Week 11 DVOA: Why DVOA Puts Eagles and Steelers Together

Something is afoot in the state of Pennsylvania, at least when it comes to DVOA ratings. Why do we have Philadelphia so low and Pittsburgh so high? They’re right next to each other, ranked 10th and 11th. Yet the Eagles lead power rankings around the Internet and have the best record in the NFL at 9-1, while the Steelers are considered frauds despite a 6-4 record and just fired their offensive coordinator this morning.

Before we talk about the Eagles and Steelers, let’s run quickly through the rest of the league. The Ravens remain No. 1 in DVOA after a Thursday night win over Cincinnati and the 49ers are second after a healthy win over Tampa Bay, and the gap between the two teams has narrowed since last week. Then there’s a huge gap of almost 13 percentage points before we get to the rest of the league. The Chiefs are still third, but lower than a week ago after losing to the Eagles. The Lions are fourth and the Bills move up three spots to fifth with a blowout of the Jets. The rest of the top 10 goes Dallas, Miami, Cleveland, Jacksonville, and then Philadelphia, which has a higher DVOA than a week ago but moves down a spot with the Jaguars passing them.

Why Doesn’t DVOA Like Philadelphia?

The short answer is “pass defense” but let’s look at the longer answer.

First of all, DVOA is lower on the Eagles than other advanced metrics but not everybody has the Eagles on top of the league. Pro Football Reference SRS puts the Eagles fifth, while ESPN FPI has them seventh. The RBSDM EPA team tiers graphic seems to have them tied with Kansas City at sixth.

Let’s compare DVOA to EPA to see where there are differences. The two biggest reasons for a gap between DVOA and EPA do not seem to apply here. Sometimes DVOA will have a team lower than EPA because of poor special teams, but that’s not the case with the Eagles, who rank fifth in special teams DVOA this year. Another frequent issue is schedule strength, but right now the Eagles’ past schedule ranks 19th, pretty close to the NFL average.

There is a small difference between Philadelphia’s rank in offensive DVOA (sixth) and offensive EPA (fifth). In particular, DVOA has the Eagles lower in passing efficiency. They rank sixth in dropback EPA per play but ninth in pass offense DVOA. However, that rank in pass offense DVOA matches the Eagles’ rank in pass success rate and net yards per pass — and is a little better, in fact. The Eagles are 10th in net yards per pass and 13th in pass success rate. They rank a little better in DVOA because they are below average in turnovers. I would have to analyze specific plays to figure out where the gap between DVOA and EPA comes from, but when you compare DVOA to success rate and net yards per pass, it doesn’t seem ridiculous to rank the Eagles’ passing game ninth (with scrambles included as pass plays).

If we split things up by down, we can see the big gap between DVOA and EPA is on third downs. The Eagles offense ranks roughly the same in DVOA and EPA on both first downs and second downs, with a much better rank on second downs in both stats. On third downs, the Eagles are currently 10th in DVOA but third in EPA. What’s interesting here is that by my definition of success rate, the Eagles have the best success rate in the league on third downs. However, they’re lower in DVOA because they only rank 10th in yards per play on third downs (11th if you include DPIs). Perhaps the problem is that DVOA needs to give more credit for getting a new set of downs and less credit for additional bonus yardage on third downs? Or perhaps this discrepancy makes DVOA more predictive of future performance?

There’s really no difference between DVOA and EPA when it comes to defense. DVOA has the Eagles 17th and EPA has them 18th. Both metrics have the run defense rated much better than the pass defense. In DVOA, the run defense is currently fifth and the pass defense is 19th. The pass rush hasn’t sacked the quarterback as much as a year ago and there are definitely issues with secondary play, in particular with whoever has to be the nickel cornerback. (Last night, it was Bradley Roby, but he was returning from an injury and isn’t as good as he once was anyway.)

One good way to see why DVOA doesn’t love the Eagles so much is to look at week-to-week single-game DVOA. This is my chance to pull out the fun colored week-to-week graph for the first time this season. (The red line is five-week rolling average.) The Eagles have played an average schedule but they haven’t had particularly dominating wins over the bad teams on the schedule. You’ll see that opponent adjustments give the Eagles DVOA below 0% for both the Week 1 25-20 win over New England and the Week 8 38-31 win over Washington. The overtime win over Washington in Week 4 is only slightly above 0%. The Eagles only have a couple of really strong wins: 25-11 over Tampa Bay back in Week 3 and then 31-17 over Miami in Week 7.

The Eagles have only two games over 40%. By comparison, the Bills and 49ers lead the league with five apiece. Five other teams have more than two. The Eagles also stand out because they don’t have any games below -40%, but they are one of seven teams with that achievement so far this year.

Essentially, DVOA sees the Eagles as a team that has a lot of close wins without ever really kicking ass. As you know from “guts and stomps” analysis, blowing out bad opponents is generally more predictive of future success than close wins over good opponents. (The best win, of course, is a healthy win over a good opponent, and the Eagles had that over Miami in Week 7.)

One last issue is fumble recovery. The Eagles have an average fumble recovery rate this year but, based on this analysis from Tom Bliss with the league office, the timing of their fumble recoveries has definitively favored the Eagles. That is luck that EPA would not filter out but DVOA would.

Why Does DVOA Like Pittsburgh?

Matt Canada got fired as the Steelers offensive coordinator this morning, because the Steelers offense looks terrible. They’re currently ranked 28th in the NFL with just 16.6 points per game. The last straw was just 77 net passing yards in a 13-10 loss to Cleveland where the defense clearly played well enough to win the game.

And yet, look down and you’ll see the Steelers are slightly above average in offensive DVOA. They rank 14th. What the hell is DVOA smoking?

I’m not going to pretend that the Steelers offense looks good, and what I’m about to write isn’t an apologetic for Matt Canada’s scheme problems. (For one thing, I don’t understand why the Steelers for years consistently haven’t been able to run play-action well no matter who the offensive coordinator or quarterback has been.) The basic stats don’t look good. The Steelers rank 28th in yards per play and 23rd in success rate. They are better running than passing, certainly, so I think criticism of Kenny Pickett makes sense. The Steelers rank seventh in run offense right now, primarily because of Jaylen Warren, but 20th passing.

There are three main reasons why offensive DVOA likes the Steelers offense much more than other metrics.

1) The average Steelers drive through Week 10 started at the 26.4-yard line. That ranked 31st in the NFL. They have to go farther to score than almost any other team.

2) The Steelers avoid turnovers. They have the fewest turnovers in the league. Although the new version of DVOA this year gives turnovers a smaller penalty than older versions of DVOA, they still matter and avoiding turnovers is a good thing!

3) The biggest issue is schedule. As bad as Canada has been criticized, it seems insane to me to fire him after a game where the Steelers couldn’t score against the Cleveland Browns. Have you seen the Browns the rest of the year? They are killing every offense on the schedule (except, for some reason, Indianapolis).

In the first 10 games of the season, the average defensive DVOA of Pittsburgh’s opponents was -8.5%. If the season ended today, that would be the third-toughest offensive strength of schdeule in DVOA history, trailing only the 1982 Buccaneers (-10.2% in a nine-game strike year) and the 2004 Browns (-8.7%). Half of Pittsburgh’s games have come against top-eight defenses, which are really top-seven defenses if you don’t consider the Steelers themselves. Things get easier for the Steelers, with an average offensive strength of schedule for the remainder of the year. It’s very likely that the offense will look better no matter what and people will attribute it to the offensive coordinator change even if Kenny Pickett isn’t really playing any better and the coordinator change hasn’t really changed things.

(Incidentally, the Bengals are currently on track to end the season with an offensive schedule strength of -7.8%, which would also be the third-toughest ever. Their remaining opponents have an average defensive DVOA of -10.9%, just in case you were thinking of picking up Jake Browning in your fantasy league. Don’t do that.)

Right now, there’s a gap of 14 places between the Steelers’ rank in offensive DVOA and points per game. Is that the biggest of all time? Nope, not even close. Thirteen different teams since 1981 have had a larger gap between those two metrics, led by the 1996 Kansas City Chiefs who ranked 24th in points per game but all the way up in fifth place for offensive DVOA. I went back and looked at these 13 teams. Some of them are similar to the Steelers. Some of them had a gap for totally different reasons. Here’s a breakdown of the top four:

1996 Kansas City Chiefs: One of the great exemplars of Martyball, the Chiefs were very similar to this year’s Steelers except that they had a veteran quarterback (Steve Bono) instead of a young first-rounder. The Chiefs were second in the league in fewest turnovers and second in the league in offensive strength of schedule. Much like this year’s Steelers, the 1996 Chiefs were a much better running team than passing team, with Marcus Allen leading them to rank second in run offense DVOA. However, they were not as good as this year’s Steelers on defense or special teams. This was one time where DVOA was definitely more predictive than points scored, because the Chiefs rebounded to fifth in the league in points scored in 1997. By the way, the quarterbacks coach of this team was a young Mike McCarthy and the offensive coordinator was Nathaniel Hackett’s dad, Paul.

2003 New York Jets: I remember writing about this team a lot in the first year of Football Outsiders. It was the first time I ever wrote about pace. The 2003 Jets had a good offense and a terrible defense (31st in DVOA) but looked like they had a good defense and a terrible offense. Pace was a big reason. The Jets played at a very slow pace, running fewer plays than other teams. Their opponents also played at a slow pace. The Jets also were second in the league in fewest turnovers and played the toughest schedule of opposing defenses. Meanwhile, their defense was 28th in takeaways and played an easy schedule of opposing offenses. So the Jets constantly had these long, drawn-out drives without turnovers against tough defenses and then their defense would face a long, drawn-out drive without turnovers against bad offenses. No wonder their offense was better by DVOA than points per game! Oh, and the 2003 Jets had no defensive touchdowns and only one special teams touchdown. The offensive coordinator of this team was … wait, this can’t be right … also Nathaniel Hackett’s dad, Paul. Yes, again.

1981 Tampa Bay Buccaneers: This one goes a ways back, far enough that we don’t have drive stats or pace stats. But we know the Buccaneers played at a slow pace because only the Oilers, with the worst offense in the league, ran fewer plays than the Buccaneers. This is another team that ranks higher in DVOA than points per game because they were low in turnovers (third) and had a tough schedule (sixth). The Bucs were fifth in the league in yards per play but only 20th in success rate, so there were probably a lot of explosive plays mixed with three-and-outs. Their low rank in points per game is particularly odd because they had five defensive touchdowns. Take those out and there’s an even bigger gap between offensive DVOA and points scored.

2008 Miami Dolphins: The original Wildcat team, the 2008 Dolphins actually ranked better in pass offense (fifth) than run offense (eighth). And this is strange: unlike our previous two teams, the Dolphins played one of the league’s easiest schedules of opposing defenses. The 2008 Dolphins were seventh in the NFL in yards per play and fifth in success rate and even had two defensive touchdowns. They led the league in DVOA in the red zone. So how did they end up just 21st in points per game? Honestly, it’s a bit of a mystery, but there were some factors. This is another offense that did better in DVOA because it excelled in preventing turnovers (first in the NFL). They faced long fields, ranking 25th in average starting line of scrimmage. And pace played a role, as the Dolphins were dead last in both situation-neutral and total pace on offense.

Here’s a table of the biggest gaps between rank in offensive DVOA and rank in points per game, with the current Steelers added on to the bottom. I’ve added a bunch of ranks in relevant stats. For the most part, you can split these teams into two categories: teams like the Steelers that weren’t good in yards or success rate but avoided turnovers and often dealt with a tough schedule, and teams where the ranks in yards and success rate match the rank in DVOA but there was a tough schedule and/or a slow offensive pace.

We’ll have to see what happens to the Steelers the rest of the way, but again, my prediction is that they will score more points simply because the schedule gets better. That doesn’t mean I think Kenny Pickett is a long-term answer at quarterback, but it does mean the Steelers are playoff contenders. They rank seventh in defense, after all. They are a deserved 6-4. Our current simulation puts them in the playoffs 68% of the time.

Hey, Let’s Update Those All-Time DVOA Lists

The Ravens are still the top 8-3 team in DVOA history and surprisingly high on the list of the best teams since 1981 by DVOA. The 49ers would also be in the top 20 if we ran a table through 10 games instead of 11 games. The 49ers have also climbed back onto the list of the top offenses in DVOA history, joining the Browns who are on the list of the top defenses in DVOA history. And we noticed this week that the Baltimore Ravens have one of the top running games ever measured by DVOA, despite using a committee of backs. Lamar Jackson matters! I’ll probably write more about the Ravens running game next week.

Remember that DVOA is normalized so that every year has an average of 0%. That’s how you can have the 1984 Dolphins on the same list here with modern teams:

I’ve gotten some questions about the 1985 Bears and the 2000 Ravens and the top defenses list. The 1985 Bears are close to this but really came on in the second half. The 2000 Ravens get dinged for an easy schedule but played their best defensive football in the playoffs, not the regular season.

Finally, the Ravens and the best running games. Remember that the new version of DVOA treats bad handoffs and aborted snaps as their own “abort” category rather than counting them as running plays for team stats. You can definitely see the effect of Lamar Jackson here when you compare the superstar backs of these other teams to the committee in Baltimore this year. Incidentally, the 2019 Ravens offense was better passing than this year’s offense but only at 14.6% run offense DVOA at this point. Last year’s Ravens offense was at 14.4% run offense DVOA at this point.

* * * * *

These are the FTN DVOA and DAVE ratings through 11 weeks of 2023. Our Defense-adjusted Value Over Average (DVOA) system breaks down every single play and compares a team’s performance to the league average based on situation in order to determine value over average. (Explained further here.)

OFFENSE and DEFENSE DVOA are adjusted for performance indoors and consider all fumbles, kept or lost, as equal value. SPECIAL TEAMS DVOA is adjusted for type of stadium (warm, cold, dome, Denver) and week of season. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.

DAVE is a formula which combines our preseason forecast with weighted DVOA (giving slightly more weight to recent games) to get a more accurate projection of how a team will play the rest of the season. DAVE is currently 24% preseason forecast and 76% actual performance for teams with 10 games and 16% forecast with 84% actual performance for teams with 11 games. DAVE now incorporates projections without Joe Burrow and Deshaun Watson for the Bengals and Browns, respectively.