2015 Final Team Strength Stats

The 2015 NFL season ended with an upset that surprised almost everyone. The Broncos, whose mighty defense carried the mostly inept offense to victory over the

Led by Von Miller and a historically great defense, the 14-4 Denver Broncos upset the 17-1 Carolina Panthers (favored by 5.5 points).1

The Panthers were superior by every traditional and advanced measure of team strength. Sure, the Broncos had a defense for the ages, but the Panthers had a stout defense of their own, coupled with the league’s top scoring offense and MVP at quarterback. The offensive deficiency of the Broncos looked like an insurmountable limitation. On paper, this was a slaughterfest waiting to happen. As they say, that’s why you play the game.2

Even though we know how the season played out, I still find it enjoyable to look back past the “any given Sunday” variability of the season and try to see which teams truly were the strongest overall. I’ll start with some very basic measures and work toward slightly more complicated ones. Without further ado, here are your 2015 final team strength stats.

Win% and Weighted Win%

Win percentage is pretty self-explanatory: Win% = (wins + .5*ties) / games played. Weighted win percentage is a pretty simple metric that gives teams extra credit for advancing in the playoffs. Using the champion Broncos as an example: Denver went 15-4 in the 19 games they played this year, giving them a .789 record. The 20 win bonus for their Super Bowl victory gives them 35.0 weighted wins for the season, resulting in a .897 weighted win percentage.3

TmWLW%WtWWtW%
DEN1540.78935.00.897
CAR1720.89527.00.931
ARI1440.77819.00.826
NE1350.72218.00.783
KC1260.66714.50.707
GNB1170.61113.50.659
PIT1170.61113.50.659
SEA1170.61113.50.659
CIN1250.70613.00.722
MIN1160.64712.00.667
WAS980.52910.00.556
HOU980.52910.00.556
NYJ1060.62510.00.625
IND880.5008.00.500
ATL880.5008.00.500
BUF880.5008.00.500
DET790.4387.00.438
OAK790.4387.00.438
PHI790.4387.00.438
STL790.4387.00.438
NOR790.4387.00.438
CHI6100.3756.00.375
TB6100.3756.00.375
MIA6100.3756.00.375
NYG6100.3756.00.375
JAX5110.3135.00.313
SF5110.3135.00.313
BAL5110.3135.00.313
SD4120.2504.00.250
DAL4120.2504.00.250
CLE3130.1883.00.188
TEN3130.1883.00.188

Carolina’s lower number of actual losses keeps them ahead of the Broncos in weighted win percentage, but Denver takes the top spot in total weighted wins by virtue of their championship. This metric will always have champions on top at the end of the season (even with 2007 type upsets).

The Seahawks won fewer regular season games than the Bengals, but they end the season with more weighted wins because they advanced further in the playoffs.

The Jets finished the season with a higher weighted win percentage than both Houston and Washington, despite bonuses for the latter two for making the tournament.

SRS score and HFA-adjusted SRS score

Simple Rating System (SRS) scores are Doug Drinen’s iterative process for rating teams by adjusting for margin of victory and strength of schedule. The average team will have a score of 0, below average teams will have negative scores, and above average teams will have positive scores. The end result is something akin to a points spread. A team with a score of 10 can be expected to beat a team with a score of 4 by 6 points (on a neutral field). Pro Football Reference publishes regular season SRS scores, but I like to keep them going through the playoffs. I also like to weight them for home field advantage (HFA), when applicable.

The following table is sorted by HFA-adjusted SRS score. Read it thus: the Seattle Seahawks had an average margin of victory of 7.78 points against a schedule of teams 2.22 points better than average, giving them an SRS score of 10.00. After adjusting for HFA, they have a margin of victory of 8.07 against a 2.26-level schedule, resulting in an adjusted SRS score of 10.32.4

TeamMOVSOSSRSHFAMOVHFASOSHFASRS
SEA7.782.2210.008.072.2610.32
KC7.831.659.498.271.669.93
CIN8.121.769.887.961.899.85
ARI8.221.469.698.221.559.78
CAR11.53-2.169.3711.25-2.199.06
PIT5.502.528.025.792.588.37
DEN4.322.967.274.043.067.10
NE8.61-1.457.168.61-1.527.09
GNB3.112.385.493.402.405.80
MIN3.652.085.733.492.195.68
NYJ4.56-3.331.244.40-3.401.00
OAK-2.502.29-0.21-2.502.37-0.13
BUF1.25-1.50-0.251.09-1.56-0.47
DET-2.632.11-0.51-2.792.21-0.58
STL-3.132.13-0.99-3.132.23-0.90
CHI-3.882.32-1.55-3.882.39-1.48
HOU-0.24-1.78-2.01-0.39-1.81-2.19
BAL-4.562.12-2.44-4.562.25-2.32
SD-4.882.26-2.61-4.882.36-2.51
WAS-0.47-2.19-2.66-0.62-2.26-2.89
NYG-1.38-2.41-3.78-1.38-2.53-3.90
ATL-0.38-3.58-3.96-0.38-3.70-4.08
PHI-3.31-1.65-4.96-3.31-1.77-5.08
SF-9.312.97-6.34-9.313.08-6.24
CLE-9.632.98-6.65-9.633.08-6.54
IND-4.69-2.14-6.82-4.69-2.22-6.91
NOR-4.25-2.61-6.86-4.25-2.73-6.98
MIA-4.94-2.17-7.11-4.78-2.27-7.05
DAL-6.19-0.98-7.16-6.19-1.06-7.25
JAX-4.50-3.29-7.79-4.34-3.39-7.72
TB-4.69-3.21-7.90-4.69-3.32-8.01
TEN-7.75-3.03-10.78-7.75-3.11-10.86

If you sort by “MOV” or “HFAMOV” you’ll see Carolina outpaced the rest of the league in average margin of victory (that’ll happen when you rank first in points scored per game and sixth in points allowed per game). However, they had one of the league’s weaker schedules, which dropped them to fifth in SRS score.

The Seahawks, by virtue of solid performance against a tough schedule, come out on top with a score of 10. Ten is a pretty low score for the top team in the league, but 2015 was a weird year. There were only twelve teams who outscored their opponents, and there were only eleven teams with positive SRS scores. The best teams weren’t historically great, and the worst teams weren’t historically bad.

Not much changes when we adjust for HFA. The bad teams are still bad, and the good teams are still good, with insignificant movements here and there. One thing to note is the strength of schedule of the Broncos, Steelers, and Packers. All three teams had great seasons despite playing difficult schedules.

On the other hand, Washington sneaked into the postseason with a mediocre record in a bad division. They played a subpar slate of opponents and still managed to cede more points than they scored. Anything can happen, of course, but don’t be surprised if they’re playing golf next January.

The Panthers may still make the playoffs, but they get this season’s reversion alert. Rarely, if ever, has a team won 14 or more games and actually been as good as its record. Carolina had some stretches where they played like a 15-1 team, but for the most part they played like an 11-5 team that flipped heads a lot. Having won their division, they’ll have a tougher schedule next year and will almost certainly win fewer games.5

Point Differential, Point Ratio, Pythagorean, Pythagenpat, SPat

This section comprises a handful of ways of looking at points scored and points allowed. Point differential is simply points scored minus points allowed, while point ratio is points scored over points allowed. Pythagorean and Pythagenpat expectation are two models for determining a team’s true winning percentage. SPat (Strength-of-schedule-adjusted Pythgenpat) is my metric for adjusting for opponent when determining Pythagenpat expectation.6

TmGmPFPAPDPRPythPPatSPat
CAR195903712191.590.7500.8080.794
SEA184573171401.440.7040.7460.755
KC184553141411.450.7070.7480.755
CIN174352971381.460.7120.7520.749
NE185103551551.440.7020.7490.746
ARI185303821481.390.6850.7320.728
DEN19422340821.240.6250.6530.695
PIT18457358991.280.6410.6740.694
MIN17374312621.200.6060.6270.640
GNB18423367561.150.5830.6030.625
NYJ16387314731.230.6210.6460.589
BUF16379359201.060.5320.5390.546
HOU17339343-40.990.4930.4920.495
WAS17406414-80.980.4880.4850.457
OAK16359399-400.900.4380.4230.451
DET16358400-420.900.4350.4200.435
NYG16420442-220.950.4700.4620.432
ATL16339345-60.980.4900.4870.430
STL16280330-500.850.4040.3880.402
CHI16335397-620.840.4010.3790.402
PHI16377430-530.880.4230.4040.401
BAL16328401-730.820.3830.3580.373
SD16320398-780.800.3740.3480.370
NOR16408476-680.860.4100.3850.361
IND16333408-750.820.3820.3560.351
JAX16376448-720.840.3980.3720.330
MIA16310389-790.800.3690.3430.327
TB16342417-750.820.3850.3590.318
DAL16275374-990.740.3250.2960.316
CLE16278432-1540.640.2600.2190.252
TEN16299423-1240.710.3050.2680.240
SF16238387-1490.610.2400.2050.236

The Panthers led the league in point differential, as is often the case for teams leading the league in scoring. The top three teams in point differential also happen to be the three teams that scored more than 500 points this year. Obviously, this undervalues teams with strong defenses, which is why using point ratios can be more instructive.

When you sort by point ratio, you’ll see that the Bengals (ranked sixth in point differential) jump to second place. This is because, while their offense wasn’t high powered, they managed to score points at a much higher rate than they allowed them (1.46 times, to be exact).

Using point ratios, you also see that the 49ers slip past the Browns for the worst rate in the NFL. Although Cleveland’s point differential was five points worse than San Francisco’s, the Browns managed to score 0.64 points for every point they allowed. The Niners, for their part, could only manage 0.61 points per point allowed.

Pythagorean expectation spits out the same ranking as point ratio but translates it into an expected win percentage. Its most useful purpose is to tell us just how much a team over- or under-performed its intrinsic strength. For example, the Broncos won 15 of their 19 games (.789 win%), but they had a Pythagorean expectation of a 12 win team (.625). This means they won about three more games than they “should have” and may be primed to backslide.

The Chargers, on the other hand, win 4 games (.250 win%) but had a Pythagorean expectation of a 6 win team (.374). This means they won about two fewer games than their talent would suggest. To put it another way, you could say the Broncos had great luck and the Chargers had bad luck.

One of the drawbacks to the Pythagorean formula is that it can overvalue defensive teams. For this reason, I prefer to use the Pythagenpat formula, which is adjusted for the scoring environment in each team’s games. Using this formula, there is only one notable difference from the Pythagorean method: the Patriots jump from fifth to third, leapfrogging the Chiefs and Seahawks.7

While the Pythagenpat formula doesn’t do much to change the ordinal rankings of the teams, it does highlight what many of us already figured: with a 11-7 record, the Seahawks were the biggest underachievers in the league (by 2.4 wins, using this metric).

Of course, my favorite of these type of metrics is the one I created and outlined for Football Perspective, SPat. I like this stat because it’s a very simple tool for measuring a team’s “true” Pythagenpat expectation. Using SPat, the Colts, of all teams, had the most wins over expectation (8 real, 5.6 expected). It also agrees with regular Pythagenpat that the Seahawks were supreme underachievers.

Looking at the middle of the pack, we can see that the Vikings, Lions, and Raiders performed almost exactly how you’d expect, given their strength and the strength of their opponents. That bodes well for the Vikings, who should contend for the NFC North crown or a Wild Card berth next year. However, it doesn’t look good for the Lions and Raiders, who are both stuck behind two superior rivals within their respective divisions. Then again, I could be wrong. As I’ve said before, games aren’t won in spreadsheets.

Expected Points Contributed

Popularized by Carroll and Palmer, and later refined by Brian Burke, expected points (EP) is a measure of the theoretical point value of having the ball at a given spot on the field. Expected points added (EPA) is a measure of how much value above or below what was expected a team actually gained on a given play. Pro Football Reference publishes total EP contributed by both team offenses and team defenses for the regular season.8 I calculated EP contributed during the playoffs as well, and ranked teams by their per-game differential.

You can read the table thus: in 19 games, Carolina’s offense contributed 128.94 EP, and their defense added 123.01 EP. That’s a total of 252.0 EP added by offense and defense. On a per-game basis, that’s 6.79 from offense, 6.47 from defense, and 13.26 overall.

TmGEXPOEXPDO+DEXPO/GEXPD/GEXP/G
CAR19128.94123.01252.06.796.4713.26
SEA18129.3252.36181.77.182.9110.09
ARI18164.6313.96178.69.150.789.92
KC1883.2377.31160.54.624.308.92
PIT18107.9945.65153.66.002.548.54
NE18128.5213.37141.97.140.747.88
NYJ1655.8254.43110.33.493.406.89
CIN1786.8028.40115.25.111.676.78
DEN19-92.98183.2890.3-4.899.654.75
GNB1868.50-2.6465.93.81-0.153.66
MIN1724.4112.8237.21.440.752.19
ATL1660.21-53.037.23.76-3.310.45
BUF1649.56-49.070.53.10-3.070.03
HOU17-65.4743.74-21.7-3.852.57-1.28
TB1658.23-80.13-21.93.64-5.01-1.37
WAS1763.18-88.27-25.13.72-5.19-1.48
DET1640.73-85.51-44.82.55-5.34-2.80
NYG1658.71-105.13-46.43.67-6.57-2.90
STL16-83.3726.21-57.2-5.211.64-3.57
OAK160.17-59.91-59.70.01-3.74-3.73
NOR16144.20-204.10-59.99.01-12.76-3.74
CHI1623.83-87.45-63.61.49-5.47-3.98
SD1622.23-91.53-69.31.39-5.72-4.33
PHI16-5.18-65.84-71.0-0.32-4.12-4.44
MIA161.89-92.50-90.60.12-5.78-5.66
IND16-42.31-52.30-94.6-2.64-3.27-5.91
JAX16-8.44-88.89-97.3-0.53-5.56-6.08
DAL16-44.61-64.24-108.9-2.79-4.02-6.80
TEN16-67.51-47.06-114.6-4.22-2.94-7.16
DAL16-42.02-87.67-129.7-2.63-5.48-8.11
CLE16-37.37-114.79-152.2-2.34-7.17-9.51
SF16-51.85-114.47-166.3-3.24-7.15-10.40

As you might expect from a team that led the NFL in point differential, the Panthers led the league expected points added per game. What you may find interesting is that, despite leading the league in scoring, Carolina was only fourth in expected points added from offense. The top honor belongs to the Cardinals, who somehow managed to fly under the radar for most of the year. Arizona’s division rival, St. Louis (at least in 2015), had the lowest offensive output on a per game basis (-5.21 EP).

The second worst offense in the league by this metric belonged to the Super Bowl champions, giving more fuel to the “defense wins championships” crowd.9 Peyton Manning was a turnover machine in the regular season before resigning himself to a game-manager position in the playoffs. In his absence, they weren’t exactly the second coming of the Greatest Show on Turf.

What interests me is the offensive and defensive splits for the Broncos and the Saints. Denver had, by far, the top defense in the league by expected points. Wade Phillips’s defense was so good, in fact, that it outweighed the offense’s futility; the team finished with 90.3 combined expected points (4.75 per game).

The Saints, on the other hand, had the league’s second best offense, contributing 144.20 expected points (9.01 per game). However, their defense may have been the worst in modern history. They added -204.10 expected points on the year. Their defensive mark of -12.76 per game makes them the most extreme unit in the NFL. In fact, their defense was so bad that, despite having the number two offense in the land, the Saints finished the year with the twelfth lowest combined expected points per game of any team.10

The Seahawks, known for a defense that has topped the league in scoring four years running, come out ahead of the higher scoring Panthers and Patriots when measured my expected points from offense. In fact, EP estimates that Seattle derived more value from Russell Wilson and company than from Earl Thomas and friends.

The Vikings and Jets are interesting too, as they also rate higher in EP from offense than from defense. This goes against my perception of both teams, who I thought were clearly defense-oriented. Of course, teams are almost always going to get more value EP value from their offenses. If defensive EP is even close to offensive EP, it’s usually safe to assume the team is either well-balanced or heavy on defense.

Total Adjusted Yards

My bread and butter metric. You can read the full definition here, but here’s the Cliff’s Notes version: TAY is a stat that measures output, translated into yardage totals, and awards first downs and touchdowns while penalizing interceptions and fumbles.

Read the table thus: the Panthers gained 10206 total adjusted yards this season at a rate of 8.06 per play, which was 0.79 TAY/P better than average. They allowed 7276 TAY at 5.87 per play, which was 1.53 TAY/P better than average. Their average marginal score was 1.16 (0.79 offense and 1.53 defense). Their weighted average, which gives more credit to offense, was 1.10, tops in the NFL.11

TmTAYoTAY/PoMTAY/PoTAYdTAY/PdMTAY/PdAVGWtAVG
CAR102068.060.7972675.871.531.161.10
SEA94318.160.8867496.331.070.980.96
ARI96048.250.9775346.760.640.810.83
NE97148.210.9481386.840.550.740.77
KC85487.770.4975016.490.900.700.67
NYJ80457.490.2164056.321.070.640.58
CIN83997.830.5674776.790.600.580.58
PIT91017.990.7185307.140.250.480.51
DEN81936.56-0.7170755.691.710.500.33
TB77607.640.3677327.40-0.010.180.20
BUF76257.500.2376177.49-0.100.070.09
MIN72397.03-0.2477727.260.14-0.05-0.08
GNB83857.03-0.2483237.260.14-0.05-0.08
DET76957.470.1978457.86-0.47-0.14-0.09
ATL78727.330.0577527.78-0.39-0.17-0.14
HOU76926.47-0.8170786.750.64-0.08-0.19
NOR91298.331.0592469.27-1.88-0.41-0.21
OAK68856.82-0.4579207.300.09-0.18-0.22
WAS80277.430.1587868.14-0.75-0.30-0.24
STL59756.49-0.7876687.030.36-0.21-0.29
JAX71427.06-0.2286457.84-0.45-0.34-0.32
MIA69717.14-0.1485537.95-0.56-0.35-0.32
NYG77677.380.1091418.27-0.88-0.39-0.32
SD80197.290.0178398.17-0.78-0.38-0.33
CHI74167.24-0.0380228.18-0.78-0.41-0.36
BAL74986.92-0.3577647.76-0.37-0.36-0.36
PHI75406.84-0.4488337.69-0.30-0.37-0.38
DAL66056.81-0.4779828.01-0.61-0.54-0.53
TEN63326.49-0.7977377.71-0.31-0.55-0.58
IND66126.29-0.9981157.57-0.18-0.58-0.64
CLE68036.53-0.7583278.34-0.95-0.85-0.83
SF61816.38-0.9090638.38-0.99-0.94-0.94

This table is initially sorted by weighted average, so I sort of Tarantinoed it by telling you the end first. If you sort by offensive total adjusted yards “TAYO” you’ll see the Panthers in front by a significant margin. However, when you look at the metric on a per play basis, the Panthers drop to fifth, and the rival Saints (perhaps surprisingly) move into first place. For the regular season, the Cardinals had a pretty big edge in offensive TAY/P. However, their play fell off pretty drastically in the playoffs, causing them to fall behind the idle Saints.

Seven of the top ten teams in offensive TAY/P made the playoffs. Here’s a sentence I never thought I’d write: the three of the four worst offenses to make the playoff belonged to teams led by Peyton Manning, Adrian Peterson, and Aaron Rodgers. Unsurprisingly, the offense led by the dynamic duo of Brian Hoyer and Ryan Mallett rated as the worst among playoff teams. It’s safe to say were it not for their residence in the dreadful AFC South, they would just be another bad offense left out of the postseason.

When you rank teams by defensive TAY, the Jets and Seahawks rate far better than any other team. Much of this can be chalked up to the fact that the third ranked Broncos played in 19 games, while the Jets and Seahawks played 16 and 18 games, respectively. On a per game basis, Denver (372.4), Seattle (374.9), and Carolina (382.5) all ranked ahead of the Jets (400.3).

Despite not playing in any extra games, the Saints, Giants, 49ers, and Eagles allowed more total adjusted yards than any playoff team. The Saints once again achieve the impressive feat of having the NFL’s best offense and worst defense as measured per game and per play. In fact, when you look at the two columns displaying marginal offensive and defensive TAY/P, New Orleans owns the two most extreme scores (-1.88 defense and +1.05 offense). This is basically the story of Drew Brees‘s career with the Saints. He has led some of the most consistently great offenses in the league and been hampered by mostly lousy defenses. The one year the Saints had a remarkable defense, Brees led them to the Promised Land.12

The column marked “AVG” gives you a pretty simple descriptive stat of how teams performed. Heading into the playoffs, I felt the three best teams in the league were in the NFC, and this metric supports that idea (the Super Bowl didn’t support that idea, of course). The Jets may rank surprisingly high for a team that didn’t make the playoffs, but they really were a solid squad. Had they been blessed with the fortune of playing in the AFC South or NFC East, they would have likely had a shot at a first round bye (the same can be said for the Chiefs).

Using this metric to rate teams, only eleven come out above average. Three of those teams missed the playoffs, while four below average teams (including three division winners) saw postseason action.

The “WTAVG” column displays a more predictive stat that shows how teams are likely to play in the future.13 Not much changes when you sort by this column, but there are a couple of notables, so strap in.

The Saints, who you’re probably tired of hearing about by now, move up nine places (26th to 17th) when I place more emphasis on offense. This ranking seems more in tune with the product I saw on the field. The Eagles drop four spots (23rd to 27th), which feels strange considering Chip Kelly’s offensive pedigree. Fortunately for Kelly, he’s taking over a team with the second worst offense, second worst defense, and worst overall ranking by these metrics. He can take a page out of the Bill Parcells handbook and take advantage of regression to the mean on a new team.

Perhaps the most surprising rating here is that of the Bucs. Despite rating rather poorly by other metrics, they come out as the tenth rated team by weighted TAY/P. The major contributing factor here is their 5.91 yards per play on offense, which ranked third in the league. They also ranked eighth in first down rate, picking up a new set of downs on 29.5% of their plays from scrimmage. Aside from those two areas, Tampa didn’t really stand out statistically anywhere. Perhaps the most significant component in their position is the mere fact that the NFL, as a whole, was rather weak this year. In a normal year, a team with Tampa’s stats would fall outside the top ten.

I did not adjust these stats for strength of schedule, but I plan on doing that soon. Come back later this week for total adjusted yardage stats with SOS-adjustments.

  1. Somehow, despite this being without a doubt the worst season of his career, this second ring significantly improved perennial bridesmaid Peyton Manning‘s legacy. The fact that he won his second title in a year in which he was arguably the worst quarterback in the league should tell us something about using team achievements to judge individual players…but it probably won’t.
  2. This was hardly a Jets-Colts, Chiefs-Vikings, or Giants-Patriots level upset, but it was another in a long line of underdog triumphs. In fact, it’s accurate to say that the best team on paper (or in a spreadsheet these days) fails to win the title more often than not. Remember, if the best team always won, every champion would be undefeated.
  3. That’s 15 wins, 4 losses, and 20 bonus wins, giving them 35 weighted wins in 39 weighted games. See weighted playoff score.
  4. I used 2.6 points for home field advantage, based on research I have done covering the past eleven seasons. To adjust, I simply add 1.3 to the score of each away team and subtract 1.3 from the score of each home team prior to running the SRS calculation. I did not include HFA adjustments for the Super Bowl or for “home” games played on foreign soil.
  5. I don’t think I’m going out on a limb by saying that. If you’re looking for hot takes, you’ve come to the wrong place. My takes are cucumber-cool.
  6. Definitions and explanations of Pythagorean, Pythagenpat, and SPat can be found here.
  7. Th Rams and Saints switch places, but really who cares that the 20th and 21st ranked teams are now the 21st and 20th ranked teams?
  8. Special teams EP is not readily available. I could use play by play data to calculate it myself, but I don’t think the payoff would be worth the work.
  9. Yes, defense does win championships. So does offense. And special teams.
  10. I haven’t research it yet, but I am willing to bet that the 348.3 expected point difference between the Saints offense and defense is the most stark disparity in NFL history.
  11. The weights are 57% offense and 43% defense. These come from personal study, as well as study from Chase Stuart of Football Perspective and Aaron Schatz of Football Outsiders. These weights are added to give more value to teams with better offenses, as offense tends to be more consistent from game to game and season to season and has shown to be a better predictor of future performance.
  12. Statistically, Brees has been on par with both Peyton Manning and Tom Brady for most of his career, but his small number of postseason appearances and regular season wins distorts the public’s perception of him. He throws more interceptions, but that is often a byproduct of having to force plays in order to win games; often, if he didn’t lead the team to 30 points, they weren’t picking up the W. He is similar to his famous interception-tossing contemporary Brett Favre in that regard.
  13. Obviously they won’t be playing in the future, as the season is over, but the stat has proven useful in picking games during the season.