide the line, stuff like that.
Losing QB, CB, S, WR - not so much.

Points: 0

#49 by Mr Shush // May 14, 2012 - 9:15am

It's so variable, though. Losing Andre Johnson hurts the Texans far more than losing Greg Jennings would hurt the Packers, not primarily because of the extent to which Johnson's a better player than Jennings, but because the rest of the Texans' competent receivers are guys like Walter and Daniels who are better at catching than getting open (never mind deep) which allows them to be hugely productive when Johnson's drawing the coverage but can render them somewhat invisible when he's not.

Points: 0

#25 by WillF (not verified) // May 11, 2012 - 10:58pm

Another point that might help explain the giants, is asking when the injuries where. I would expect that for the Giants the fact that the team was much healthier in the playoffs than the regular season (that is, Osi and Tuck playing close to 100%, health along the OL, etc.) explains the huge discrepancy in the team's performances.

Points: 0

#28 by AJ (not verified) // May 12, 2012 - 3:15am

this is really echoing what danny said earlier. We still can't parse what effect certain players have on the total of their team. For instance, the giants could probably survive most of their dline being injured, but would they have really been the same if osi had survived and say jpp was out? Or take the giants in 09 whos defense went from being very good to abysmal because kenny philips got hurt and cc brown was the replacement. My point is, its nice to quantify injuries, but only after the fact can we make judgements about how impactful they were.

I hate bringing up the dead horse manning brady thing, but i do want to add, back in 2007, at least on the surface, the vast majority of us might have treated both manning and brady's injuries as having relatively symmetric results. After the fact, what conclusions can we make? I think its obviously too simplistic and naive to conclude well brady is a overrated qb and manning is underrated, but rather, it shows us how little we or really football people truly understand the dynamics and inter-player connections there are within games or seasons.

Points: 0

#40 by takeitdown // May 13, 2012 - 1:13pm

I agree the difficulty is pretty enormous. It's so multifactorial it becomes quite tricky quite quickly. For instance, how large a drop a team sees when it misses a starter depends first and foremost on how good the 2nd stringer is at that position. Per your example, Matt Cassel, though not great, is likely a big step up from anything Indy trotted out last year. A David Garrard as backup will have a lot better time than a 6th round rookie (usually.)
It also depends, as you mention, on the team around. A team with a great D, and very solid OL may have a great QB and be dominant. But even if they lose their QB, as long as the replacement is decent, that will still be a good team. Whereas a team which isn't as strong across the board, but has developed a scheme over time to fit the QB, may utterly collapse without said QB.
Another factor, which is HUGE, is coaching ability. If you have a coach who plays his scheme, no matter what happens, injuries can be pretty crippling, sometimes the backups simply can't get it done. While if you have a coach who is able to adjust his schemes, he won't have a Ben Roethlisbergeresque backup trying to play in a Peyton Manning offense. He won't keep the safety as his focal point on defense if he loses his star. This makes as large a difference as any of the rest...can the coach actually adjust his schemes to the players? If so, injuries are far less damaging.
So, it seems, to actually come up with something predicting how much certain injuries will affect a team, you have to at least 1) compare each player to replacement level, and compare their backup to replacement level, and note the difference, 2) see how strong the team is outside of that position group, for relative importance of the position group to the team and 3) somehow assign a coaching value to capture some of the difference between a coach who will completely adjust his scheme if Tebow comes in for Sanchez, versus one who sticks with a "next man up has to do the same job" philosophy.
The problem is a number of these are quite difficult to apply non opinion based numbers to (like coaching), but I think that set of 3 gets you a lot closer to the answer. For instance, you know sort of intuitively that Indy can't survive without Peyton, because their backups were poor, their other groups were fairly weak and even their defense depended on scoring heavily (pass D defense) and so even a decent QB might leave the defense flailing, and their coach isn't known as a guy who can adjust on the fly.
Conversely, when a similar thing happens in New England, or Philly, or, I'd guess, New Orleans, you still lose a lot, but the ability of the coaches to adjust, and the strong position groups outside of QB give the teams a much better chance of succeeding.

Points: 0

#36 by RC (not verified) // May 13, 2012 - 12:02pm

" or playing shenanigans with reports (a la Bill Belichick) "

Why does this keep coming up? Belichick follows the letter of the law on injury reports, IE, if the player is held out of any portion of practice for health (preventative or recuperative) he must be listed on the injury report.

Brady is listed every week as probable because he misses certain throwing drills that the backups are required to attend.

Points: 0

#39 by Thomas_beardown // May 13, 2012 - 12:50pm

I'm pretty sure they changed the rule actually. I player has to be listed as full participation, partial participation, or did not participate in practices now. However you don't have to list a player who has partial participation as probable any more. He just does it for fun.

Points: 0

#45 by Jimmy Oz (not verified) // May 13, 2012 - 8:30pm

Brady was listed 4 times last year, & two of those were play-offs v Broncos & Ravens.

Points: 0

#37 by Topas // May 13, 2012 - 12:11pm

Interesting article.

But from a stats side. A graph which axis starts at 1000 on a stats site so that the increase in injuries looks more impressive...
Would have expected better than that.

Otherwise great article.

Points: 0

#41 by Danny Tuccitto // May 13, 2012 - 2:58pm

Just for the record, are you seriously insinuating that a near doubling of AGL over the past 10 years is merely a byproduct of my axis decision-making? I mean, I share your dismay when I see people using axes in the way you describe, but what's displayed isn't that at all. It's not like total AGL "rose" from 1,000 to 1,200 or something.

Points: 0

#47 by Dr. Statistics… (not verified) // May 14, 2012 - 8:54am

Darrell Huff and Edward Tufte devotees are almost always going to complain when an axis for a counting statistic doesn't originate at 0, especially if there's no squiggle to indicate a discontinuity in the axis. As a scientist, I would prefer that almost all axes for counting statistics originate at 0, but as a realist, I've learned to adapt to a world with suboptimal graphs.

Points: 0

#62 by Pat (filler) (not verified) // May 14, 2012 - 10:45pm

It doesn't make sense to start at zero, though. The idea of a team going a season without ever listing a player on the injury report is completely ludicrous. It seems silly to complain about the axis starting at 1000 when the 0-value of the statistic is unattainable. You could've just made the statistic "AGL' = AGL - 1000."

It's like plotting temperature. You don't want to start every graph at absolute zero. It doesn't help anything for weather, for instance, because the 'typical' temperature isn't anywhere near absolute zero.

Points: 0

#42 by AJ (not verified) // May 13, 2012 - 3:14pm

LOLOL- my personal favorite axis shenanigans involved my friend trying to get me to invest in some random tech company. He showed a grap with the earnings showing a strong steady rise that traveled across the map. Only, the graph only traveled across three months and basically started and 25 dollars and ended at 35.

Points: 0

#43 by DisplacedPackerFan // May 13, 2012 - 3:27pm

I am curious if you have 2001 and 2000 data (or earlier). If 2001 and 2000 were in the 1500 - 1600 range then suddenly the low points in 02-03 and 05-06 look more like the anomalies and last year looks like a lock out anomaly.

Other than that I think you covered things well. It's just something I was curious about. Like when I look at economic indicators and chuckle at a single curve fit when it looks way more like a step function with exponential growth at the start of the steps and you can draw some decent correlations with new economic drivers. It's not just the boom/bust/bubble type stuff, but the economy grows when there is a new popular industry, there can often be bubble around it too (like the tech bubble, but again tech/information has been driving the economy since the late 80's there was a bubble with it too, but it's still a big driver).

So as others have said just doing a simple curve fit is not the always the best way to represent things. I'm not saying that is the case with this data, but I could see it being that way if prior data looks different than expected based on the curve fit.

Points: 0

#51 by Marima (not verified) // May 14, 2012 - 10:36am

I'm curious as to exactly what shenanigans with the injury report Bill Belichick played. If you really are "agnostic" on the subject, then why bother inserting that accusation in your column at all, without anything to back it up? If there are no facts behind it, then a baseless rumor shouldn't be perpetuated here by FO.

Points: 0

#63 by Pat (filler) (not verified) // May 14, 2012 - 11:09pm

Seriously? Belichick's injury report 'shenanigans' have been around for years now. Brady was probable with a shoulder injury for something like 3 years in a row, if memory serves. It was a big deal a little while ago when Brady was actually removed from an injury report.

Points: 0

#65 by LionInAZ // May 15, 2012 - 12:37am

See comment 45. Someone is lying, mistaken, or has an axe to grind against Bill Belichick.
Someone else can sort it out.

Points: 0

#67 by BaronFoobarstein // May 15, 2012 - 2:33am

I couldn't find any zrchived injury reports, but I did find numerous references to Brady's constant appearance as "probable." Here is the article with the most concrete data I read: http://www.boston.com/sports/football/patriots/articles/2008/09/06/bradys_three_year_streak_on_injury_report_comes_to_an_end/

Points: 0

#75 by LionInAZ // May 16, 2012 - 12:50am

And, of course, Brady went down in the season opener and missed the rest of that season.

OK, so I'll retract my previous harsh words. Still, considering that starters listed as 'probable' have about a 95% actual game participation rate, I'm not sure how serious these supposed 'shenanigans' really are. If Brady had been listed as 'questionable' all that time, I'd be much more inclined to cry foul.

Coaches everywhere try to game the injury reports. Jim Schwartz infuriates Lions with his consistent refusal to discuss injuries.

Points: 0

#80 by Pat (filler) (not verified) // May 17, 2012 - 11:20am

Shenanigans (n):

1) Secret or dishonest activity or maneuvering.
2) Silly or high-spirited behavior; mischief.

No one's saying it's "serious" - even the first definition carries the connotation of the second (seriously, who would use the word 'shenanigans' to mean something serious? Just look at it!). In fact, I'm virtually certain that Belichick was doing it to basically state his opinion to the NFL that he thought injury reports were stupid, and 'probable' is extremely stupid.

Belichick's "shenanigans" mostly stopped recently because the NFL told teams very clearly to quit it, and when they have to list players on the report (it must be reported that a player missed practice).

Plus I think Belichick got bored with it.

Points: 0

#85 by BarbaraB (not verified) // Jun 15, 2012 - 3:45pm

Whatever the stats say, Patriots fans believe that Tom and the rest of the guys can still make it next season, injured or not.

  • There are several gambling web-sites on the market which claim to be the perfect on-line gambling site offered but the fact is this fact is definitely the one with the most important gambling online bonus deals plus the highest payouts.

Points: 0

Save 10%
& Support Danny
Support Football Outsiders' independent media and . Use promo code WRITERS to save 10% on any FO+ membership and give half the cost of your membership to tip Danny.