website statistics
Even as the season winds down, the team efficiency rankings haven't crystallized as clearly as one might assume with the larger sample size of games.  While the days of the big 10-spot swings are gone, multiple teams moved three to four spots, the most meaningful variety of changes we've seen in roughly a month.

That's not entirely surprising when considering the unusual hierarchy that has taken place this season.  With weaker teams in both the elite and lower classes, the middle class is relatively bloated this season, hence the abundance of teams sitting at 7-7 or 8-6.  Whether or not they remain in the playoff race is largely circumstantial (see: NFC South), but these rankings all year have suggested that the gaps between teams in the middle third are not as wide as they typically are.

This week's write-up will focus on a quartet of teams in this range who hold varying levels of hope in the postseason race.

AFC Fringes

A look at two AFC teams hanging on for dear life.

- Though their odds of snapping the NFL's longest current playoff drought are low, the Buffalo Bills have sat in the top 10 for most of the past month.  The Bills somehow fell one spot in the rankings despite a season-defining victory over Green Bay last Sunday, though their Gross Winning Percentage (GWP) did jump up a point. 

Regardless, it's not hard to believe that a team with the second-ranked defense (behind only Denver) would be a top 10 team, even with a decisively below-average offense.  Buffalo does rank first in Expected Points Added (EPA) with -41.1 cumulative EPA this season, nearly twice as much as second-place Detroit.  That's not nearly as dominant as some of the other top defenses—for reference, Seattle led the league with -93.6 EPA last year, while Carolina was second with -73.5—though the Bills' figures to drop a bit more with two games remaining.

Ironically, Buffalo's pass defense has peaked the past month since losing top corner Leodis McKelvin to a season-ending ankle injury.  McKelvin was probably only Buffalo's top corner in name—Stephon Gilmore is a budding star, while underrated offseason acquisition Corey Graham actually leads the Bills DBs in EPA per game.  But McKelvin was still a quality No. 2 CB and very high end No. 3 CB, yet the Bills have forced negative EPA performances in every game since his injury, with the Green Bay masterpiece serving as their best non-Jets EPA showing of the season:

However, when I analyzed the Bills earlier this season, I noted how very few teams had made the postseason with their kind of offense.  At that point, Buffalo was averaging 3.8 yards per carry and 6.5 net yards per attempt (NY/A).  Nine teams had made the playoffs with that combination since the 2004 rules changes, but most of them either vastly exceeded their win expectancies (2012 Colts) or benefited from exceedingly poor division competition (2010 Seahawks).

Buffalo is still averaging the same number of yards per attempt, but the Kyle Orton-led passing game has dipped to 5.8 NY/A on the season, the seventh-worst mark in the league.  Expanding the (admittedly arbitrary) endpoint to the last 20 years, these are the teams to make the postseason with that combo:

As was the case initially, most of those teams are scattered on the early end of the time point.  The aforementioned 7-9 Seahawks are the only team to make it if we limit the time period to our previous 2004-present scope, while just five have made it since 2000.  Unless you're a team like the 2002 Bucs with one of the greatest defenses in NFL history (-139.7 regular season EPA), playoff success isn't happening.

Still, that doesn't illegitimize what will likely be Buffalo's first winning season since 2004.  We often see strong defensive teams with shaky quarterback play exceed their expected win probabilities, with Arizona and Philadelphia being the most blatant examples from 2014.  But all the metrics portray the Bills as a legitimate eight-win team—AFA's current 0.57 GWP projection would translate to 7.98 wins, while Pro Football Reference's Pythagorean win expectancy pegs Buffalo with 8.4 expected wins.  Considering that the Bills have had less offensive skill position talent than nearly any team in the league this year, that highlights how truly impressive this young defense has been.

- Whereas the Bills have come on strong, the San Diego Chargers continue to dip, falling to 23rd and seeing their GWP dip to a season-low 0.42 this week.  There's no shame in losing games to Denver and New England, but the way those losses unfolded were both surprising and concerning.

After racing out as the early MVP favorite, Philip Rivers experienced a midseason dip before returning to form during a three-game winning streak against Oakland, St. Louis and Baltimore.  But over the past two weeks, Rivers has regressed in leading the offense to 17 points.  Over that span, Rivers ranks last in adjusted yards per attempt (AY/A) among quarterbacks with at least 40 pass attempts in the past two weeks.

It's a tiny sample size, of course, and it's unfair to skew the numbers against Rivers when he has faced two of the best pass defenses in the league.  But even if you want to limit the scope to Rivers' three-game winning streak, a randomly plucked sample that should make him look good, he ranked ninth in AY/A and 11th in plan yards per attempt.  That's perhaps not what we'd expect knowing that the Chargers won all those games (as a side tangent, despite losing all his starts in that span, Zach Mettenberger actually ranked ahead of Rivers in AY/A, perhaps providing a shred of hope for disillusioned Titans fans).  Moving back to Rivers, this is looking like another typical career-average season, despite the hot start:

Apart from the hellfire that was the twilight Norv Turner years, an average Rivers season has been good enough to get San Diego to the playoffs, even with a below-average defense.  Rivers has made the playoffs five times, but apart from his first two years as a starter, he's been saddled with a defense that has conceded a positive EPA to opponents:

The 2014 rendition is giving up 4.66 EPA/G, and although that's much better than the 7.46 EPA/G mark last season's team conceded, it would still rank as San Diego's third-worst defensive season since 1999.  Most of those problems stem from a poor month from Weeks 6-9, when the Chargers gave up 6.2 yards per play, fourth-worst in the league. 

Still, there aren't a whole lot of areas where the Chargers defense stands out in any meaningful way.  Some metrics have implied that the Chargers defense has improved over the past month, but the larger sample size portrays a bleak picture.  San Diego possesses a roughly league-average pass defense and a run defense success rate that ranks 28th.  The Chargers don't generate many big plays either, ranking 28th in interception rate, 29th in total takeaways per game and 29th in sack percentage.  San Diego also ranks 28th in third down defense, though they're closer to average in red zone defense, sitting 14th overall.

To be frank, it's a fairly boring unit that lost its best cornerback in rookie Jason Verrett and has seen exciting outside linebacker Melvin Ingram miss most of the year with a hip injury.  Even its top players, Eric Weddle and Corey Liuget, bear a closer resemblance to the steady jack-of-all-trade types rather than a big-play generator. 

There's nothing wrong with being fundamentally sound, of course, but San Diego's defense suggests the need for an infusion of playmaking.  Rivers has been forced to pick up the slack, and while he succeeded for most of the season, his feeble performances against elite competition the past two weeks illustrate the need for a better supporting cast in San Diego (or Los Angeles?) next season.

On Shaky Ground

Two division leaders are big underdogs to stay in first.

- This column has already spilled plenty of ink on the Arizona Cardinals, so I won't dig as deep into the general trends and long-term outlook as I typically do with other teams.  Arizona has already clinched a playoff berth, so for models like GWP and DVOA that place them well outside the top 12, Cardinals fans will have the #WINZ retort to throw in the face of analytics.

Still, the comeuppance may arrive in January, when Ryan Lindley and Kerwynn Williams will likely comprise the most anonymous quarterback-running back combination in playoff history.  In seven career games, Lindley has compiled an astounding -70.7 EPA and 34.5 percent success rate.  For reference, that EPA is more than double what the worst 2013 quarterback, Geno Smith, compiled in over twice as many starts (-31.9 over 16 starts).  That success rate would also have been the worst in the league last season, right below the exiled Josh Freeman and Brandon Weeden.

We shouldn't blame all of that on Lindley, for six of those games came in his rookie season, when he was unprepared sixth-round pick on a wretched 5-11 Cardinals team.  Bruce Arians had not yet arrived in the desert, as he was still working his magic with another then-rookie, Andrew Luck, in place of a leukemia-striken Chuck Pagano. 

Still, it's hard to conceive of a playoff team with a less prepared quarterback in recent history.  The Joe Webb game immediately comes to mind, but that start came about when normal starter Christian Ponder was a gametime scratch.  I compiled some recent quarterbacks with similar profiles to Lindley, with the parameters being: a QB who played from 2000-2014, a QB in his first three years in the league, at least three starts, no more than one win and a AY/A of 4.0 or less (Lindley sits at 3.0 after his first brief appearance).  This was the horrorshow I got:

It's probably not a good sign that Lindley himself is already on that list.  I took away the wins restriction and found that the most obvious comp was probably 2005 Kyle Orton, who "led" the 11-5 Bears to the playoffs despite posting the lowest AY/A and second-lowest NY/A rates in the league.  Lindley will only get two regular-season starts at most, but Bears fans had to endure 15 games of the Purdue rookie, and only made the playoffs behind the league's best defense, a component the Cards do not possess.

The narrative surrounding Williams is less discouraging, though running backs themselves are fungible commodities.  Over 34 carries in the last two weeks, Williams has compiled -2.5 EPA on a 38.5 percent success rate.  That latter mark would rank 35th out of 40 qualified running backs, but it's a small sample size.  Moreover, the Cards have had some O-line turnover due to injuries, one of those two games came against St. Louis' top-ranked run defense (based on SR%). 

Arizona probably has less margin for error than any other playoff team, even if they end up playing the NFC South champ in the Wild Card round, the most likely possibility.  AFA still gives the Cards a 62 percent chance of reaching the divisional round, but that likely overestimates their chances of beating Seattle and earning a first-round bye this week.  The bot can't account for the Lindley factor, but there is a strong possibility that the Cardinals will end the season as AFA's lowest-ranked playoff team.

- The Cincinnati Bengals moved up three spots this week's rankings as a result of their 30-0 thrashing of the Browns (who remain the second-highest ranked AFC North team).  The Bengals still rank a middling 14th, however, and a closer look suggests that they belong behind the Steelers and Ravens, as most analytics models portray.

AFA's playoff probabilities give the Bengals a solid 80 percent chance to reach the postseason, though that still ranks behind both Baltimore and Pittsburgh.  However, Cincy figures to be the underdog in their remaining two games, a Monday night home contest against Denver and a visit to Heinz Field in Week 17. 

A final record of 9-6-1, which would equate to a 0.60 winning percentage, would still probably exceed the numbers projections.  GWP pegs the Bengals as a 0.54 level team, while PFR has them at with 7.6 expected wins at the moment, nearly a win and a half less than they currently have.

This isn't to pin anything on Andy Dalton in particular.  The Red Rifle has almost exactly duplicated his EPA/P and success rate from 2013, with a demonstrable increase in Win Probability Added (WPA) suggesting that he has done better in high leverage situations this year.  Considering that Dalton's playoff failures have earned him a "choker" reputation, that's not insignificant.  Moreover, with Cincy making the move to rookie Jeremy Hill as the primary running back, the Bengals should improve their roughly league-average ground game by finally leaning towards the more efficient back:

Hill might be the type of boost the Bengals need to vault themselves into the same tier as AFC contenders like the Steelers, Ravens and Colts.  Many might perceive Cincy as on that level already, but in reality, the Bengals are essentially a league-average team in everything besides run defense, where their 51.8 percent success rate is the second-worst in the league.  In many ways, this isn't particularly different from the San Diego unit we dissected earlier, in that the Bengals lack the playmaking ability (22nd in takeaways per game, 31st in sack percentage) to compensate for relatively mediocre per-play defense.

Indeed, Cincy bears a strong resemblance to some recent teams with stellar records but underlying mediocre success rates.  The Bengals average 5.5 yards per offensive play and concede 5.3 yards per play on defense, a profile that roughly aligns them with these recent squads (arbitrary endpoint= 2000):

Excluding the 2014 squads, six of the previous 13 squads made the postseason.  However, most of those squads ended up exceeding their Pythagorean win expectancies, with the 2013 Patriots (12 wins vs. 10.5 expected), 2003 Titans (12 vs. 10.7) and 2004 Seahawks (9 vs. 7.9) standing out as the most prominent examples.  That '04 Seattle squad, with a young Matt Hasselbeck-Shaun Alexander combo on offense and a middling defense, bears the strongest resemblance to this year's Bengals squad.

Given the likelihood that they hit the road to face a slightly superior opponent in the wild card round, there is a very solid chance the Bengals end up extending their playoff winless streak to seven.  It's impressive for a team in today's NFL to make the postseason five times in six years, as Cincy is on the verge of doing, but for those frustrated by their inability to get off the B-list contender list, this year's squad does not appear to represent the answer to that trend.

Pooch Punts

Looking at teams facing the best and worst of times.

- This week's rankings have established a clear line of division among the top six teams.  The seventh-place Ravens sit with a 0.59 GWP, while the top six are all clustered at 0.62 GWP or better.  The Broncos are actually still way ahead of the pack with 0.76 GWP, a byproduct of their status as a top-three offense and defense.  Only one other team, the Seahawks, even has two top-10 units on offense and defense.

The obvious outlier there is the Dolphins, who still curiously sit at second.  Perhaps this is something I'll look into more after the season, but considering that passing offense is the most important variable, it's shocking that the Fins remain this high despite a 5.7 NY/A, the sixth-worst mark in the league.  Miami does possess an excellent rushing attack and pass defense, but that would seem more in line with placing them in the borderline top 10, as DVOA and and PFR's Simple Rating System (SRS) both do.

The Patriots are one team that many might feel deserve better, even though they've been a top six fixture for a few weeks.  The model isn't as impressed with New England's defense as the mainstream public, rating the Pats as a roughly league-average pass defense and a below-average run defense.  EPA/P doesn't really jibe with that, though, ranking New England 16th in run defense and eighth in pass defense.

AFA has some wackier rankings than some other analytics sites, especially early in the season, but like most models, they see the Broncos, Patriots, Seahawks and Packers as part of the NFL aristocracy.  The inclusion of Miami and Indianapolis in that grouping is the unusual factor, as few would see the Patriots as the fourth-best team in the AFC, as well as the second-best in the division they have already clinched.  But just as the model is not an end-all-be-all ranking of the league, we also shouldn't be too quick to dismiss things that look out of whack.  Hopefully I'll have time to take a more in-depth look at the model soon.

- On the opposite end of the spectrum, AFA has established the Oakland Raiders as the clear-cut worst team in the league.  While it's hard to deny that Oakland belongs near the basement, AFA pegs the Raiders with a 0.25 GWP, five percent worse than the next worst team in the rankings. 

Interestingly, no other model seems to see Oakland as that poor of a team.  DVOA has the Raiders 25th, SRS has them 28th and even FiveThirtyEight's Elo ratings, which bear a closer resemblance to the mainstream perception because of their incorporation of past seasons' success, puts Oakland 29th.  Even AFA's own success rate graphs suggest that Oakland possesses a roughly average defense, hardly the characteristic of the league's worst team:

This isn't to suggest that the Raiders are anything but one of the three to five worst teams.  Oakland is not above average in any stat category, and no team has a larger gap between yards per play and yards allowed per play.  An already thin squad has gotten worse due to injuries, with the secondary and offensive skill position corps particularly decimated this season.

Still, in Khalil Mack and Derek Carr, the Raiders appear to have hit on their first two picks of the previous draft (though the jury is obviously still very much in session on Carr).  With a full complement of draft picks in each of the next two seasons, the Raiders can finally begin to correct for their decade-long sins in personnel mismanagement.

Here are the updated team efficiency rankings after 15 weeks.  As always, observations, questions and snide remarks are welcome in the comments section.

1 DEN 1 0.76 0.51 3 1
2 MIA 2 0.65 0.52 11 6
3 SEA 4 0.65 0.50 9 4
4 GB 3 0.64 0.50 1 19
5 IND 5 0.63 0.50 4 18
6 NE 6 0.62 0.53 7 15
7 BAL 7 0.59 0.49 8 10
8 KC 11 0.58 0.51 19 8
9 NO 13 0.57 0.50 5 29
10 BUF 9 0.57 0.53 26 2
11 DET 8 0.56 0.48 18 3
12 CLE 10 0.54 0.47 17 7
13 DAL 12 0.54 0.47 6 26
14 CIN 17 0.54 0.50 15 13
15 PIT 16 0.53 0.48 2 30
16 SF 15 0.52 0.51 25 5
17 PHI 14 0.52 0.50 14 16
18 HOU 19 0.51 0.48 13 17
19 CAR 20 0.50 0.50 20 11
20 ARI 18 0.48 0.49 22 12
21 NYG 21 0.46 0.49 21 20
22 STL 23 0.43 0.49 27 9
23 SD 22 0.42 0.52 23 24
24 TEN 26 0.42 0.50 24 21
25 MIN 27 0.41 0.49 28 14
26 WAS 25 0.40 0.47 12 27
27 CHI 24 0.40 0.52 16 31
28 ATL 28 0.36 0.49 10 32
29 NYJ 29 0.33 0.53 29 25
30 JAC 30 0.30 0.53 31 23
31 TB 31 0.30 0.48 30 22
32 OAK 32 0.25 0.53 32 28

ATL 7.0 36 2.4 1.3 7.9 57 2.8 0.41
ARI 6.6 34 1.7 1.5 6.6 64 3.5 0.39
BAL 6.9 42 1.9 1.3 6.4 65 1.7 0.43
BUF 5.8 40 2.2 1.8 5.4 62 3.9 0.51
CHI 6.0 46 3.4 1.7 7.4 58 2.3 0.49
CAR 5.9 43 2.2 2.2 6.2 60 2.2 0.37
CIN 6.6 40 3.2 1.9 6.0 53 2.7 0.39
CLE 6.7 38 3.1 1.6 5.6 54 3.8 0.44
DAL 7.2 44 2.4 2.2 7.0 61 3.0 0.41
DEN 7.5 42 2.1 1.3 5.4 66 2.8 0.52
DET 6.4 39 1.9 1.6 5.8 64 3.6 0.49
GB 7.4 43 1.3 1.6 6.3 51 3.4 0.38
HOU 6.8 39 2.7 2.0 6.4 55 3.2 0.38
IND 7.2 47 2.4 2.4 6.4 57 2.4 0.41
JAC 5.0 35 3.6 1.2 6.7 60 1.1 0.29
KC 6.0 42 1.4 2.3 5.3 54 0.8 0.34
MIA 5.7 51 2.2 2.3 5.7 59 2.7 0.30
MIN 5.5 42 3.4 0.9 6.2 56 2.7 0.42
NE 6.8 45 1.5 1.1 6.2 56 2.9 0.53
NO 7.1 46 2.1 1.5 7.0 57 2.3 0.34
NYG 6.3 39 2.5 1.9 6.7 60 3.3 0.35
NYJ 4.8 45 3.2 2.5 6.8 61 1.1 0.49
OAK 5.1 33 2.7 1.9 6.8 57 1.6 0.46
PHI 6.6 41 3.5 2.2 6.4 61 1.9 0.44
PIT 7.4 43 1.5 1.2 7.3 61 2.0 0.44
SD 6.8 32 2.7 1.3 6.5 54 1.5 0.50
SF 5.7 41 2.3 1.7 5.9 63 3.7 0.45
SEA 6.2 49 1.5 2.3 5.7 62 2.3 0.53
STL 6.0 39 2.9 2.5 6.2 66 2.5 0.59
TB 5.9 34 3.7 2.7 6.8 61 2.3 0.50
TEN 6.3 38 3.6 2.0 6.5 57 2.5 0.51
WAS 6.6 42 3.1 2.6 7.1 64 1.1 0.59
Avg 6.4 41 2.5 1.8 6.4 59 2.5 0.44