Titans vs. Worst No. 1 Seeds Ever
NFL Week 17 - We thought there might be a lot of shuffling on top of the Football Outsiders DVOA ratings this week after some surprising results in Week 17. But instead, the top three teams stayed exactly the same, although all three saw their actual DVOA ratings drop. Dallas is still No. 1 despite a loss to Arizona. Tampa Bay is still No. 2 despite a close win over a bad Jets team. Buffalo is still No. 3 despite playing a game against Atlanta that was much closer than the final score of 29-15 indicates. The Falcons narrowly outgained Buffalo on a per-play basis and won the turnover margin as well. Our Post-Game Win Expectancy (PGWE) formula suggests that given the VOA splits in this game, we would usually expect Buffalo to only win 20% of the time.
Dallas was at 30.0% DVOA last week and drops to 28.3% DVOA this week, which means we don't have any teams above 30% right now. This would be only the eighth year since 1983 where no team topped 30% DVOA during the regular season. This is where I note that I've made a decision not to do a bunch of comparisons of historical numbers to 2021 DVOA through 16 games in this week's analysis. Yes, the season is one game longer now and so the most accurate comprison of "like to like" would be to compare this year's ratings now to historical ratings, rather than doing a bunch of comparisons and lists next week. But the fact is that the 17-game season is here to stay, until it inevitably becomes the 18-game season, and it's easier to just do a big end of the season wrap-up when the season actually ends. So that's what I'm going to do next week.
Returning to a look at this week's ratings: New England moves up one spot to fourth in total DVOA after beating the Jaguars 50-10 on Sunday. The Patriots also move back to the top spot in weighted DVOA. Yes, I know it seems a bit ridiculous to have the Patriots and their rookie quarterback as our best team going into the final week of the season. Didn't we just run a Film Room piece pointing out Mac Jones' weaknesses and the reasons why the Patriots are not built for success in the postseason? Yes, we did. We also know that historically, big dominating wins are the best indicator of a team's success going forward. The Patriots now have three of the top six games of the season by DVOA, all coming during the last two months. The dichotomy between the Patriots' performance in their big wins and their performance in losses certainly suggests we once again have to look at how much weight to give blowouts. But as we have written many times, we give a lot of weight to blowouts because history suggests they're predictive.
As usual, there's more movement among the teams in weighted DVOA than in total DVOA, as we not only add on Week 17 games but also lower the weight of any game played more than five weeks ago. Dallas drops one spot to No. 2. Then Kansas City moves up to No. 3 in weighted DVOA. Kansas City had a DVOA over 30% for its loss to Cincinnati, with a PGWE of 78%. That game turned heavily on penalties, which are mostly not included in DVOA. The Chiefs only had 21 more penalty yards than the Bengals, but the timing of the penalties was terrible, starting with the illegal use of hands that essentially gave the Bengals the game at the end.
Behind the Chiefs are the Packers, who move up four spots in weighted DVOA to No. 4. I've been saying for a few weeks that what we really needed to see from Green Bay was a big, dominating victory. Well, they finally got one, 37-10 over Minnesota, although with the caveat that it came against a backup quarterback. Green Bay's DVOA for this game was 77%, the Packers' best single game of the season. It moves them up not only to fourth in weighted DVOA but to eighth in total DVOA and to the top spot in our Super Bowl odds.
The Packers are one of three playoff teams whose DVOA rating has been lower than their win-loss record for most of the season. Let's talk a little bit about the other two.
Tennessee Titans: Lowest DVOA Ever for a No. 1 Seed
Kansas City's loss to Cincinnati means that the Tennessee Titans will probably earn the No. 1 seed in the AFC. All they have to do is get past the Houston Texans this week. I got in a lot of trouble with Titans fans when I pointed this out on Twitter earlier this week, but the Titans will go into the playoffs with the lowest DVOA rating ever for a No. 1 seed. Their current DVOA of -2.4% would make them the first team to ever get a No. 1 seed with a below-average total DVOA. Here's a look at the worst No. 1 seeds by DVOA since 1983:
|Worst DVOA for No. 1 Seed, 1983-2021|
|*Strikebreaker games not included.|
In fact, right now the Titans have a lower DVOA than any No. 2 seed ever, not just any No. 1 seed ever, although a win over Houston in Week 18 is likely to change that. Three teams have earned No. 2 seeds with negative DVOA for the season: the 2005 Chicago Bears (-1.8%), the 2004 Atlanta Falcons (-1.7%), and the 2000 Minnesota Vikings (-1.3%). All of those teams were slightly ahead of the 2021 Titans.
There are a couple of things to point out here to give Titans fans reason for optimism despite their low DVOA rating. First, you may notice that almost none of the teams on the table above lost their first playoff game. Only the 1985 Raiders and 1992 Steelers lost in the divisional round. The power of that bye week is strong! The 2015 Broncos are on that table and they won the Super Bowl. But the 2015 Broncos also had a much better regular season than the 2021 Titans, with almost 20 points of DVOA separating them.
The other reason for optimism, as Titans so enthusiastically pointed out to me, is that the Titans have suffered a lot of injuries this year. They lost Derrick Henry for half the season. They played some games without either of their top two receivers, veterans A.J. Brown and Julio Jones. They've had a number of offensive line and defensive injuries as well. The Titans will be a lot healthier in the playoffs than they were in most of their losses this season. This is true! With a healthy roster, the Titans aren't really the worst team to ever win a No. 1 seed despite having the lowest DVOA. They're certainly better than the 2000 New York Giants, for example. The question is: how much do we want to mentally adjust the Titans' DVOA rating for the injuries they had to deal with? We don't have full AGL numbers for this season yet but the Titans certainly haven't faced as many injuries as, for example, the Baltimore Ravens. They'll be high in AGL, but not the highest team in the league. You may have seen a stat that the Titans have set an NFL record for the number of players used on the active roster, but even our own Titans expert Tom Gower told me that's often because Jon Robinson has "smartly been very aggressive with changing out bottom-of-the-roster players for specific matchup reasons." It's not entirely because of injuries.
Let's say, for example, that we say that the Titans real offense is only the offense when both A.J. Brown and Julio Jones are active. Now you've improved the Titans' offensive DVOA from -6.6% to 8.4%, which is pretty significant! You've also picked a sample that has just seven games, almost all of which occurred more than two months ago. And if you're a Titans fan, you also would probably argue that the Week 1 loss to the Cardinals doesn't matter anymore, just like that Packers Week 1 loss to New Orleans, so we would want to take that game out, and now we've got 16.0% offensive DVOA but it's a six-game sample. Don't even get me started trying to figure out which games should and shouldn't count on defense because like many teams during an NFL season -- especially during the last two NFL seasons with COVID -- the Titans have been shuffling players in and out of the lineup all season. Most of the other playoff teams could make a similar argument that we should only be looking at a certain subset of their games when their rosters were healthier. Obviously, a couple of playoff teams have been particularly healthy; the Patriots come to mind. But it's very difficult to determine the specific impact of injuries to players other than at the quarterback position, and the Titans never had to use their backup quarterback all season.
Cincinnati Bengals: EPA vs. DVOA
I've written plenty about the Green Bay Packers and the Tennessee Titans this year. I've written less about the Cincinnati Bengals. That's because I was able to explain what was going on with the Packers and Titans a lot easier. The Bengals situation is much more complicated. Right now, we have the Bengals 17th overall despite the 10-6 record and the win over Kansas City this week. This is different from some other metrics on the Internet, though not all of them. ESPN's FPI has the Bengals only 14th, close to where DVOA has them. PFR's Simple Rating System puts them 13th. But there's a big gap on both offense and defense between the Bengals' DVOA and their rating by raw EPA (Expected Points Added). Right now, the Bengals come out 18th in offensive DVOA and 20th in defensive DVOA. By comparison, they rank 11th in offensive EPA/play and 12th in defensive EPA/play according to RBSDM.com.
We're going to have to go deep into the weeds to try to determine what's going on here. But if you want to go weed-diving, come and join me. Note that most of the stats listed below do not include Monday Night Football, so they may differ from stats you see other places on this website.
Let's start with offense. There's a big first reason why DVOA differs from EPA when it comes to the Cincinnati offense, and that's strength of schedule. No team has played an easier slate of opposing defenses this season, especially against the pass. By DVOA, the best pass defense Cincinnati has faced this year is No. 10 Chicago, back in Week 2.
Here's a table comparing DVOA to VOA (without opponent adjustments) as well as EPA:
|Bengals Offense, 2021|
You can see from that table the gap between DVOA and VOA. You can also see that there's still a big gap to make up between Cincinnati's rank in VOA and EPA on pass plays. On run plays, there really is no gap. The difference is all on pass plays.
The next thing I looked at was down-by-down splits to see if that pointed to a reason for the difference. It certainly doesn't look like it:
|Bengals Pass Offense by Down, 2021|
The Bengals end up with a higher rank in EPA than in VOA on every down. So it's not an issue of whether a specific down is being overvalued by one of the metrics.
Here are three other theories about why the Bengals' pass EPA ranks higher than their pass VOA.
1) Penalties are an issue. The RBSDM EPA model includes all penalties, as long as it can determine that a penalty took place on a called run or pass play. As I noted earlier, we leave penalties out of DVOA for the most part, since they don't tend to be predictive. (We do include defensive pass interference, since it counts for such big yardage totals.) And Cincinnati has the league's biggest gap between its own penalties and penalties called on its opponents. This is a factor on both sides of the ball. In total, the Bengals are last in the NFL with 68 accepted penalties and fifth with 112 accepted penalties for their opponents. But it may not be as big an issue as it seems at first glance. Specifically on offense, the Bengals have 356 penalty yards by opponents but have lost 282 yards on penalties themselves. That gap isn't too big, since the Bengals opponents have a lot of smaller defensive penalties. (We'll get to the Bengals' penalty totals on defense below.)
2) Long plays are an issue. As you may know, DVOA starts to reduce the value of each additional yard on a long play over 40 yards. The reason for this is that often the only difference between a 60-yard pass play and a 40-yard pass play is the line of scrimmage and how much space was left before the ballcarrier reached the end zone. Overall, we've found that these super-explosive plays aren't necessarily as predictive as shorter successful plays. And the Bengals have a lot of these plays. They are second in the NFL with 16 pass plays of 40 or more yards. (The Rams are first at 17, by the way.)
As a result, there's a clear dichotomy between the Bengals' success rate on plays and yards per play. DVOA considers every play separately, it isn't like passer rating where you put a bunch of numbers into mixing bowl and a rating comes out. Nonetheless, you can simplify DVOA by thinking of it as a combination of three things: success rate, yards per play, and turnovers. The Bengals are pretty average in turnovers on both sides of the ball. However, they have 7.42 net yards per pass attempt (including sacks and DPIs), which ranks second in the league. But they also have a 48.1% success rate on pass plays, which ranks 12th. You can see that their VOA ranking on passes (11th) is much closer to success rate, while EPA (fourth) is much closer to net yards per attempt.
3) Touchdowns might be an issue. I'm less sure about this one, but I went and looked at the highest plays in EPA for the Bengals compared to their highest plays in our success value over average (aka the numerator of VOA). The top 19 plays in our values are all touchdowns. In EPA, six of the top 19 plays are not touchdowns. The Bengals have a surprising number of long plays that aren't touchdowns. For example, six of their 16 plays of 40 or more yards did not score. DVOA's success values give a bonus for a touchdown, and it's certainly possible that bonus is a little too high which is causing us to value short touchdowns more than other metrics while undervaluing long non-touchdown gains.
OK, that brings us to defense.
|Bengals Defense, 2021|
Once again, the issue is clearly passes rather than runs, as both DVOA and VOA are similar to EPA on running plays. And unlike for the offense, strength of schedule isn't really an issue on defense. Cincinnati's strength of schedule on defense ranks 12th, close to NFL average.
Let's look at things by down:
|Bengals Pass Defense by Down, 2021|
Same problem we had with offense. There's no reason to believe that DVOA is doing something specifically different than EPA on any particular down. On each down, EPA ranks better than VOA.
Next, I looked at penalties. This is a bigger issue with Cincinnati's defensive ratings than with the offensive ratings. The Bengals have had only 257 yards on defensive penalties this year. Their opponents have 448 yards on offensive penalties. That's the biggest gap in the league in favor of a defense. (The biggest gap in the other direction: San Francisco has had 596 yards on defensive penalties compared to just 291 yards on offensive penalties for opponents! Tampa Bay also has a huge gap, with 487 yards on defensive penalties and just 209 yards on offensive penalties for opponents.)
This is where I admit that otherwise, I am stumped right now at what is causing this difference between how EPA rates the Cincinnati pass defense and how DVOA rates the Cincinnati pass defense. Big gains are not the issue, as the Bengals have allowed an average 10 gains of 40 or more yards. And looking at success rate and yards per play gives you a result that's the opposite of what we would expect. You may remember that on offense, the success rate was closer to the rank for VOA and the yards per play was closer to the rank for EPA. That's because DVOA is essentially built around success rates with bonuses for extra yardage, while EPA is more built around raw gains for each play. OK, but the success rate for the Bengals pass defense is 46.1% (14th) while net yards per play allowed is 6.36 (20th). That means the rank for success rate is closer to the rank for EPA, while the rank for net yards per play allowed is closer to the rank for DVOA!
Anyway, more research will need to be done on why DVOA has the Bengals so much lower than EPA, particularly on defense. Which rating is more accurate? I suppose we'll find out come the postseason, although we won't really find out in the postseason because there is too much randomness in the postseason to judge metrics based on what happens when one team plays 1-4 games. Research on how to improve DVOA's predictive ability is based on years and years of data rather than a handful of games.
* * * * *
Football Outsiders playoff odds, snap counts, and the FO+ database are now all updated through Week 17. A reminder that all our free stats pages, including DVOA and player position stats, require registration to view. This is not a paywall! You only need to register (for free) and then log in to the site to view these pages. While you're at it, you can get a seven-day trial of FO+ and check out the FO+ features like a deeper DVOA database, weekly fantasy projections, fantasy football research tools, and picks against the spread.
* * * * *
Here is the Football Outsiders Top 16 through 17 weeks of 2021, measured by our proprietary Defense-adjusted Value Over Average (DVOA) system that 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 opponent and 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.
WEIGHTED DVOA gives recent games more strength than older games to get a better idea of how well teams are playing now.
To save people some time, please use the following format for all complaints:
<team> is clearly ranked <too high/too low> because <reason unrelated to DVOA>. <subjective ranking system> is way better than this. <unrelated team-supporting or -denigrating comment, preferably with poor spelling and/or chat-acceptable spelling>
Click here for the full table.