2015 Adjusted Games Lost
by Scott Kacsmar
If the Seattle Seahawks are a DVOA dynasty, then the New York Giants are an AGL dynasty after leading the NFL in adjusted games lost to injury for the third consecutive season. Not only have these been league-leading totals, but they are the three-highest totals in our database going back to 2000. Needless to say, there will not be a fireworks display to celebrate this unfathomable achievement. No other team had led in back-to-back years since 2000, because injuries are expected to regress towards the mean. But for the Giants, 2015 was just another year of black-and-blue hurt with no postseason play.
The Giants have had very durable quarterback play with Eli Manning's ironman streak alive and well, but almost everywhere else this team has just been wrecked by injury for three years running. This year, the Giants went into Week 1 in bad shape with left tackle Will Beatty (torn pectoral in May while lifting weights), Victor Cruz (setbacks in PCL recovery) and Jason Pierre-Paul (fireworks accident) all on the mend. The Giants placed four safeties on injured reserve before September. You knew a significant Jon Beason injury was right around the corner, and even blocking tight end Daniel Fells had the battle of his life with MRSA.
New York earned the three-peat the hard way. Teams like Baltimore and New England may have had more high-profile injuries, but by sheer volume, no one topped the Giants again.
These numbers do not simply add up the number of games missed. With Football Outsiders' adjusted games lost (AGL) metric, we are able to quantify how much teams were affected by injuries based on two principles: (1) Injuries to starters, replacement starters, and important situational reserves (No. 3 wide receiver, nickelback, etc.) matter more than injuries to bench warmers; and (2) Injured players who do take the field are usually playing with reduced ability, which is why AGL is based not strictly on whether the player is active for the game or not, but instead is based on the player's listed status that week (IR/PUP, out, doubtful, questionable or probable).
As long as NFL teams are solely responsible for producing weekly injury reports, we cannot say that every single injury has been accounted for, but secrecy is an unavoidable aspect of this part of the game.
Note: these numbers are subject to change between now and the release of Football Outsiders Almanac 2016, as we may adjust which players qualify as "replacement starters" or "important situational players."
2015 Results
The following table lists the AGL totals and rankings along with the results from 2014 for comparison. This data is only for the regular season.
| Team | 2015 AGL | Rk | 2014 AGL | Rk |
| CIN | 28.2 | 1 | 71.7 | 16 |
| ATL | 28.8 | 2 | 93.8 | 25 |
| SEA | 40.1 | 3 | 72.9 | 18 |
| CAR | 50.9 | 4 | 51.4 | 6 |
| DAL | 51.7 | 5 | 76.1 | 19 |
| PHI | 52.0 | 6 | 48.6 | 5 |
| KC | 54.9 | 7 | 98.8 | 26 |
| NO | 56.1 | 8 | 58.0 | 9 |
| GB | 56.2 | 9 | 41.9 | 3 |
| DEN | 56.7 | 10 | 36.9 | 1 |
| OAK | 57.6 | 11 | 103.6 | 29 |
| MIN | 59.0 | 12 | 56.1 | 8 |
| NYJ | 61.6 | 13 | 41.5 | 2 |
| ARI | 62.5 | 14 | 72.8 | 17 |
| MIA | 63.4 | 15 | 79.5 | 21 |
| HOU | 64.8 | 16 | 59.9 | 11 |
| Team | 2015 AGL | Rk | 2014 AGL | Rk |
| IND | 65.1 | 17 | 104.7 | 30 |
| TEN | 65.2 | 18 | 79.7 | 22 |
| PIT | 67.1 | 19 | 42.8 | 4 |
| JAC | 68.9 | 20 | 77.8 | 20 |
| CLE | 70.7 | 21 | 67.0 | 14 |
| TB | 75.0 | 22 | 87.2 | 23 |
| DET | 76.7 | 23 | 67.6 | 15 |
| STL | 80.3 | 24 | 64.1 | 13 |
| BUF | 80.5 | 25 | 59.0 | 10 |
| SF | 83.9 | 26 | 101.8 | 28 |
| SD | 88.5 | 27 | 119.1 | 31 |
| CHI | 92.8 | 28 | 101.6 | 27 |
| NE | 93.3 | 29 | 62.0 | 12 |
| BAL | 96.1 | 30 | 52.6 | 7 |
| WAS | 119.1 | 31 | 89.5 | 24 |
| NYG | 138.7 | 32 | 137.1 | 32 |
The average AGL dropped from 74.3 in 2014 to 68.9 this season, the first time we did not have an increase since 2010. The correlation for AGL between 2014 and 2015 was 0.40, or essentially the same as it was between 2013 and 2014. These are the highest correlations in our database going back to 2002.
The correlation between 2015 AGL and 2015 team DVOA was -0.23, and the correlation between 2015 AGL and 2015 regular-season wins was -0.32. Both numbers are essentially the average of what we have observed over the years.
Cincinnati having the No. 1 "healthiest" season is not going to excite many Bengals fans, because they really only had the healthiest season for 13 weeks. This was shaping up to be the year for Marvin Lewis and the Bengals to end the playoff drought. The team was playing its best since the 1988 Super Bowl team, Andy Dalton and Tyler Eifert were having career seasons, but it's the loss of those two players to injury in Week 14 against Pittsburgh that will always stand out here. The Bengals could still think about the No. 1 seed in the AFC at that point before a harmless-looking play ended Dalton's season. Eifert eventually returned for the end of the season, but he sure could have come in handy in Denver when the Bengals lost in overtime, missing out on a playoff bye again. AJ McCarron did what he could in relief of Dalton, who never missed a start (81 in a row including playoffs) before this season's thumb injury.
In many ways the Bengals were similar to the 2014 Broncos: both teams led the league in AGL but had the quarterback, the most important player on most teams, get hurt late in the season. Peyton Manning tore his quad in Week 15 of the 2014 season and was never the same quarterback. A year earlier, the Broncos did not have Von Miller and Chris Harris available in the Super Bowl due to injuries. This season, Denver finished 10th in AGL, but had its key players healthy in time for the playoffs to fuel a title run. The other Super Bowl team, Carolina, also had another healthy year, joining the Eagles as the only teams to rank in the top six in AGL in the last two seasons. Say what you want about Chip Kelly, but his AGL rankings of first (2013), fifth (2014) and sixth (2015) in Philadelphia look great for his foray into sports science. He now has to turn around a San Francisco squad that has been beat up the last two years.
Some teams just have lingering injury issues for whatever reason. For example, the Giants have ranked 22nd or worse in AGL in six straight seasons now. Indianapolis was finally able to end its ugly streak with an average season in 2015. The Colts ranked 24th or worse in AGL in every season from 2006 to 2014. Of course, a lacerated kidney for Andrew Luck was the worst kind of injury that could happen to a quarterback-dependent team like the Colts. Some observers believe Luck was actually injured as far back as September.
Obviously, AGL is about totals rather than measuring the quality of the specific players injured. Dallas had a pretty good AGL count this year, but it was really a lost year for the Cowboys, and we could have called it after Week 2. By then, the Cowboys basically lost Tony Romo (Week 2 collarbone injury), Dez Bryant (Week 1 foot injury) and Orlando Scandrick (preseason torn ACL) for the year. Mostly everyone else stayed healthy, but you would trade a lot of those parts to keep your franchise quarterback, dominant wideout and best defensive back healthy all year. This is why the correlation between AGL and team performance is not quite as strong as you might expect, because teams can still get by as long as certain players are not accumulating the AGL.
The big-name injuries are why many thought Baltimore and New England would finish with the worst AGL numbers in 2015. No team had a bigger increase in AGL from 2014 than Baltimore. By the end of the season, Baltimore was without its starting quarterback, top two running backs, top two wide receivers, top two tight ends, center, and left tackle. On defense, Terrell Suggs went down in Week 1. That is a killer season, but at least most of those players were able to get in a handful of games before shutting it down until next year.
[ad placeholder 3]
Similarly with New England, not many of its best players went to injured reserve early. The constant piling up of injuries midway through the season is what created the perception of a league-leading wounded team. Dion Lewis was lost for the season in Week 9. A week later, Julian Edelman was lost until the playoffs. Danny Amendola failed to escape the Buffalo game intact. Then Rob Gronkowski suffered a major scare in Denver in Week 12. Before you knew it, LeGarrette Blount was done for the year. That was a rough patch that led to four losses in six games, but most of those players were back for the playoffs.
New England and Washington were the only teams ranked 20th or worse in AGL to make the playoffs. Jay Gruden is also quickly entering into Bill Belichick territory when it comes to gamesmanship with the injury reports, which we'll look at more in a few paragraphs. Washington may surprise a lot of people at No. 31, but there are injuries here many don't remember, such as new arrival Junior Galette going down before the season even started. The secondary was a mess with safety Duke Ihenacho going down in Week 1, and games missed by starters such as DeAngelo Hall and Chris Culliver. Guard Shawn Lauvao missed 13 games, DeSean Jackson missed six games and the tight end depth was severely tested for an offense that likes to use two tight ends, with Niles Paul, Logan Paulsen and Derek Carrier all finishing on injured reserve.
Six of the top 10 teams in AGL made the playoffs, but Atlanta was not one of them. The Falcons had the third-largest year-to-year decline (minus-65.0) in AGL since 2002, which only adds to the disappointment of finishing 8-8 after a 5-0 start. This team managed to knock off Carolina, but lost to the Blaine Gabbert-led 49ers and blew a 14-point lead at home against the Colts. Julio Jones never missed a game, though he did appear to be slowed by injuries in the weeks he was limited in practice. This team should have achieved more with that much help from the schedule and health.
Injury Reporting Tactics
One of the most interesting things you can do with this data is to see which teams are trying to play some mind games with their opponents by listing a player with a designation that may not be very indicative of his true probability of playing. You have to be gullible to take any team's injury report at face value as the gospel truth, but some teams are known to bend the rules more than others.
The NFL's actual guidelines are for "probable" players to have approximately a 75 percent chance of playing, while "questionable" is 50 percent and "doubtful" is 25 percent. There is a league-wide disregard of those figures.
[ad placeholder 4]
Probable basically means the player is going to play. For the second year in a row, just a shade under 95 percent of probable players played that week. Only the Ravens (86.2 percent), Browns (86.3 percent) and Patriots (88.6 percent) were under 90 percent. This is essentially teams covering the bases on reporting the minor injuries that will rarely ever prevent a player from playing.
This is kind of weird, but the Texans and Jets have finished first and second, respectively, in total probable players in each of the last three seasons. Bill O'Brien took the job from Gary Kubiak in 2014, but he sure loves listing his team as probable. The Texans led the league in 2014 with 170 probable designations, and did so again in 2015 with 169 (and 167 of them played). Even though the Jets switched coaches from Rex Ryan to Todd Bowles, a first-time coach, New York still finished second in consecutive years at listing probable players. Meanwhile, we see a handful of the same teams from last year that used probable the least.
Fewest players listed as probable, 2014-15 pic.twitter.com/d7aUEOjJpH
— Scott Kacsmar (@FO_ScottKacsmar) March 1, 2016
Why spill the beans that a player is going to play when you can keep it a guessing game with questionable to play? 2014 was trending in the right direction with 55.7 percent of questionable players playing, but it went back up to 62.4 percent this season with three teams (Detroit, Miami and Cleveland) all topping 80 percent. For the second year in a row, the Steelers played the lowest percentage of questionable players (30.8 percent) while barely using that designation again. New Atlanta coach Dan Quinn's staff only used it five times. The following compares the questionable player results of 2015 to 2014.
| Rk | Team | 2015 Questionable | Played | Played% | 2014 Questionable | Played | Pct. | Rk |
| 1 | PIT | 13 | 4 | 30.8% | 12 | 1 | 8.3% | 1 |
| 2 | TEN | 24 | 10 | 41.7% | 48 | 23 | 47.9% | 11 |
| 3 | JAC | 33 | 14 | 42.4% | 15 | 7 | 46.7% | 9 |
| 4 | KC | 25 | 11 | 44.0% | 26 | 16 | 61.5% | 21 |
| 5 | STL | 36 | 16 | 44.4% | 31 | 14 | 45.2% | 8 |
| 6 | OAK | 15 | 7 | 46.7% | 42 | 23 | 54.8% | 16 |
| 7 | NO | 17 | 8 | 47.1% | 29 | 18 | 62.1% | 24 |
| 8 | PHI | 28 | 14 | 50.0% | 15 | 3 | 20.0% | 2 |
| 9 | SEA | 32 | 16 | 50.0% | 22 | 9 | 40.9% | 5 |
| 10 | CAR | 10 | 5 | 50.0% | 28 | 12 | 42.9% | 6 |
| 11 | GB | 57 | 29 | 50.9% | 21 | 13 | 61.9% | 23 |
| 12 | IND | 34 | 18 | 52.9% | 26 | 15 | 57.7% | 19 |
| 13 | NYG | 33 | 19 | 57.6% | 23 | 15 | 65.2% | 26 |
| 14 | NE | 71 | 41 | 57.7% | 104 | 69 | 66.3% | 28 |
| 15 | SF | 27 | 16 | 59.3% | 60 | 22 | 36.7% | 4 |
| 16 | TB | 45 | 27 | 60.0% | 74 | 36 | 48.6% | 12 |
| Rk | Team | 2015 Questionable | Played | Played% | 2014 Questionable | Played | Pct. | Rk |
| 17 | ATL | 5 | 3 | 60.0% | 26 | 17 | 65.4% | 27 |
| 18 | WAS | 61 | 37 | 60.7% | 62 | 49 | 79.0% | 32 |
| 19 | CIN | 18 | 11 | 61.1% | 21 | 13 | 61.9% | 22 |
| 20 | DAL | 20 | 13 | 65.0% | 25 | 16 | 64.0% | 25 |
| 21 | HOU | 24 | 16 | 66.7% | 27 | 12 | 44.4% | 7 |
| 22 | SD | 57 | 38 | 66.7% | 43 | 24 | 55.8% | 17 |
| 23 | CHI | 97 | 65 | 67.0% | 33 | 18 | 54.5% | 15 |
| 24 | DEN | 41 | 29 | 70.7% | 34 | 12 | 35.3% | 3 |
| 25 | BAL | 22 | 16 | 72.7% | 21 | 12 | 57.1% | 18 |
| 26 | BUF | 15 | 11 | 73.3% | 22 | 12 | 54.5% | 14 |
| 27 | ARI | 47 | 35 | 74.5% | 34 | 23 | 67.6% | 29 |
| 28 | NYJ | 26 | 20 | 76.9% | 24 | 12 | 50.0% | 13 |
| 29 | MIN | 19 | 15 | 78.9% | 30 | 18 | 60.0% | 20 |
| 30 | DET | 56 | 45 | 80.4% | 19 | 9 | 47.4% | 10 |
| 31 | MIA | 36 | 29 | 80.6% | 33 | 24 | 72.7% | 31 |
| 32 | CLE | 59 | 50 | 84.7% | 42 | 30 | 71.4% | 30 |
| - | NFL | 1103 | 688 | 62.4% | 1072 | 597 | 55.7% | - |
Chicago was the big questionable team this season in John Fox's debut season, and over two-thirds of those players were still active for the Bears. Bill Belichick (Patriots) and Jay Gruden (Redskins) were a little more restrained in their use of questionable this season, but both still ranked in the top three for total times used.
Some of Gruden's tactics can be described as questionable at this point too. Washington linebacker Jackson Jeffcoat was placed on injured reserve in early November despite not appearing on the injury report since a thigh injury had him on there in Weeks 1-2. There was no indication Jeffcoat was injured at the time his season was ended. Gruden nonchalantly explained to media that Jeffcoat "had back spasms or something like that, so he had an injury settlement." Or something you just made up? Players on a team's injured reserve list can be released with an injury settlement, but there is no proof that has ever happened between Washington and Jeffcoat, who is still listed as active on an updated roster page on the team's official site. Apparently "injury settlement" has a different meaning to Gruden than the rest of the league.
One thing teams used to be able to agree on was that a player listed as "Out" on Friday meant that he was out on Sunday. However, there were some situations in 2015 where a player listed as out or doubtful ended up playing that week. I never encountered this in the previous two years of overseeing this project, but apparently what happens is the team upgrades the player's status on Saturday even though it is not recognized in the injury reports online. For example, Cincinnati's Michael Johnson was listed as being doubtful for a Week 11 game in Arizona, but only a writer for the official Bengals site noted on Saturday -- in an article posted on that Friday -- that Johnson was upgraded to questionable. Johnson started and played 51-of-60 snaps, yet the injury report on his own team's website still lists his status as "Doubtful" for Week 11, as does the NFL's official website.
The same "Saturday upgrade" happened for Oakland's T.J. Carrie in Week 5 against Denver, and that was after he was ruled out on Friday. Trevor Riley (Jets) and Mark Herzlich (Giants) were the other "out players" to play this season, which can raise the eyebrows of even those covering the team.
Jets OLB Trevor Reilly (finger infection) was listed as out two days ago. Upgraded to doubtful yesterday. ACTIVE today. #nyj
— Manish Mehta (@MMehtaNYDN) October 4, 2015
LB Mark Herzlich, who was "out" earlier in the week, then upgraded to "questionable" on Saturday, is also active.
— Ralph Vacchiano (@RVacchianoNYDN) November 29, 2015
Seattle's Bruce Irvin appears to be the only player in 2015 who was listed as doubtful, did not receive a magical Saturday upgrade, and still played that week against the Steelers in Week 12. Oddly enough, of the 216 doubtful players in 2014, only Seattle's Alvin Bailey was active that week, but he did not play.
Shady Saturday reports aside, the teams are actually not taking advantage of their right to list doubtful players as only having 25 percent likelihood of playing. Instead, doubtful has come to mean zero percent outside of a few special circumstances. So we have found one area where NFL teams do not choose to be deceitful in reporting injuries.
Later this week we will look at the AGL breakdown by units. The musical-chairs game remains strong with the San Diego offensive line.
Comments
24 comments, Last at 04 Mar 2016, 10:46am
#1 by RickD // Mar 01, 2016 - 12:53pm
In case it's not obvious, we should note that the various injury list probabilities are likely not to match exactly the "real probabilities" of a player missing a game. For example, let's say a player has a 60% chance of playing. With "questionable" at 50% and "probable" at 75%, the appropriate thing for a team to do would be to list him as "questionable". With "probable" the upper bound probably goes even higher than the halfway point of 87.5%, since there's likely to be a lot more backlash against not playing somebody who was 90% probable than there would be against listing him erroneously.
For the "questionable - did they play?" list, we'd hope to see teams playing these guys between 37.5% and 62.5% of the time, and most teams fall in that range, with the real percentages skewing a little higher. But teams playing their "questionable" players 75% of the time or more might want to think hard about what they're doing.
#19 by Scott Kacsmar // Mar 02, 2016 - 1:10pm
That was added a few years ago, probably in an attempt to make these reports more legitimate. We track it, but I'm not sure how useful it really would be since there are three reports of it for Wed-Fri. If a guy is limited on Wed. and Thu. but full on Friday, then his status for the week will be Full participation. We only keep track of the final designation. I guess it would be interesting from a fantasy perspective to see if limited practice skill players perform worse than average that week.
#18 by InTheBoilerRoom // Mar 02, 2016 - 10:56am
It seems to me the designations are being treated more like bins. Players reported as "probable" have somewhere between a 75-100% chance of playing, "questionable" a 25-75% chance, and "doubtful" a 0-25% chance. And I feel like that is a reasonable way to report it.
#3 by Tomlin_Is_Infallible // Mar 01, 2016 - 2:17pm
Seems really lacking in value to count JPP or other similar things.
The idea being either FOOTBALL injuries are bad luck, bad training/conditioning, or some portion of both.
Off the field self-inflicted stupidity doesn't seem to have any predictive or explicative power for understanding why a team might underperform unless you maintain some sort of "character" data in the equation. In other words, factor in things like suspensions?
--------------------------------------
The standard is the standard!
#7 by brian30tw // Mar 01, 2016 - 3:09pm
If you get more granular than the team level statistics, are there any players that are consistently active/inactive at rates that differ greatly from the implied rates of the injury designation? One way to game the system and cause confusion for the opponent is to always list, someone, maybe Gronk, as questionable, even if you know he's going to play, but offset that by listing someone else, who you know will be a healthy scratch anyway, as questionable as well. Then, 1 of 2 questionable listings will play, making it look like Billy B is following the rules.
#21 by Noahrk // Mar 03, 2016 - 9:07am
Yes, but that doesn't help competitively because it's no good lying when everyone knows you're lying. What you need to do is what the Pats do, put most of your designations on the questionable bucket and play them 50% of the time. Then opponents do not know what to expect.
------
Who, me?
#11 by FireSnake // Mar 01, 2016 - 4:44pm
"quickly entering into Bill Belichick territory when it comes to gamesmanship with the injury reports"
I might be wrong, but according to the percentages in the table, the Patriots are clearly one of the three to five teams, who are crossing their ts and dotting the is on their reports, and doing absolutely everything in compliance with the - idiotic - league rules. Go criticitze teams like Miami, Arizona or the Jets.
Because you call yourself educated NFL followers:
In case you haven't noticed, the Patriots stopped screwing with the injury designations YEARS AGO. BB told he ignores the reports anyway, and thinks every other coach does. It may have been relevant the year the Pats faced off with the Colts in the playoffs with all 22 starters for both teams carrying "probable", it definitely hasn't been ever since. I am sick of reading about it.
The only thing BB fiddles with injured players is that they may put players on IR who are not severely injured in order not to have to cut them (Spikes and probably Dobson)
#12 by FireSnake // Mar 01, 2016 - 4:54pm
And just in case: Please tell us, where the "probable designation implies 75% chance of playing" (nonsense) originates, because that makes no sense at all.
The NFL does not want the following to happen: Player not on injury report but cannot play due to injury. In order for this to happen, teams simply have to put more players on the report. That doesn't mesh at all with the 75% probability.
#13 by LyleNM // Mar 01, 2016 - 5:19pm
I believe the NFL rule is that if a player is listed as Probable (or not on the injury list) and then does not play, the team is required to explain in writing to the league office the reasons for not playing said player.
#14 by Snoth // Mar 01, 2016 - 7:29pm
On the NE one you're missing Solder and Wendell(gone for the season), Develin, Lafell who played with a really bad foot injury the entire season through the playoffs, Dominique Easley(IR), and every starting offensive lineman(excludes Andrews) was dealing with an injury. So In terms of games missed and how those injuries affected the team "not many of its best players went to injured reserve early" doesn't matter since everyone else on the team was injured at some point late in the season just look at how many important players were questionable going into the Championship game, i mean Patriots fans were worried about the availability of Marcus Cannon!
#20 by Nahoj // Mar 02, 2016 - 2:58pm
Here's some interesting reading and stats I found on this topic (though for the life of me I couldn't locate the 4th part about Doubtful players).
It breaks down which position is most likely to play through different categories (your "probable" QB is #1, playing 80% of the time) and sorts probability by types of injuries (if you are listed as Probable but the reason is a Concussion, you're only 56% likely to play, so you really should be considered Questionable).
If you are Questionable and the reason is Illness or Undisclosed, you're actually Probable, around 70%.
https://www.teamrankings.com/blog/nfl/nfl-injury-analysis-how-often-do-hurt-players-actually-play-1-of-4
https://www.teamrankings.com/blog/nfl/nfl-injuries-how-often-do-probable-players-play-2-of-4
https://www.teamrankings.com/blog/nfl/nfl-injuries-questionable-players-and-leg-injuries-3-of-4
#22 by nat // Mar 03, 2016 - 10:37am
The NFL's actual guidelines are for "probable" players to have approximately a 75 percent chance of playing, while "questionable" is 50 percent and "doubtful" is 25 percent.
I think Scott's 25-50-75% buckets (which also aren't ranges) are from an outdated policy.
The actual 2015 policy says this:
Out(definitely will not play)
Doubtful(at least 75 percent chance will not play)
Questionable(50-50 chance will not play)
Probable(virtual certainty player will be available for normal duty)
It's inconsistent nonsense, like many of the NFL's policies.
"Out" and "Doubtful" give exact ranges. (0% and 1-25% chance of playing)
"Questionable" gives an exact value, but clearly does not mean to. (Exactly 50%?)
"Probable" gives an inexact (virtual certainty) phrase intended to be a range, and further qualifies "play" as "normal duty", which would allow a "probable" designation for a slightly injured bench-warmer, but would forbid it for a definite starter who is being given otherwise limited play-time because of his injury.