Strength of schedule is a fickle variable, one that should tell us more about a particular opponent but really does not. Though this formula does account for S.O.S, that's not always the best indicator of whether or not a team has faced "legitimate" competition.
Despite what Bill Parcells may have you believe, a team is not always what their record says they are. They are often a flawed representation of late-game variance and fluky injuries, especially when you're splitting hairs among closely contested teams. That's how you end up with Tampa Bay being the 20th-ranked team in Football Outsiders' DVOA despite entering the week a single win.
Still, there are instances where a team shows some pretty alarming splits among the haves and have-nots. In the case of one likely playoff team, that makes for some troubling signs going forward.
Fool's Gold?
The San Francisco 49ers are a top-10 team in these rankings, which seems to fit with their public perception as a borderline Super Bowl contender. But while the NFC champs started the season with an impressive victory over the Packers, they've since beaten the Rams, Texans, Cardinals, Titans and Jaguars; they have lost to the Seahawks, Colts, Panthers and Saints. Only the Arizona game (which was at Candlestick) stands out as a semi-impressive win, and all four losses are to likely playoff teams.
Beyond the schedule, the red flags are clearly there, nearly all of them on offense. The passing game is where things have most painfully gone wrong, as Colin Kaepernick has not really progressed in his first full season as a starter. A year after finishing 10th in EPA per play, Kaepernick is down to 17th in that category this season, dropping from 0.16 to 0.09. As you can see, an increase in traditional drop-back reads from the pocket have had an adverse affect on his numbers:
The high sack number there is particularly alarming for such a mobile quarterback. Plays like this and this are indicative of a player who simply is not processing information fast enough. In both instances, Kaepernick could probably have scrambled away from the impending pass rush, but spent too much time trying to decipher his reads and consequently failed in his pocket awareness.
However, it's shortsighted to blame the Niners struggles solely on their young quarterback. San Francisco seems intent on playing anachronistic 1 WR-2RB-2TE sets on the majority of their plays, greatly limiting their speed and athleticism in a league where passing is predicated on those very traits. For all his strengths, Anquan Boldin is not a burner, and his production has been wildly erratic as teams have focused on taking away Kaepernick's primary read.
Vernon Davis is still a fine tight end, and Michael Crabtree's return is imminent (though it's tough to ask a player returning from a torn Achilles to be the offensive savior). But with a full offseason for opposing teams to adjust, the 49ers offense is not creating the same amount of confusion as they did when teams had to prepare for them on the fly in 2012.
Kaepernick has not taken a decisive step in his development as a passer, and the 49ers have not maximized his chances for success by asking for traditional reads from the pocket while not surrounding him with adequate weapons. San Francisco seems to realize this, but their significantly dialed-back offense seems unlikely to cause much damage against top-notch playoff defenses.
With games against the Redskins, Rams, Bucs and Falcons, the Niners should be able to squeeze out 10 to 11 wins and make it into the postseason, albeit likely as a wild card team. Kaepernick is a talented high variance quarterback, so it's not out of the realm of possibility that he pulls off a Flacco-esque hot streak through the playoffs. But as currently constructed on offense, the 49ers seem unlikely to redeem themselves for their near-miss in the Superdome last February.
A Turbulent Flight
The New York Jets became the first team to alternate wins and losses in their first 10 games of the season with their no-show in Buffalo last week. New York's Jekyll-and-Hyde identity seems confounding, but digging deeper, we can find some patterns in an absurdly scattershot team.
First off, the Jets defense is a rock-solid foundation, one that essentially shows up every week. New York has the league's best run defense success rate, anchored by arguably the league's best young defensive line in Muhammad Wilkerson, Sheldon Richardson and Damon Harrison. The pass defense is a little more hit-and-miss, but is about league average and generally reliable enough.
As most people can understand, it's the offense that fluctuates severely from game to game, particularly rookie quarterback Geno Smith. Smith's performance is not really an every-other-week type of deal, but rather a home-road split, a fairly common malady for a rookie QB. Indeed, per Pro-Football-Reference, his home-road splits and win-loss splits match almost perfectly:
But opponents obviously matter more than location, and in that sense, Smith's struggles are a bit more confounding. If he played better defenses on the road and worse defenses at home, then his big splits would be a bit more understandable, though still concerning. But to date, the Jets have played five home and five away games. Take a look at how the opposing pass defenses from home games have compared to opposing defenses from road games on average:
New York has actually faced slightly better pass defenses at home! Trying to make sense of Smith and the Jets is probably a futile exercise, a product of their roster composition. The Jets rely on a ton of young players to play important roles. While many of those players are talented and performing at the higher end of their expectations this season, their youth generally means they are higher variance players. Thus, the Jets as a whole have built their 2013 foundation upon a boom-or-bust base.
The Jets have already exceeded expectations this season, and Rex Ryan deserves credit for extracting so much out of a raw defensive unit. It's impossible to say who will emerge out of the AFC's mediocre morass to snatch the sixth seed, but the Jets probably have the highest ceiling of an (admittedly uninspiring) eight-team smorgasbord of 5-5 and 4-6 teams.
Quick Hits
- The Saints are the new number one team in the rankings, narrowly ahead of Seattle after taking down the previously sixth-ranked Niners. If New Orleans can get past Atlanta on Thursday, that will give the 9-2 Saints a chance to snatch the top seed from the 10-1 Seahawks on Monday Night Football in two weeks. ESPN producers are thanking the heavens for Cam Newton vs. Tom Brady and Drew Brees vs. Russell Wilson within a three-week span.
- I've yet to write a single word about the Buffalo Bills this season, which is what happens when you start Thad Lewis and Jeff Tuel for a month. But the Buffalo defense has turned a decisive corner in 2013, with promising young stars in all three units. Marcell Dareus, Kiko Alonso and Stephon Gilmore are all 24 or younger. Mario Williams is still not worth $100 million, but with 12 sacks and the fourth-highest WPA among defensive ends, he's been excellent. If the team can appease Jairus Byrd and sign him to a long-term deal, this could be one of the league's top units very soon.
- Remember when the Chargers looked very sixth-seedish? Seems like a while ago. San Diego has dropped three in a row, and probably deserves their status as the least efficient defense. The Chargers are the worst run defense and third-worst pass defense, which is unsurprising when you look at their non-Eric Weddle defensive personnel. Five of San Diego's last six games are against the Chiefs, Broncos, Bengals and resurgent (!) Giants, so the 4-6 Chargers can probably kiss those wild-card ambitions goodbye.
Check out all the happenings below:
RANK | TEAM | LAST WK | GWP | Opp GWP | O RANK | D RANK |
1 | NO | 2 | 0.80 | 0.47 | 2 | 8 |
2 | SEA | 1 | 0.76 | 0.44 | 5 | 1 |
3 | DEN | 3 | 0.73 | 0.43 | 1 | 16 |
4 | PHI | 4 | 0.67 | 0.48 | 4 | 19 |
5 | GB | 8 | 0.65 | 0.52 | 3 | 27 |
6 | CIN | 5 | 0.65 | 0.49 | 11 | 2 |
7 | CAR | 7 | 0.61 | 0.49 | 12 | 10 |
8 | ARI | 10 | 0.60 | 0.51 | 19 | 6 |
9 | DET | 9 | 0.59 | 0.51 | 7 | 22 |
10 | SF | 6 | 0.59 | 0.54 | 16 | 4 |
11 | NE | 12 | 0.55 | 0.50 | 13 | 14 |
12 | NYG | 11 | 0.53 | 0.54 | 24 | 3 |
13 | CHI | 13 | 0.52 | 0.54 | 8 | 26 |
14 | IND | 16 | 0.49 | 0.47 | 10 | 29 |
15 | CLE | 20 | 0.49 | 0.49 | 28 | 5 |
16 | SD | 17 | 0.48 | 0.46 | 6 | 32 |
17 | KC | 14 | 0.47 | 0.47 | 30 | 7 |
18 | PIT | 22 | 0.46 | 0.46 | 17 | 23 |
19 | HOU | 18 | 0.46 | 0.48 | 20 | 9 |
20 | DAL | 15 | 0.46 | 0.54 | 14 | 21 |
21 | TEN | 19 | 0.46 | 0.47 | 15 | 18 |
22 | MIA | 25 | 0.44 | 0.50 | 25 | 15 |
23 | BUF | 23 | 0.44 | 0.51 | 22 | 13 |
24 | NYJ | 21 | 0.42 | 0.51 | 23 | 12 |
25 | WAS | 24 | 0.42 | 0.54 | 18 | 25 |
26 | ATL | 26 | 0.39 | 0.52 | 9 | 31 |
27 | MIN | 28 | 0.36 | 0.55 | 26 | 24 |
28 | STL | 27 | 0.34 | 0.50 | 21 | 28 |
29 | BAL | 29 | 0.34 | 0.53 | 29 | 17 |
30 | TB | 30 | 0.34 | 0.56 | 27 | 11 |
31 | OAK | 31 | 0.29 | 0.49 | 31 | 20 |
32 | JAC | 32 | 0.20 | 0.52 | 32 | 30 |
TEAM | OPASS | ORUNSR% | OINT% | OFUM% | DPASS | DRUNSR% | DINT% | PENRATE |
ARI | 6.2 | 37 | 4.2 | 2.6 | 5.6 | 60 | 3.4 | 0.33 |
ATL | 6.4 | 38 | 2.9 | 1.8 | 7.2 | 57 | 1.8 | 0.41 |
BAL | 5.5 | 31 | 3.4 | 1.6 | 6.2 | 57 | 2.0 | 0.45 |
BUF | 5.3 | 39 | 2.4 | 2.6 | 5.8 | 59 | 4.0 | 0.42 |
CAR | 6.0 | 44 | 2.7 | 1.6 | 5.6 | 57 | 4.1 | 0.35 |
CHI | 6.7 | 41 | 2.2 | 1.6 | 7.2 | 55 | 4.4 | 0.37 |
CIN | 6.5 | 39 | 3.6 | 2.0 | 5.0 | 60 | 2.8 | 0.49 |
CLE | 5.1 | 42 | 2.8 | 1.5 | 5.1 | 62 | 2.1 | 0.47 |
DAL | 6.4 | 41 | 1.6 | 1.9 | 7.0 | 55 | 2.8 | 0.44 |
DEN | 8.2 | 42 | 1.5 | 2.9 | 6.5 | 63 | 3.2 | 0.44 |
DET | 7.2 | 40 | 1.9 | 2.0 | 6.9 | 64 | 2.8 | 0.44 |
GB | 7.6 | 44 | 2.9 | 2.0 | 6.7 | 57 | 1.2 | 0.43 |
HOU | 5.9 | 40 | 3.0 | 2.0 | 5.6 | 59 | 1.5 | 0.53 |
IND | 6.3 | 48 | 2.0 | 0.9 | 7.0 | 56 | 2.6 | 0.25 |
JAC | 5.4 | 31 | 4.3 | 1.9 | 7.4 | 56 | 1.2 | 0.38 |
KC | 5.1 | 38 | 1.1 | 1.6 | 5.5 | 61 | 3.3 | 0.32 |
MIA | 5.4 | 42 | 3.0 | 1.6 | 6.1 | 55 | 3.5 | 0.29 |
MIN | 5.6 | 38 | 3.8 | 2.5 | 6.8 | 53 | 2.1 | 0.31 |
NE | 5.8 | 44 | 1.8 | 1.7 | 5.7 | 54 | 3.2 | 0.33 |
NO | 7.5 | 46 | 2.0 | 1.2 | 5.4 | 58 | 3.1 | 0.35 |
NYG | 6.2 | 37 | 4.9 | 2.4 | 6.0 | 62 | 2.9 | 0.42 |
NYJ | 5.7 | 39 | 5.2 | 1.9 | 6.2 | 70 | 1.3 | 0.48 |
OAK | 5.4 | 39 | 3.6 | 2.2 | 6.6 | 57 | 2.0 | 0.46 |
PHI | 7.5 | 48 | 2.0 | 1.9 | 6.5 | 54 | 2.7 | 0.41 |
PIT | 6.3 | 35 | 2.6 | 1.9 | 6.3 | 61 | 1.8 | 0.36 |
SD | 7.6 | 45 | 2.2 | 1.1 | 7.4 | 46 | 1.4 | 0.41 |
SF | 6.1 | 39 | 2.8 | 2.2 | 5.7 | 60 | 3.0 | 0.43 |
SEA | 7.4 | 47 | 2.1 | 2.8 | 5.0 | 60 | 4.4 | 0.59 |
STL | 5.8 | 38 | 1.7 | 2.4 | 7.0 | 59 | 3.6 | 0.46 |
TB | 5.2 | 39 | 2.0 | 1.8 | 6.3 | 60 | 3.1 | 0.64 |
TEN | 6.1 | 42 | 2.5 | 2.7 | 6.1 | 56 | 2.5 | 0.45 |
WAS | 6.4 | 47 | 3.1 | 2.4 | 7.5 | 53 | 2.9 | 0.41 |
Avg | 6.2 | 41 | 2.7 | 2.0 | 6.3 | 58 | 2.7 | 0.42 |
You still haven't fixed the strength of schedule issue I brought up last week. You're counting the bye week as an opponent, dividing everyone's SOS by N+1 games instead of N games. That's why only two teams have Opp GWP above 500, when all teams should be converging towards 500.
That's also why Dallas dropped 7 spots this week - they had their bye last week and their Opp GWP dropped from 0.54 to 0.49. Why? Because the bye week was included as an opponent: 0.54 * 10 / 9 = 0.49
Same thing for St. Louis, the other team that had a bye last week. Previous Opp GWP was 0.51, bye week included as an opponent, now Opp GWP is 0.45 because 0.51 * 9 / 10 = ~0.45
If what James is saying is right, which I believe he is, then the four teams that have not had their bye yet are all overanked as their SOS is artificially higher than other teams. Those teams are Sea (#2), Phil (#4), Cincy (#5), and Buffalo (#16). Seattle should probably be closer to Denver than NO, Philly & Cincy should prob below GB, and Buffalo may rank as low 22-23.
I think this makes a lot of sense as being a Seahawks fan I've noticed how easy their schedule has been (there are 5 teams with 2 wins or less - they have played all 5). I have a hard time believing their SOS is not the lowest. When you adjust Denver's SOS it would end up being: 0.39 * 11 /10 = 0.43, right in line with Seattle's as the easiest in the league
Apparently Sterling doesn't read the comments. Brian, Brian! Can you do something to fix this? Please! --Alessio
You may have answered this in another column, but how come the team data visualizations don't match these rankings? The visualization shows the Panthers as having the most efficient defense in the league, but here their ranking is 11.
Do one of these not include D-Int% as those are fluky?
Jonathan,
I believe the viz is the EPA on the season so it's looking at past performance which includes variables that aren't necessarily predictive of future success. As you mentioned, D-Int% is probably one factor that is weighted less in the rankings than in the EPA viz.
Jonathan-- The visualizations on the front page are defensive/offensive EPA/game. The efficiency rankings aren't the same, they're based on a somewhat more complex formula.
-Jason
The good news about the SOS error is everyone will have had their byes after next week, so it will mostly fix itself on its own. Of course, averaging in a 0 will effect high SOS's more than lower ones, but that should be a minor issue.
Also I inadvertently switched the 10 and 9 when I typed the equation for Dallas. It should be 0.54 * 9 / 10 = 0.49
funny how this article arrogantly mocks Bill Parcells and assumes that the actual records of teams are wrong while some analysis is the thing that really is right (even though it therefore follows logically that the analysis is not capable of predicting wins).
Then add in the irony of not being able to handle bye weeks, and boldly pointing out how the cowboys dropped 7 spots.
There's nothing "bold"er than making a joke about Tony Romo visiting the beach.
I'd like to see this fixed too, though. Or at least explained if it is, for whatever reason, not an error. Even though it'll "fix itself" after this week, may as well fix it now for the sake of future seasons.
not only is Mr. Xie, or whatever his last name really is, not reading the comments about the bye weeks, and dissing Bill Parcells, he riders the media wave about Tony Romo choking, which tells me he knows nothing about football because anybody who follows the media basically knows nothing about football. okay, exaggerating a little, but not by much.
@ "funny how this article arrogantly mocks Bill Parcells and assumes that the actual records of teams are wrong while some analysis is the thing that really is right (even though it therefore follows logically that the analysis is not capable of predicting wins)"
Come down to earth. If you wouldn´t be a one time lurker, you would know what this Parcells thing means. That was not offending at all, but going into the direction that game outcomes are 50%+ random as Brian Burke explained some years ago. He just used the Parcells quote back then, and it´s repeated now. Nothing wrong w/it at all.
No one said this system can predict wins. But what it can do is showing tendencies for future outcomes. In that case the system was great. For example calling the ATL collapse before it was happening, having GB on top before they won the SB, having the NYG on 4th before the won the SB. No one else saw that coming.
Your usual Prisco power rankings can´t do that either, since their rankings go by W-L record. Every idiot can do that!
Greetings, Karl, Germany
I'll fix it. I knew there was a problem but I thought I had it licked. One of my data sources has been changing their format.
@ Wizard
Don´t pretend to be wiser than you are...
About Parcells, see my last post.
And to enlighten you, Mr. Xie was just kidding about Romo. Everybody on this site knows that he is no choker. He was just sarcastic, b/c mainstream thinks he is.
Next time better read and understand before posting. Otherwise you´d look like a fool.
Karl, Germany
I dont understand the ranking of saints and seahawks. Seahawks has a stronger sos and higher ranked offense and and defense, but have the same score as the saints.
Should everyones Opp GWP be below 0.50?
> I dont understand the ranking of saints and seahawks. Seahawks has a stronger sos and higher ranked offense and and defense, but have the same score as the saints.
Check the penrate for the Hawks, it's miserable.
Still working out some bugs. The Opp GWP is going to be off in the table, but in the calculations it gets normalized to a .50 average point.
Now finally corrected. (I know, never say finally when dealing with computer code.) The Opp GWP was being averaged over the # of opponents+1...
What is penrate? Why is it not in the site glossary?
Yay! It's fixed. And there were only minor changes in the team's GWP.
Anon - penrate is the penalty rate of the team, which means the Seahawks commit (or are called...) for way more penalties than the average team, particularly the Saints who have fewer than average penalties.
It looks like changes were made to the first table, but not the second table :(
>What is penrate? Why is it not in the site glossary?
See http://www.advancednflstats.com/2013/01/dont-overlook-effect-of-penalties.html for some sort of explanation.
TL;DR: Penalty rate is defined as penalty yards per snap.
There were never any problems with the 2nd table.
Seahawks still hands down the best team in the league, this will be proven in the playoffs, barring injuries to any key players.
Parcels was a master motivator, that quote was meant as nothing more then a motivator to his team, it's not like he actual believed it.
Parcels was great for the media and fans in post game interviews, because of such things, mostly said to influence his teams, people make to big a deal of it.
The big game this week the model based on GWP has Denver -3 VS NE.
The model did win last week's biggest difference between it's spread and the actual spread with the Eagles -3.5 VS Redskins.
There's no big differences this week, the biggest being Bears -3, actual +1, 4 point difference.
Cardinals -6.5, actual -2.5, another 4 point difference.
RE: Karl from Germany,
In regards to predicting future outcomes, the model does have some success and some failures.
Of coarse one could point-out a few successes, one could do that using any model, even the worst model.
Last year the model had Carolina top 5 team and they finished 7-9, certainly not a top 5 team.
This year the model had Carolina 12th when the should of been a top 5 team a few weeks back.
Being 12th was not a good indication of the quality of Carolina, they now are 7th, but should be top 5. They are better than Bengals and Packers even with Rodgers.
And let's not forget last season the model said Ravens were not as good as their 9-2 record, all the Ravens did was win the Super Bowl.
The model had Denver no.1 heading into the playoffs and the largest GWP of any playoff game was Denver over Baltimore. Not very predictive was it.
Mitch,
Seattle has had a fairly weak schedule and a number of close games against crappy teams (TB anyone?). Wilson has been good enough, but certainly not great. Soon they will play the 49ers in Candlestick, and the Saints will be no pushover, either.
The model had Carolina as very strong last year... right as they went on a midseason tear that has continued into this season.
Models that Brian Burke has referenced on this site show that only 1 in 4 years has the 'best' team in the NFL as the Superbowl winner. Variance happens good sir. We can both cherry-pick data but overall the system you seem to question does a better than average job of predicting outcomes.
Cheers,
J