Week 10 DVOA Ratings
Another week, another flip on top of the Football Outsiders DVOA ratings. New England and San Francisco have been No. 1 and 2 in our DVOA ratings since after Week 4, but this is the fourth time they've switched places. This week, the Patriots are back on top as San Francisco's rating takes a hit from last night's close loss to Seattle.
It's a game that DVOA does not think was as close as it looked. Seattle ended the game with 46.8% DVOA compared to -3.1% DVOA for San Francisco. The problem wasn't defense; in fact, San Francisco's defensive DVOA for the season is better after last night's game than it was beforehand. No, the problem was San Francisco's worst offensive performance of the season by far. The 49ers averaged just 3.9 yards per play for the game (compared to 4.6 for the Seahawks). Very poor play on first and second down (-58.6% DVOA) was saved by reasonable play on third down (28.4% DVOA).
But, like I said, their defense improved last night, going from -31.4% last week to -32.2% this week. So San Francisco, like New England, is still on the list of the best defenses ever tracked by DVOA through this point in the season. Now that both teams have had their bye weeks, we only need one of these tables:
| BEST DEFENSIVE DVOA THROUGH 9 GAMES, 1985-2019 |
|||
| Year | Team | W-L | DVOA |
| 2012 | CHI | 7-2 | -39.9% |
| 2002 | TB | 7-2 | -39.1% |
| 1991 | NO | 8-1 | -37.1% |
| 1991 | PHI | 4-5 | -35.5% |
| 2019 | NE | 8-1 | -33.7% |
| 2019 | SF | 8-1 | -32.2% |
| 1997 | SF | 8-1 | -32.2% |
| 1992 | PHI | 6-3 | -31.7% |
| 1993 | PIT | 6-3 | -29.9% |
| 1990 | CHI | 8-1 | -28.8% |
| 2015 | DEN | 7-2 | -27.4% |
| 1996 | GB | 8-1 | -27.2% |
| 1986 | CHI | 7-2 | -27.2% |
Of course, there's a lot of discussion of the Patriots and their super easy schedule in 2019, and whether they belong on this list given that they had their first below-average defensive game of the year when they played the first tough offense on their schedule. Our adjustments suggest that the Patriots still qualify as one of the best defenses ever because while they've beat up on bad offenses, they've beat up on them a lot worse than the rest of the league has. This is the week that our opponent adjustments reach full strength in the DVOA system, and you'll notice the Patriots still are on this list.
As good as the Patriots defense looks after opponent adjustments, it looks a lot better before opponent adjustments. The Patriots' VOA without opponent adjustments (and without making all fumbles equal) is currently -45.9%, which means their adjustment is over 12 percentage points! That -45.9% rating would easily be the best defense ever tracked if we compared it to all other unadjusted defenses. The only other defense to ever clear the -40% even without opponent adjustments was the 1991 Eagles, the best defense in DVOA history. The DVOA-VOA gap for the Patriots defense is almost three times as big as the DVOA-VOA gap for the 49ers defense.
Week 10 is a good point to review schedule strength numbers and the effect that adjustments are having on DVOA. For example, other advanced stats around the Web have the Baltimore Ravens now challenging the Patriots and 49ers at the top of the league after this week's huge dominant blowout of Cincinnati. DVOA does not, because the Baltimore Ravens also have a huge adjustment for opponent strength. The Ravens move up two spots to No. 5 this week but they remain below Kansas City at No. 3 and Dallas at No. 4. Kansas City stands out in particular because the Chiefs are the only team currently in the DVOA top five that doesn't also have one of the five easiest schedules in the league. Instead, Kansas City has had the fourth-hardest schedule so far based on average DVOA of opponent. Without adjustments, the Chiefs would be a distant fifth, trailing Dallas and Baltimore by over 10 percentage points and only slightly ahead of No. 6 Minnesota and No. 7 New Orleans.
Let's take a look here at the teams with the biggest difference between DVOA and VOA. Again, the VOA I'm using here not only removes opponent adjustment but also the adjustment that makes all fumbles equal and the weather/altitude adjustments for special teams.
Top 5 Teams Adjusted Downward
1) New England, from 52.6% (1) to 37.3% (1). Easiest schedule of the year so far.
2) Buffalo, from 3.9% (13) to -11.1% (24). No. 31 ranked schedule so far. The Bills actually move up one spot in DVOA this week, passing Cleveland, because DVOA believes they were the better team in their close loss to the Browns.
3) Tennessee, from 3.7% (14) to -7.1% (23). No. 23 ranked schedule so far, and second-best rate of fumble recovery (71%) behind only New Orleans (82%). Tennessee has 14 fumbles on offense and has recovered 11 of them.
If I may digress, this is also a good time to point out how well Ryan Tannehill is playing as Tennessee's starting quarterback. In Weeks 1-6, with Marcus Mariota behind center, the Titans ranked 29th in offensive DVOA at -18.2%. In Weeks 7-10, with Tannehill, the Titans have improved to seventh in offensive DVOA at 16.6%. Based on Tannehill's past performance, it's unlikely they'll stay quite this good for the rest of the year, but they're definitely better than they were with Mariota and that makes a playoff run more likely. This week, I made an adjustment to Tennessee's weighted DVOA in the playoff odds simulation to reflect a projection that only considers a preseason projection with Tannehill and the offensive DVOA since Week 7.
Back to the countdown...
4) Dallas, from 36.8% (2) to 26.7% (4). No. 30 ranked schedule so far, and third-best rate of fumble recovery (69%). Dallas has 10 fumbles against its defense and has recovered eight of them.
5) Baltimore, from 34.3% (4) to 24.7% (5). No. 28 ranked schedule so far. Their schedule against bad run defenses has been particularly easy, and of course Baltimore runs a lot so those plays are getting adjusted more.
Top 5 Teams Adjusted Upward
You'll notice these adjustments are smaller than the top downward adjustments.
1) Cincinnati, from -49.0% (31) to -42.1% (31). Hardest schedule of the year so far.
2) Green Bay, from 9.2% (10) to 15.9% (8). No. 2 ranked schedule so far.
3) Oakland, from -1.1% (18) to 5.4% (12). No. 3 ranked schedule so far.
4) Kansas City, from 22.3% (5) to 28.4% (3). No. 4 ranked schedule so far.
5) New York Giants, from -29.7% (28) to -24.4% (28). Only the No. 13 ranked schedule so far, but Giants have recovered only 36% of fumbles including just 6 of their mind-boggling total of 20 fumbles on offense. The worst fumble recovery rate, not including muffed kicks, belongs to the Atlanta Falcons at 29%. Atlanta has had nine fumbles on defense and all nine were recovered by the opponent.
A discussion of Cincinnati and the Giants brings us to a discussion of this year's worst teams, and that brings us to those tables I've been running every week of the worst DVOA through X games. Except we don't need as many of those tables this week. The New York Jets offense was good enough against the Giants that the Jets climbed off the list of the worst offenses in DVOA history. And the Miami defense was so good against Indianapolis that the Dolphins not only aren't the worst defense in DVOA history anymore but aren't even on the list of the worst defenses. They aren't even the worst defense of 2019 anymore; Cincinnati now is.
Since they rank 31st on both offense and defense with below-average special teams, Miami is still on the list of the worst teams in DVOA history. However, if the trend of the last few weeks holds then this is probably the last week they'll appear on this list.
| WORST TOTAL DVOA THROUGH 9 GAMES, 1985-2019 |
|||
| Year | Team | W-L | DVOA |
| 2005 | SF | 2-7 | -65.6% |
| 2013 | JAX | 1-8 | -62.3% |
| 1993 | TB | 2-7 | -62.0% |
| 1996 | STL | 2-7 | -56.5% |
| 2009 | DET | 1-8 | -56.4% |
| 2008 | STL | 2-7 | -53.1% |
| 2019 | MIA | 2-7 | -51.6% |
| 2009 | OAK | 2-7 | -51.2% |
| 2002 | HOU | 2-7 | -50.5% |
| 2007 | SF | 2-7 | -50.3% |
| 1991 | IND | 0-9 | -50.1% |
| 1999 | CIN | 1-8 | -50.0% |
The Dolphins' trend since their Week 5 bye is a steady move upwards, even though they have yet to have a single game with a positive overall DVOA. Every week is better than the week before it. To show you the trend, let's break out the first DVOA week-to-week graph of the 2019 season. This one is a little different, because I've graphed both Miami's DVOA and their VOA rating for each week so you can see the effect of opponent adjustments for playing teams ranging from New England on one hand to Washngton and the Jets on the other hand.
If you like this DVOA vs. VOA chart, let me know in the comments and I'll run a couple more of them in future weeks for teams such as the Patriots and 49ers.
* * * * *
Stats pages should now be updated through Week 10, including playoff odds, the FO Premium DVOA database and snap counts.
* * * * *
These are the Football Outsiders team efficiency ratings through 10 weeks of 2019, 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 to 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 represents an attempt to figure out how a team is playing right now, as opposed to over the season as a whole, by making recent games more important than earlier games.
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>
| RK | TEAM | TOTAL DVOA |
LAST WEEK |
WEI. DVOA |
RANK | W-L | OFF. DVOA |
OFF. RANK |
DEF. DVOA |
DEF. RANK |
S.T. DVOA |
S.T. RANK |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | NE | 37.3% | 2 | 34.9% | 1 | 8-1 | 3.6% | 11 | -33.7% | 1 | 0.0% | 18 |
| 2 | SF | 31.2% | 1 | 30.5% | 2 | 8-1 | -0.5% | 17 | -32.2% | 2 | -0.5% | 20 |
| 3 | KC | 28.4% | 3 | 26.7% | 4 | 6-4 | 26.7% | 2 | -0.8% | 17 | 0.9% | 12 |
| 4 | DAL | 26.7% | 4 | 27.3% | 3 | 5-4 | 29.6% | 1 | 1.4% | 20 | -1.4% | 25 |
| 5 | BAL | 24.7% | 7 | 23.8% | 5 | 7-2 | 20.2% | 3 | -1.2% | 14 | 3.4% | 5 |
| 6 | MIN | 22.0% | 6 | 22.7% | 6 | 7-3 | 14.3% | 7 | -7.1% | 8 | 0.6% | 13 |
| 7 | NO | 17.0% | 5 | 18.3% | 7 | 7-2 | 8.0% | 8 | -7.7% | 7 | 1.3% | 9 |
| 8 | GB | 15.9% | 8 | 15.3% | 8 | 8-2 | 16.0% | 6 | 0.6% | 19 | 0.5% | 15 |
| 9 | SEA | 14.3% | 9 | 14.6% | 9 | 8-2 | 19.1% | 4 | 2.5% | 22 | -2.3% | 28 |
| 10 | HOU | 9.1% | 10 | 9.3% | 10 | 6-3 | 7.1% | 9 | 0.3% | 18 | 2.2% | 7 |
| 11 | PHI | 7.1% | 13 | 8.3% | 11 | 5-4 | 2.7% | 13 | -5.0% | 9 | -0.6% | 21 |
| 12 | OAK | 5.4% | 15 | 6.2% | 12 | 5-4 | 17.2% | 5 | 12.3% | 29 | 0.5% | 16 |
| 13 | LAR | 3.5% | 11 | 1.8% | 15 | 5-4 | -5.4% | 21 | -8.3% | 4 | 0.5% | 14 |
| 14 | JAX | 2.9% | 14 | 3.1% | 13 | 4-5 | 0.3% | 14 | -1.8% | 12 | 0.9% | 11 |
| 15 | IND | 2.0% | 12 | 2.5% | 14 | 5-4 | -1.3% | 18 | -4.9% | 10 | -1.6% | 27 |
| 16 | DET | 0.4% | 17 | 0.8% | 16 | 3-5-1 | 3.3% | 12 | 7.0% | 25 | 4.2% | 2 |
| 17 | CHI | -0.3% | 20 | -1.3% | 20 | 4-5 | -12.2% | 26 | -8.3% | 5 | 3.6% | 4 |
| 18 | CAR | -1.2% | 16 | -0.2% | 17 | 5-4 | -3.2% | 19 | -0.9% | 15 | 1.1% | 10 |
| 19 | LAC | -1.9% | 19 | -0.9% | 19 | 4-6 | 6.7% | 10 | 7.5% | 26 | -1.1% | 22 |
| 20 | PIT | -2.0% | 21 | -0.5% | 18 | 5-4 | -14.5% | 28 | -12.4% | 3 | 0.2% | 17 |
| 21 | DEN | -4.8% | 22 | -4.2% | 21 | 3-6 | -8.0% | 23 | -7.9% | 6 | -4.8% | 32 |
| 22 | TB | -6.1% | 18 | -6.6% | 22 | 3-6 | -6.2% | 22 | 2.0% | 21 | 2.0% | 8 |
| 23 | TEN | -7.1% | 23 | -8.1% | 23 | 5-5 | -4.5% | 20 | -0.9% | 16 | -3.5% | 30 |
| 24 | BUF | -11.1% | 25 | -10.3% | 24 | 6-3 | -11.0% | 25 | -1.2% | 13 | -1.3% | 23 |
| 25 | CLE | -12.1% | 24 | -11.5% | 25 | 3-6 | -10.1% | 24 | 4.7% | 23 | 2.6% | 6 |
| 26 | ARI | -13.5% | 26 | -12.8% | 26 | 3-6-1 | 0.2% | 15 | 11.3% | 28 | -2.4% | 29 |
| 27 | ATL | -19.5% | 28 | -19.4% | 27 | 2-7 | -0.3% | 16 | 14.4% | 30 | -4.7% | 31 |
| 28 | NYG | -24.4% | 27 | -24.1% | 28 | 2-8 | -14.2% | 27 | 8.9% | 27 | -1.4% | 24 |
| 29 | WAS | -28.6% | 29 | -31.3% | 30 | 1-8 | -23.1% | 29 | 5.2% | 24 | -0.4% | 19 |
| 30 | NYJ | -30.8% | 31 | -31.1% | 29 | 2-7 | -36.8% | 32 | -2.0% | 11 | 4.1% | 3 |
| 31 | CIN | -42.1% | 30 | -42.4% | 31 | 0-9 | -24.1% | 30 | 23.2% | 32 | 5.2% | 1 |
| 32 | MIA | -51.6% | 32 | -48.9% | 32 | 2-7 | -28.8% | 31 | 21.3% | 31 | -1.6% | 26 |
- NON-ADJUSTED TOTAL DVOA does not include the adjustments for opponent strength or the adjustments for weather and altitude in special teams, and only penalizes offenses for lost fumbles rather than all fumbles.
- ESTIMATED WINS uses a statistic known as "Forest Index" that emphasizes consistency as well as DVOA in the most important specific situations: red zone defense, first quarter offense, and performance in the second half when the score is close. It then projects a number of wins adjusted to a league-average schedule and a league-average rate of recovering fumbles. Teams that have had their bye week are projected as if they had played one game per week.
- PAST SCHEDULE lists average DVOA of opponents played this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
- FUTURE SCHEDULE lists average DVOA of opponents still left to play this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
- VARIANCE measures the statistical variance of the team's weekly DVOA performance. Teams are ranked from most consistent (#1, lowest variance) to least consistent (#32, highest variance).
| RK | TEAM | TOTAL DVOA |
W-L | NON-ADJ TOT VOA |
ESTIM. WINS |
RANK | PAST SCHED |
RANK | FUTURE SCHED |
RANK | VAR. | RANK |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | NE | 37.3% | 8-1 | 52.6% | 8.4 | 1 | -18.5% | 32 | -4.8% | 24 | 12.8% | 23 |
| 2 | SF | 31.2% | 8-1 | 36.0% | 8.1 | 3 | -9.8% | 29 | 6.0% | 8 | 17.7% | 31 |
| 3 | KC | 28.4% | 6-4 | 22.3% | 7.1 | 5 | 7.0% | 4 | 5.6% | 9 | 11.3% | 20 |
| 4 | DAL | 26.7% | 5-4 | 36.8% | 8.2 | 2 | -10.9% | 30 | 1.2% | 15 | 6.4% | 7 |
| 5 | BAL | 24.7% | 7-2 | 34.3% | 6.9 | 6 | -9.3% | 28 | -1.8% | 21 | 11.9% | 21 |
| 6 | MIN | 22.0% | 7-3 | 20.1% | 7.9 | 4 | 1.1% | 19 | 3.9% | 12 | 7.7% | 9 |
| 7 | NO | 17.0% | 7-2 | 20.9% | 6.2 | 10 | 1.9% | 16 | -0.3% | 18 | 10.3% | 15 |
| 8 | GB | 15.9% | 8-2 | 9.2% | 6.7 | 7 | 8.2% | 2 | 0.1% | 16 | 17.6% | 30 |
| 9 | SEA | 14.3% | 8-2 | 14.5% | 6.7 | 8 | -1.9% | 21 | 8.2% | 4 | 8.3% | 11 |
| 10 | HOU | 9.1% | 6-3 | 9.9% | 5.8 | 12 | 4.0% | 9 | 5.6% | 10 | 4.6% | 4 |
| 11 | PHI | 7.1% | 5-4 | 7.6% | 6.4 | 9 | -2.8% | 25 | -7.2% | 25 | 18.9% | 32 |
| 12 | OAK | 5.4% | 5-4 | -1.1% | 6.2 | 11 | 7.9% | 3 | -7.9% | 26 | 13.5% | 24 |
| 13 | LAR | 3.5% | 5-4 | 6.9% | 5.4 | 15 | -2.3% | 22 | 10.0% | 3 | 14.3% | 26 |
| 14 | JAX | 2.9% | 4-5 | -1.2% | 5.5 | 13 | -2.5% | 24 | -3.6% | 23 | 4.9% | 5 |
| 15 | IND | 2.0% | 5-4 | 2.5% | 5.4 | 14 | -4.9% | 27 | 2.5% | 14 | 10.6% | 17 |
| 16 | DET | 0.4% | 3-5-1 | -1.9% | 5.2 | 16 | 4.3% | 7 | 3.5% | 13 | 4.6% | 3 |
| 17 | CHI | -0.3% | 4-5 | -4.9% | 4.9 | 17 | 3.6% | 11 | 10.4% | 2 | 10.4% | 16 |
| 18 | CAR | -1.2% | 5-4 | -5.0% | 4.8 | 18 | 3.3% | 12 | -2.5% | 22 | 11.1% | 19 |
| 19 | LAC | -1.9% | 4-6 | 0.4% | 4.1 | 22 | -3.3% | 26 | 13.7% | 1 | 8.4% | 12 |
| 20 | PIT | -2.0% | 5-4 | 0.0% | 4.8 | 19 | 1.9% | 15 | -13.9% | 30 | 5.8% | 6 |
| 21 | DEN | -4.8% | 3-6 | -2.3% | 4.7 | 20 | 3.7% | 10 | 7.5% | 6 | 8.3% | 10 |
| 22 | TB | -6.1% | 3-6 | -3.4% | 3.8 | 24 | 2.1% | 14 | -1.1% | 20 | 3.9% | 1 |
| 23 | TEN | -7.1% | 5-5 | 3.7% | 4.4 | 21 | -2.3% | 23 | 7.6% | 5 | 13.8% | 25 |
| 24 | BUF | -11.1% | 6-3 | 3.9% | 3.7 | 25 | -16.9% | 31 | -0.1% | 17 | 3.9% | 2 |
| 25 | CLE | -12.1% | 3-6 | -14.8% | 3.9 | 23 | 6.4% | 5 | -18.4% | 32 | 15.7% | 28 |
| 26 | ARI | -13.5% | 3-6-1 | -14.0% | 3.7 | 26 | -0.6% | 20 | 6.4% | 7 | 7.3% | 8 |
| 27 | ATL | -19.5% | 2-7 | -23.5% | 3.1 | 28 | 6.0% | 6 | 5.2% | 11 | 10.6% | 18 |
| 28 | NYG | -24.4% | 2-8 | -29.7% | 1.3 | 29 | 2.3% | 13 | -8.4% | 27 | 12.7% | 22 |
| 29 | WAS | -28.6% | 1-8 | -29.8% | 1.3 | 30 | 4.1% | 8 | -0.9% | 19 | 14.8% | 27 |
| 30 | NYJ | -30.8% | 2-7 | -32.5% | 3.2 | 27 | 1.4% | 18 | -15.0% | 31 | 16.9% | 29 |
| 31 | CIN | -42.1% | 0-9 | -49.0% | 1.0 | 31 | 8.3% | 1 | -9.4% | 28 | 9.5% | 13 |
| 32 | MIA | -51.6% | 2-7 | -51.4% | 0.0 | 32 | 1.8% | 17 | -10.9% | 29 | 10.1% | 14 |
Comments
120 comments, Last at 15 Nov 2019, 12:38pm
#3 by theslothook // Nov 12, 2019 - 5:38pm
Ok, no one has discussed them, but has anyone watched the Bengals closely? I sort of miss Rob Weintraub's commentary on the Bengals as they are largely ignored by the posters on this site.
I ask because depending on the responses, it informs what their team building strategy ought to be. Its clearly more than the quarterback and more than the coach since they went on an abysmal losing streak a year ago.
#11 by PackFan // Nov 12, 2019 - 6:45pm
I've been curious about other measures of toughest schedules for a while. I'm not sure that averaging it really captures it.
I would think that once a team is historically bad, if they get 10% worse, my team's odds of winning aren't helped out as much as if an average team drops 10%. Yet comparing strength of schedule by examining the average of opponents' DVOA would count these changes equally.
Would it be possible to run something like the Forrest Index in reverse? I'm thinking you could take the stats for an average team and running it up against each team's schedule and then counting the expected number of wins.
#23 by Will Allen // Nov 12, 2019 - 9:26pm
That's kind of what I'm driving at; I think different kinds of teams would benefit from different kinds of opponent strength distributions, and I wonder how that could be captured mathematically.
Just for ease of illustration, assume there are no divisions, and each team has 16 opponents every year, and the 32 teams are evenly distributed along the DVOA curve. Now imagine the most perfectly average team ever. To maximize regular.season wins, would Team Average rather play a schedule consisting of A) Teams 1-8 and Teams 25-32 or B)Teams 8-15 and Teams 17-24? Now ask the same question of the best team and worst team? Should they have a preference for a schedule at the extremes, or one bunched in the middle? Would it make any difference?
My intuition is that Team Average and Team Awful should prefer a schedule of extremes, and that Team Wonderful should prefer a schedule of mediocrities, and I know the answer may change depending on the spacing between each team, but my sense of it is also that average or median DVOA isn't capturing the nuance of schedule strength.
#57 by Pat // Nov 13, 2019 - 9:49am
" but my sense of it is also that average or median DVOA isn't capturing the nuance of schedule strength."
Yeah, but there's no statistic that would. You can't combine 14 pieces of information (the team + their opponents) into 1 number without losing information.
#72 by Richie // Nov 13, 2019 - 12:17pm
It's something I have thought a little bit about, but haven't gotten very far. Since game results are binary, schedule strength should reflect that somehow.
For instance, if you are a slightly above average team and play 4 games and the opponents have extreme DVOA of 26, 11, -19 and -24, your average opponent strength is -1.5. Which would indicate that the team should maybe be 4-0 since their average opponent is worse than them. But in fact, it would be more realistic for the team to be 2-2, since two of their opponents were clearly better and two were clearly worse.
Maybe a summation of the probabilities of an average team beating a team with the DVOA of each of your opponents?
#74 by TGT // Nov 13, 2019 - 12:47pm
I don't know about a summation, but an "average record against schedule" would be great. Even better would be 3 columns using 0 DVOA for the team, and +/- 15% DVOA (or, say, whatever is 8th and 24th that year). I think that's a nice middle ground between data dump and single average.
#12 by Richie // Nov 12, 2019 - 6:47pm
Good point. Prior to the New England game, Baltimore's offense had a 12 DVOA. After New England they were at 16 DVOA. Then after Cincinnati they were at 20 DVOA. Scoring 30 against New England had a similar effect on their DVOA to scoring 42 on Cincinnati.
#14 by ssereb // Nov 12, 2019 - 7:10pm
They went from 12.8% OFF DVOA immediately before beating the Patriots to 16.6% immediately after, which might not seem huge but opponent adjustments were increasing at the same time, discounting the Ravens' blowout of the Dolphins and emphasizing fairly unimpressive offensive performances against the Bengals, Seahawks, and Cardinals.
EDIT: The Ravens still have games against the defenses currently ranked 2, 3. and 4 (SF, PIT, and LAR) so they'll have every opportunity to prove they're the offense I think they are, having already beaten the defenses ranked 1 and 3.
#100 by Aaron Schatz // Nov 13, 2019 - 4:12pm
I believe a lot of the issue here has to do with changes in opponent adjustments and in the gradual weighting of opponent adjustments.
Baltimore had 53.2% offensive DVOA against New England, its best game of the year and one of only 10 single games with offensive DVOA above 50% this season.
Baltimore had 42.1% offensive DVOA against Cincinnati.
#16 by Q // Nov 12, 2019 - 7:42pm
There is likely no way to definitely prove this but I suspect that the NE Defense as constituted is set up to absolutely demolish bad Offenses. BB is legendary for showing tricky/confusing looks that absolutely baffle young qbs (much less bad young qbs). Furthermore, the lack of huge glaring weaknesses in the secondary means they do not get easy reads on who to target.
However, when faced with Qb's with an actual pulse (even more so if they actually have talent) all the looks that confound the young/bad qbs are nothing more than window dressing. Yes, their defense is still decently good but far, far removed from any historic context.
My continued prediction is that for the next 4 Weeks their defense will look quite mortal and people will begin wondering what has happened to their defense. Then, they will absolutely destroy the Bengals, Dolphins, Bills at the end of the year where they basically get as many INT's as they allow TD's followed by everyone proclaiming that the D is back and that the Master BB has them peaking for the Playoffs.
#18 by theslothook // Nov 12, 2019 - 8:08pm
Why is this your view? What about this defense feels like a mirage to be exposed by savvier quarterbacks? Why can't it be that this defense had a bad day against a tough opponent with an unorthodox offense?
And since that loss, there's been an avalanche of revisionism. Now, the Patriots are not only not the best defense ever(not even close), they're not even one of the best in the nfl? Maybe they might even be just good to ok?
I won't accuse you of that line of thinking, but its just amazing to me how much we overreact to one game. Should we btw be burying the saints today?
#22 by dank067 // Nov 12, 2019 - 9:19pm
Here's one thing I'm going to try to watch for with New England. A lot of film breakdowns that I've seen have pointed out that their execution has been near-perfect: everyone is playing their assignments correctly, the secondary is doing a phenomenal job of passing off coverages, etc.
Their next four games (the Eagles perhaps a little less so than the other three) come against teams who have dominant skill position players and quarterbacks with excellent improvisation ability. The Pats are going to have to tilt their gameplan towards stopping certain players and hold up in coverage on extended plays. Gilmore can certainly cover anyone in the league, can the rest of the defense as a whole keep things locked down? I'm really looking forward!
#24 by Q // Nov 12, 2019 - 9:33pm
This was my view before the Ravens game. The Ravens are their own unique subset since they run an Offense pretty much different from any other team.
I saw many people suggesting even before Balt that SF might have had a better Defense. There is nothing revisionist about that. SF's Defense is built upon High Draft Pick after High Draft Pack. Historically, Dominant Defenses have generally been comprised of Numerous High Picks/All Pro Caliber Players/Dominant Front 7's. SF matches this template far more than does NE. We also know that BB is a Defensive genius. It isn't going out on a limb to suggest that his wizardry and tricks are much more effective vs new/young qbs than others.
The last 2 good QBs NE faced put up 30+ on them (Mahomes in the Title Game and Lamar this year). The Saints shouldn't be buried but their Offense has been concerning for a while now.
If NE is some type of historic Defense then let's see them hold the Cowboys to 10, the Texans to 7, the Chiefs to 13 or 14. My guess those teams combine to put up 70-80 points against NE.
#52 by Scott P. // Nov 13, 2019 - 8:47am
This is revisionist history -- Sean McVey's offense was considered world-beating until the Super Bowl, and Mahomes was shut out in the first half, so I see it as shutting down the highest-powered offenses 50% of the time, which is pretty decent, in my book.
#53 by Boots Day // Nov 13, 2019 - 8:59am
If you want to give the Pats defense credit for shutting out Mahomes for a half, then you also have to ding them for allowing the Chiefs to score at a 60-plus points-per-game rate in the second half. Great defenses don't give up 60-point games!
You can't only choose those halves of games where your theory looks good.
#60 by Will Allen // Nov 13, 2019 - 10:07am
I sure didn't see the McVay offense as world beating until the Super Bowl, and I wasn't alone. There were more than a few of us who saw Goff as a significantly flawed qb who had his flaws obscured very efficiently from the beginning of the season, but the obscuring schemes were becoming less and less effective as the season wore on. To the point that some of us were writing that the Rams offense wouldn't do anything against The Dark Lord's defense.
#63 by Pat // Nov 13, 2019 - 10:35am
There was plenty of evidence that the Rams offense was cooling down significantly before the Super Bowl, too. Goff's performance specifically was called out as having cooled off a ton, from 105 DYAR/game from weeks 1-11 down to -4.3 DYAR/game for weeks 13-20.
The whole "the Patriots shut down this world-beating offense!" is really very revisionist. The Rams passing offense specifically had basically ground to a halt by the Super Bowl.
#80 by Pat // Nov 13, 2019 - 1:43pm
That's due to the early-season success. Rams scored 35.4 ppg from week 1-11 (a 566-point per season pace) and 27.7 ppg the remainder of the season up to the Super Bowl (a 443-point pace). Excluding the playoffs it's basically the same (27.6 ppg instead of 27.7 ppg). Weighted DVOA exists for a reason: by the Super Bowl the Rams were a worse offense than the Patriots, even though the full-year Patriots were over 10% DVOA lower.
This isn't to say that holding them to 3 points isn't damn impressive. Aaron doesn't do "single-game DVOA using weighted DVOA opponent corrections" but if he did it would probably be something like -40% defensive DVOA for that game. The single game DVOA using full-season corrections was something like -50%.
#90 by Noahrk // Nov 13, 2019 - 3:10pm
I think the original idea is interesting, that a team (offense or defense) might be relatively better against tougher or weaker opponents. Kind of like what the Steelers are perceived as, playing better against better opponents, and worse against lesser ones. It would be great if it could be measured or tested, but alas.
Ermm, I think I replied to the wrong comment. Anyway, I meant the OPs comment.
#67 by Aaron Brooks G… // Nov 13, 2019 - 11:40am
Until the NFC Championship game, anyway. The Saints bottled them up pretty well and Goff looked terrified the entire game. That all repeated in the Super Bowl, which is why I didn't even bother watching it.
#62 by Pat // Nov 13, 2019 - 10:20am
Keep in mind you're jumping a bit in context here: he called it a "good defense" and now you're saying "what [..] feels like a mirage to be exposed"? You're making it sound as if he's saying there's no substance to the defense, which isn't what he's saying.
So if we rephrase what you're saying to "what about this defense feels like they're overperforming relative to how they'd do later," then the answer was just posted by Aaron the other day: interceptions.
They're way overperforming on interceptions. They have 19 interceptions, leading the league by a lot, but they're projected to only be in the middle-of-the-pack for the remainder of the year. San Francisco is too, as a note, although it's not as dramatic. By the projections, Pittsburgh will close the gap in interceptions with them significantly by the end of the year (24.7 projected vs 21.8 projected, versus the current 19 to 14).
#25 by Anon Ymous // Nov 12, 2019 - 9:38pm
Odd that BB would be "notorious" for something he hadn't done in over a decade. Go to any Pats message board circa 2018 and a long-standing complaint would be how NE's vanilla defenses made inexperienced and mediocre QBs look decidedly capable. Instead of sending an array of dizzying looks, NE would play it straight up and let the QBs make their own mistakes. The whole "Bill always boggles young QBs" rhetoric is more myth than fact.
This doesn't mean the criticism isn't true of this year, of course. Just know the foundation of your argument isn't as sound as you think.
#28 by Q // Nov 12, 2019 - 9:48pm
The fact that every time NE plays they show the stat the BB had won like 20 in a row against 1st and 2nd String QBs seems to suggest that they are doing pretty well against them. When I have watched Pats games in the past, usually they have played more Vanilla vs better Qbs and less Vanilla vs young/bad ones.
#30 by Anon Ymous // Nov 12, 2019 - 9:56pm
I haven't a clue what games you have watched prior to this year, but what you espouse is the exact opposite of how the team has actually played.
A couple notes about the winning streak:
1. Those are home games, where NE wins a lot anyway. They lost to Big Ben, Sanchez and Colt McCoy on the road. (Perhaps more, those are off the top of my head.)
2. Those are *team* wins that don't have much to do with unusually flummoxed QBs.
#31 by Q // Nov 12, 2019 - 10:01pm
We'll see how Other Worldly NE looks in the next 4 games. It is hard to call a Defense Great, much less Historic if every time they face a Top Qb they give up 30. If they hold DAL to 3 and KC to 10 I'll openly proclaim them as Great. On the other hand, if they are giving up 25-35 then that should pretty much end all conversation about them.
#34 by Q // Nov 12, 2019 - 10:13pm
I also disagree with your claim that NE throughout the years plays as Vanilla against Rookie QB A as they do against Aaron Rodgers or more veteran QBs. You are basically trying to argue that BB through the years does put his Defenses in the best possible position since he has tons of things he can do to trick inexperienced/bad qbs that would be much less effective vs more experienced/better QBs.
#35 by Anon Ymous // Nov 12, 2019 - 10:19pm
I only speak the facts, you can take from that what you will.
Edit: As far as Rodgers goes, NE consistently uses containment rush tactics against mobile QBs. This is regardless of their skill level; they try just as hard to keep Josh Allen in the pocket as they do Russell Wilson. Perhaps this is the source of your confusion?
#78 by sbond101 // Nov 13, 2019 - 1:19pm
I really think the argument that DVOA is not properly capturing just how uniquely awful this year's slate of bad teams are should be losing steam at this point. Yes this years DVOA has a distribution with a cluster at the bottom which is somewhat unusual but the teams at the bottom of that chart have shown distinct signs of life compared to the worst teams of all time - I think it is currently much much more difficult to sustain that argument now that the Jets & Dolphins both have wins over non-awful teams (the Bengals being clearly not on the scale of historically laughable based on their close loses to non-awful teams). This year is looking a lot more like a typical year on the bottom end of the distribution - which really undermines the currency of the arguments against the predicative value of the defensive DVOA of the 49ers and the Pats.
#19 by oaktoon // Nov 12, 2019 - 8:44pm
They currently trail two teams that they defeated, and which have worse records-- DALL and MINN.
A win over the Niners in two weeks will leave the two teams tied at 9-2-- but GB holding the tiebreaker-- so at that point the Packers could have the most anti-DVOA record in many a year.
Who knows? Who knows anything about the NFC? DVOA has been grossly overrating SF from the get-go. I would say 12-4 is the likeliest final record for them, which means a 4-4 second half. All the Cowboys do is lose to teams with winning records, but DVOA must assume that they have a different coach and are actually winning those games. The Vikings might just be for real, however.
#26 by Will Allen // Nov 12, 2019 - 9:41pm
For purposes of ranking teams in a discrete season, one really shouldn't invest too much meaning in a data point labeled "win" or "loss" especially when that win or loss was decided by 6 or fewer points. Especially when each team only has 16 such data points.
If you insist on ignoring the largest sample sizes available, which is play-by-play, or drive results, looking at schedule strength and point differential is probably more useful.
#27 by Q // Nov 12, 2019 - 9:42pm
Green Bay has been penalized by DVOA for playing a really hard schedule. Unlike DAL and NE they have not had the opportunity for Huge DVOA destruction games vs the Dolphins, Bengals, etc. I understand about opponent adjustments, but the fact that 4 of the Top 5 teams in DVOA have 4 of the 5 easiest schedules in the League seems to suggest that perhaps the DVOA adjustment penalty is not big enough (at least this year).
The seeming not big enough opponent adjustments has NE and SF looking like Other Worldly Defenses.
In regards to GB, I do think they are seriously flawed ( I don't trust any part of their team). However, they have quite possibly the best resume in the entire league so far.
#29 by Anon Ymous // Nov 12, 2019 - 9:50pm
Let's pretend for a minute that DVOA isn't capturing just how terrible NE's opposing Os were - and given how many were terrible and beat up, I'm willing to grant that. What does this have to do with Green Bay? Are you saying their opponents have been historically great to the point that DVOA isn't adequately capturing their difficulty? If you aren't making that claim, what claim are you making? Why else would a statistic designed to account for opponent strength not recognize it in GB's schedule?
#40 by Perfundle // Nov 12, 2019 - 11:49pm
He isn't saying that Green Bay's opponents have been historically great, but rather that they haven't played the historically awful ones that allow them to pad their numbers beyond what opponent adjustments can account for.
#42 by Anon Ymous // Nov 13, 2019 - 12:16am
He was responding to a comment that specified where GB stands in relation to Dallas and, more importantly, Minnesota. In this context a GB "penalty" must be more than you describe. If all Q meant was that a few other teams have potentially inflated DVOAs due to feasting on lousy competition - a point that isn't lost on me or anyone else at this site - then it wasn't well elucidated.
#45 by Perfundle // Nov 13, 2019 - 1:35am
Well I understood his point fine. Also, I'm not sure why you need his point to apply to Minnesota, when he specifically singled out Dallas and New England, as well as SF and Baltimore in terms of the teams with the easiest schedules, seeing as oaktoon also mentioned where GB stands in relation to SF. Also, oaktoon finished with a caveat that Minnesota might be ranked correctly.
This isn't to say that I really agree with either of them. I think GB is ranked where they are because they didn't beat the bad teams on their schedule by what a top-5 team should have. Ditto NO and Seattle, which is why all three are clustered below several teams with worse records.
#55 by Cheesehead_Canuck // Nov 13, 2019 - 9:45am
I'm not going to nitpick their place in DVOA rankings, but what bad teams have GB even played?? The "worst" team they've played is probably the Chargers (which ironically was their worst game as well), and the Chargers are better than all but one of NE's opponents so far. That's a hella tough schedule they've played through 10 weeks and the only break they get the rest of the way are back to back WAS-NYG games.
#95 by theslothook // Nov 13, 2019 - 3:48pm
We need to define bad team. I know Detroit's record is bad, but theyve been in all of these games and had chances to win a lot of them too. Just as there are teams that win nail biters in a season to go 11-5, there are teams that fall on the wrong side of things.
Detroit strikes me as a decent team. I can probably name 10 teams that are worse than Detroit
#110 by scraps // Nov 13, 2019 - 8:37pm
You say "Ditto NO and Seattle" who "didn't beat the bad teams on their schedule" for why the Seahawks are ranked relatively low. Now, I'm not complaining (as a Seahawks fan); they are ranked pretty much where they deserve to be ranked (though as a road team they are 5-0, which is nice). But the two teams they lost to were the Ravens and the Saints! I assume the Seahawks are ranked relatively low not because they didn't beat the bad teams, but because they beat the bad teams (Bengals, Bucs, Browns, Falcons) by less than they should have.
#68 by Aaron Brooks G… // Nov 13, 2019 - 11:43am
The argument is that DVOA only partially accounts for schedule difficulty. That is, you would expect DVOA to over-value teams with comparatively easy schedules and under-value teams with comparatively hard ones.
Green Bay has had a comparatively hard schedule, thus, the argument is that they are better than their DVOA level.
#92 by Q // Nov 13, 2019 - 3:31pm
Something being the most predictive historically doesn't mean that is true for this year.
You can just hand wave something off as random noise if you want. However, it is very reasonable for people to observe than 4 out of the Top 5 teams in DVOA have had 4 out of the 5 Easiest Schedules this year and to wonder if this really is just random chance or not.
#37 by jh_eldred // Nov 12, 2019 - 10:56pm
Green Bay is adequately ranked. Their special team and defense aren't very good, and they have poor receiving talent which hinders their offense at times. Winning one-score games the way they have isn't sustainable. And they just got trucked by a mediocre Chargers team. They're not terrible by any means, but at no point this year have I watched them and felt like they were a juggernaut.
#41 by oaktoon // Nov 13, 2019 - 12:11am
Just that they beat the crap out of a 5-4 Dallas team for 2 1/2 quarters, and also beat a 7-3 MINN team... and they are 8-2 and ranked below both. I'm just saying-- and I understand "sample sizes" of plays and drives and turnovers and all the rest-- that DVOA or no DVOA, odds are they might get a home game or two at Lambeau in January. (Yes-- NYG 2007/2011 and SF 2013 for games when that didn't work out so good)
#43 by Anon Ymous // Nov 13, 2019 - 12:21am
Considering that playoff seeding is determined by wins, I doubt you'll find anyone who disputes this. But that really has nothing to do with a discussion of GB's DVOA.
I must ask.... this is nothing new, DVOA has viewed GB differently than you for several weeks now, and the reasons for the discrepancy haven't changed. So why do you keep harping on it?
#113 by oaktoon // Nov 13, 2019 - 11:48pm
Team A defeats Team B and Team C-- one of them on road, handily...
Team A has a better record than Team B and Team C...
DVOA says, OTOH.... oh come on, where's the humanity here? Of course it is at the very least quirky, interesting, perhaps humorous-- maybe correct--- but dang it's a fun one, isn't it... And those Ws are a heck of a lot more important in how this season will play out than anything DVOA says. Rodgers will surely be claiming "Dang, why aren't we as good as the Vikings/Cowboys/Niners?" when one or more of them comes to Lambeau this January. Or maybe the Packers will go 10-6, end up as the wild card, and lose to DALL/MINN in the first round. Entirely possible as well.
I watch the games. I don't know who the best team is in the NFC. I don't think DVOA does either. Is that so awful?
#69 by Aaron Brooks G… // Nov 13, 2019 - 11:51am
I used 2000 as a default starting place.
Kansas City and Pittsburgh also have five home playoff losses. Arrowhead is old (1968) but Heinz Field is new (2001).
Next are Cincinnati, Chicago, and Philadelphia. Soldier Field is old (1922), but the UFO is new (2003). The Linc (2003) and Paul Brown Field (2000) are new.
I'm not sure age matters.
#71 by Will Allen // Nov 13, 2019 - 12:13pm
I'm kind of surprised that the acoustic properties of each stadium is not data that has been carefully observed and published widely as of yet. We know the Seahawks' stadium was specifically designed to accentuate crowd noise, but I'd think this stuff would have been carefully measured across the league, and disseminated to the public by now. Did Lambeau and Soldier have their noise levels change with renovations? I'd love to know for sure. I do know that the University of Michigan's 100,000 fans, in their classic bowl configuration, has always been strangely quiet, while, for instance, the mere 50,000 that can attend a University of Minnesota game, in their 7 year old stadium were so loud against Penn St. last Saturday that James Franklin had a tough time using his headset on the sidelines.
#97 by Sixknots // Nov 13, 2019 - 3:54pm
We know the Seahawks' stadium was specifically designed to accentuate crowd noise
Yeah, I went to a game without ear plugs and I'll not go to another game there without them. The noise was truly painful and the ride home in the car was strangely quiet. I couldn't hear fully until the next morning.
#47 by theslothook // Nov 13, 2019 - 2:41am
You keep questioning DVOA every week and every week you seem to be ignoring the answers. Your default position seems to revolve around wins( the very reason DVOA was invented), which places your whole arguments in the land of answers in search of a question.
I am even more astounded at the assumptive leaps you have made. Assume GB beats SF, what will its DVOA be? SF is likely headed for 4-4 as if your hand waving assumes it is all pre-ordained. Tell me, did you put money on those games and if so, which one's so that we may extoll you on your extraordinary foresight. Meanwhile, Gb losing to a plucky(but hardly dominant) Philly team at home and a blowout at the hands of a bad Chargers team are seemingly dismissed as usual any given Sunday events. And that near loss to Detroit and allowing the Cowboys to come back are both barely relegated to mere footnotes in your GB summary.
I get it. You are a fan of GB and clearly feel that dvoa is a flawed stat because your preferred team seems unfairly ranked. But rather than see GB's ranking as evidence of a flawed system, maybe you ought to consider why DVOA speaks unfavorably for GB relative to its record. Once you do so, you learn not to treat DVOA as a be all and end all but rather a good sanity check for your own preconceptions.
Finally - I saw no similar rancor out of you when GB lost convincingly to a mediocre(to bad?) Chargers team. Maybe this wouldn't come off as obvious fan whining if you're timing wasn't always so convenient.
#112 by oaktoon // Nov 13, 2019 - 11:40pm
Any Packer fan can see that a) our team has flaws and b) Despite those flaws, they have a halfway decent chance of getting a #1 or #2 seed. Sample size does not determine record or home field-- and the only quirk I am noting is that we have a metric which concludes that despite Team A defeating Teams B and C-- and having a better record than each-- and now the chance, however small, that there could be a Team D that they also defeat and enjoy a tiebreaker over for home field-- and yet will also almost still trail by this metric.
I find it curious. Not necessarily wrong, as I've said before. But I have grown somewhat tired-- to change sports for a second-- of a bunch of analysts concluding (HOU or NYY or LA) are the best teams, but if they lose in the playoffs, it's just random you-know-what again so who gives a hoot?
I mean we now have a separate rating on this site which claims that Alabama is better than LSU, and that Wisconsin-- losers to Ohio State (easily) and Illinois (somehow)-- is the #5 team in the country. I live in Madison. I don't beleive that for a second.
Every system has flaws. None are perfect. But I will challenge those who start to hoist any of them as "the it"-- did anyone in their right minds truly believe the 49ers-- or the Patriots with a 42 yr old Brady, a weak OL, and no Gronk-- were a historically great team? Of course not. One of them might still win a championship. But so might BAL, KC, NO, GB, SEA. MINN, DALL and maybe even one or two other teams not on the radar yet.
Play the games. We shall see... What i would love to see is someone doing a thorough analysis-- based on games played after some reasonable passage of time-- maybe 4-5 weeks (and forget week 17 for all the problems associated with whether teams even care about winning then)-- to see of Vegas, DVOA, W/L record or some other metric is a better predictor of outcomes. If DVOA outshines all the rest, I would gladly bow to that. My sense is it is one of several ways to determine just how good these teams are.
And as for the Packers-- if Rodgers spots one wide open receiver they probably beat Philly, and if one GB defender doesn't jump MCCaffrey they might lose to Carolina. The record says they split those two games-- whatever DVOA thinks. And that's fine.
But of course a quirk like DALL/MINN ranked ahead of GB despite both their comparative records plus the result of their head-to-head games is worth discussing. We can't just ignore them and say "DVOA says GB isn't as good". Maybe. Maybe not.
#115 by theslothook // Nov 14, 2019 - 11:02am
If I understand your argument, you seem to think Dvoa should be predicting wins within a season. And therefore the correct seeding and ultimately the super bowl participants and the winner.
And as the season progresses, it should be updating correctly based on the outcomes of the past, hence your using of Dal and Min along with GBs 8-2 record.
That seems sensible but it's not what DVOA is doing. It's meant to try and capture a team's intrinsic value, which is defined by how predictive a team's quality is year-to-year. The logic goes like this... If a team has intrinsic quality it should show up over time. that explains why Minnesota and Dallas are ranked ahead, something about their performances are more predictive overtime than Green Bay's.
Thus, the discussion really revolves around which type of model more accurately captures a team's intrinsic quality.
You might think on the surface that the answer is obvious...optimize on wins but that to me a mistake. As discussed before, wins are fraught by sample size and variation in play.
Just look at Green Bay, it's play on offense and defense has been all over the place. When both units play well they win handily. When each unit plays well and badly at the same time, you get uneven games that they might lose. And when both play bad they get blown out. And it doesn't take much of an imagination to see Green Bay having a much worse record if certain bounces didn't go their way. And that's the point, when you're using wins as your primary metric, it becomes extremely hard to disentangle.
And besides, I dont think super bowl winners are necessarily expected to be at the top of DVOA. That's sort of a happy result. The super bowl is much more about which team plays well in four games.
#32 by Raiderfan // Nov 12, 2019 - 10:06pm
For me, It presents more clearly than a thousand words how “eye tests” and counting stats (interceptions, TDs) can obfuscate the “ground truth” Of capability—which I think DVOA does better than any other single stat.
#46 by Perfundle // Nov 13, 2019 - 1:43am
Well this page shows 2007 NE's weekly offensive and defensive DVOA ratings:
https://www.footballoutsiders.com/game-previews/2008/super-bowl-xlii-preview
Assuming no extreme ST ratings, it looks like they had positive DVOA in every single game, and we know what happened in the playoffs.
#101 by Aaron Schatz // Nov 13, 2019 - 4:17pm
That was an older version of DVOA. Based on the current version, the 2007 Patriots had negative DVOA twice: Week 13 (Ravens, 27-24) and Week 15 (Jets, 20-10).
There is a team that made it through an entire year without a negative DVOA for a single game, and that's the best team in DVOA history, the 1991 Washington Redskins. Amazingly, they had 49.9% DVOA for their Week 16 game against Philadelphia where they sat their starters in the second half. Their lowest rating all year was 19.5% for Week 9 (17-13 over the Giants).
#104 by jheidelberg // Nov 13, 2019 - 5:03pm
I figured that the answer to my original question would be no, that no one had positive DVOA for all 16 games or negative DVOA for all 16 games (the Dolphins don't realize this opportunity to accomplish the feat this year, yet win at least 2 games). DVOA tanking does not get you the number one pick.
As for the 1991 Redskins, 19.5 % as their worst game, amazing! Only 6 teams this year have a DVOA better than 19.5%. I see that only 2 other teams in 1991 had DVOA better than 19.5% and one of those teams was at 19.6%. Am I reading this correctly? That Redskins team is beyond belief.
So I see that the 1991 Redskins won the Super Bowl. Did they do the incredible feat of have positive DVOA in all playoff games to make it a 19-0 DVOA season?
#38 by RickD // Nov 12, 2019 - 11:19pm
"but Giants have recovered only 36% of fumbles including just 6 of their mind-boggling total of 20 fumbles on offense. " Yeah, well, the box score says Daniel Jones "fumbled" and Jamal Adams recovered it, but really, Adams just picked his pocket. Let's not pretend the Giants were "unlucky" on that one. Jones personally is credited with 13 of the Giants' fumbles this season, of which the Giants have recovered 4. His fumble issues have become so noticeable SBNation has dedicated an article to exploring all 13 of them. https://www.sbnation.com/nfl/2019/11/12/20959164/daniel-jones-fumbles-giants-rookie-nfl-ranking-oof The funny thing is that QB fumbles are supposed to be easier to recover than running play fumbles. (47% vs 55%: archive.advancedfootballanalytics.com/2010/01/fumble-rates-by-play-type.html ) I don't quite know what to think about the Jones fumble luck. If I discard the pickpocket, they've recovered 4 out of 12, which is hardly statistically interesting from such a small sample size. The Giants could bemoan their fumble luck. But, really, Jones has to do a much better job controlling the ball. He's fumbled more than half the _teams_ in the NFL.
#54 by sbond101 // Nov 13, 2019 - 9:05am
"The Giants could bemoan their fumble luck. But, really, Jones has to do a much better job controlling the ball. He's fumbled more than half the _teams_ in the NFL." - This is interesting and goes to something that's bothered me for a long time about how fumbles are talked about by the stats community. Fumble recovery isn't luck, it's a skill, we all know this and think about it that way when we watch games - but when we do stats analysis we take the position that every team is so close together in the skill of recovering fumbles and the sample size is so small that the difference in fumble-recovery skill will be forever obscured by the randomness that forms a part of fumble recovery (e.g. if you had 500 fumbles/team in a season you would see the skill difference as the randomness would cancle, but you never get a sample large enough to draw a conclusion). (I suspect that there is a tightly clustered distribution on the upper end with a series of low-end outliers as a result of comparably poor effort).
All that said when some terrible team like the 2019 Giants is described as having "poor fumble luck" I have often been suspicious that they might actually be downward outliers with regard to recovery, and have never quite gotten over the fact that i'll never get a large enough sample size to make the arguement convincingly.
#77 by TGT // Nov 13, 2019 - 1:04pm
Maybe blitzing defenses recover more fumbles as QBs are often lion's share of fumbles, and blitzing teams would have more bodies back to pick them up.
I can come up with all sorts of possible reasons like that, but with 20+ players rotating on one side of the ball, and fumbles being an individual stat, it seems likely that any information would be too noisy to handle. If everyone on an offense is 10% less likely to recover a fumble than the average at their position, the team is still going to be positive on fumble recoveries something like a third of the time.
#79 by mrh // Nov 13, 2019 - 1:29pm
This is interesting and goes to something that's bothered me for a long time about how fumbles are talked about by the stats community. Fumble recovery isn't luck, it's a skill, we all know this and think about it that way when we watch games
Recovery of a fumble, despite being the product of hard work, is almost entirely random.
Stripping the ball is a skill. Holding onto the ball is a skill. Pouncing on the ball as it is bouncing all over the place is not a skill. There is no correlation whatsoever between the percentage of fumbles recovered by a team in one year and the percentage they recover in the next year. The odds of recovery are based solely on the type of play involved, not the teams or any of their players.
Fans like to insist that specific coaches can teach their teams to recover more fumbles by swarming to the ball. Chicago's Lovie Smith, in particular, is supposed to have this ability. However, in Smith’s first three seasons as head coach of the Bears, their rate of fumble recovery on defense went from a league-best 76 percent in 2004 to a league-worst 33 percent in 2005, then back to 67 percent in 2006.
Fumble recovery is equally erratic on offense. In 2008, the Bears fumbled 12 times on offense and recovered only three of them. In 2009, the Bears fumbled 18 times on offense, but recovered 13 of them.
#81 by Pat // Nov 13, 2019 - 1:49pm
"Fumble recovery isn't luck, it's a skill, we all know this and think about it that way when we watch games"
Being able to recover a loose ball might be a skill. Having the ball bounce in such a way that it's possible for you to recover it is random. So the result for a team is some part skill, some part random. The evidence suggests that it is dominantly random.
Keep in mind that a part of the reason that the football is shaped the way it is is because the players liked the fact that the bounce was erratic and random. The game was intended to have loose-ball recovery be unpredictable.
#86 by sbond101 // Nov 13, 2019 - 2:43pm
"Being able to recover a loose ball might be a skill. Having the ball bounce in such a way that it's possible for you to recover it is random. So the result for a team is some part skill, some part random. The evidence suggests that it is dominantly random." - This is definately true; as a result an excellent group of fumble recovers might hypothetically recover 5%-10% more fumbles then expected with a large standard deviation in the expected results function. As a result in order to see the effect you would need a huge sample of fumbles with substantially the same group of players having an opportunity to recover them. Given the dynamics of football it is not possible to make a claim on the issue because large sampels of fumbles with the same players involved never exist. This is a similar effect to winning at gambling on blackjack (with basic competence in the rules of the game) being dominantly random but producing a distribution of results depending on freuqency of play mistakes and understanding of the math behind how betting works in an iterative reinvestment game. It's not random - but it looks like it is if you can't see enough trials.
"There is no correlation whatsoever between the percentage of fumbles recovered by a team in one year and the percentage they recover in the next year. The odds of recovery are based solely on the type of play involved, not the teams or any of their players." - This is BS; what it should say is "There is no statistically observabale correlation whatsoever". most groups of players get <10 chances to recover fumbles in their time playing together. The vast majority of coaches (who should have somewhat less effect) get <100 chances to make the effect known. There are no where near enough trials available for any given dataset to derrive a clear conclusion about whether effects like this exist when they are expected to be fairly small and high-variance. I appreciate the above statement by the outsiders is intended to address a fairly simple critque and is functional for that purpose - but it's false at face value. Recovering a fumble is obviously a combination of the play on which it occurs, randomness, and player skill, where the play and the randomness are dominant.
#91 by Noahrk // Nov 13, 2019 - 3:22pm
I just can't see how skill would come into it much beyond practicing falling on the ball or being stronger and taking the ball away from a weaker guy. A few practices and it would seem everyone would be as skilled as anyone else who had practiced it a bit. I can't recall ever watching a fumble and thinking "wow, look at how that guy recovered it!" Well, maybe occasionally, but typically it's ordinary stuff.
#87 by Joe Pancake // Nov 13, 2019 - 2:51pm
The discussion of "fumble luck" is a philosophical/semantic one about what constitutes luck.
It's not luck in the sense most people think of it, where the outcome is totally random no matter who's involved in the experiment (the way flipping a coin is). If a bunch of schmoes played an NFL team in a game of football they would undoubtedly fumble a lot and undoubtedly lose almost all of those fumbles, because they lack the recognition, agility, and strength of their opponent.
However, what has been demonstrated is that in the NFL "fumble luck" is not a repeatable skill and past performance has little to do with future success. So, in analyses, you can typically treat it as if it is luck -- it's tantamount to luck -- and that's the important thing for sites like FootballOutsiders.
You can call it something else if you like, but "positive fumble recovery variation" doesn't have the same ring to it.
#64 by big10freak // Nov 13, 2019 - 11:18am
Here I was looking forward to a DVOA discussion and lo! there are charts to boot and then I find that someone who is part of the GB fanbase has already been here and salted the earth.
Oh well. I need to be quicker next time pre-contamination.
#70 by theslothook // Nov 13, 2019 - 12:02pm
Green Bay is an interesting team because they are so damn hard to pin down exactly how good they are. Theres been very little consistency to any parts of their team it seems. First few games the defense was dominant, even against a good vikings offense. It has since become flat bad. The offense will take turns looking listless and dominant in the same damn game. And god knows their running game is so schizophrenic.
I am trying to remember a team like GB in the past and the most imperfect one that immediately comes to mind is to 2003 Rams. They were a team with a recent pedigree of success coming off a bad prior season. They were also schizophrenic offensively and defensively they survived on a lot of turnovers. So maybe stylistically the comparison doesn't work.
#85 by Cheesehead_Canuck // Nov 13, 2019 - 2:25pm
I can't really disagree. On one hand, I'm impressed with an 8-2 record with the schedule they've had. On the other, I think GB got away with one in the Detroit game and had a semi stroke of luck by playing a Mahomes-less KC. But then so did Minnesota, and they lost. I don't trust the defense but I trust it more than I have in the past 4 seasons.
I think if the good injury luck continues (good in comparison to most of their other seasons) and the right match-ups line up in January, they could make a run. Or we could end up with another Divisional exit when the defense gives up 37 points and I'll have to live with same old casual observers saying Rodgers can't perform in the playoffs because ringz.
#93 by Q // Nov 13, 2019 - 3:39pm
GB is indeed incredibly hard to judge. Not only are both their Offense and Defense seemingly unreliable, but also they leave you with a weird feeling after the game.
Often times this year, GB goes up early by 14/20+ points (Minn, DAL, Car, kinda Den) and then they end up holding on and winning by 5-10 points where they come close to blowing it but not really.
It leaves you wondering: Were they lucky to get the big lead? Were they lucky to hold on or is it more them shutting it down?
Likewise, games such as vs DAL people see as kind of a blowout but probably seem much closer by DVOA
#96 by theslothook // Nov 13, 2019 - 3:52pm
I'll go one further, where does this leave us when evaluating Aaron Rodgers at this point?
If I were to tier qb play, then I would say there are two elite qbs in the league right now, though Rodgers inclusion is as as much about his past performances than his current level of play. He's maddeningly inconsistent and I keep waiting for that to end. Before this year I would have put Rodgers as the best qb in the nfl. Right now I think its Mahomes, who when healthy is guaranteed to throw for 300 and 3tds at a minimum every week. I used to say that about Rodgers.
#102 by Q // Nov 13, 2019 - 4:31pm
Rodgers is indeed hard to judge. I don't know how many QBs could have done nearly as well with the Packers instead when Adams was hurt and all he had were #4 and #5 receivers. GB desperately needs to add playmakers to the Offense in the Offseason which is why I think they are still 1 good draft away from being a SB favorite.
Mahomes is hard to judge because of the opposite reason. With Reid, Hill, Kelce, etc it is an embarrassment of riches. With even just Matt Moore and QB they were scoring almost 30/game. Obviously Mahomes is much better than Moore, but I think almost any QB would look amazing on that team
#106 by Q // Nov 13, 2019 - 5:49pm
I don't recall it ever being said about either:
-When Manning didn't play IND usually struggled mightily. I don't ever remember his backups lighting it up. Manning did though get to play with Great skill position talent so Brady fans did used to say that Brady could have the same #'s playing indoor with Faulk/James, Harrison, etc
-Brady never had high level Offensive Skill position talent around him until 07. His early years the team was carried by the Defense. Backups in NE though have generally looked quite good when they have played, nothing like Hundley or Kiser playing in GB when Rodgers went down
#107 by theslothook // Nov 13, 2019 - 5:54pm
Ok, but it's a variant of - the disparity in production between the two is explained by supporting casts/ coaching - with Mahomes in Manning's role and Rodgers as Brady.
The irony of course is I'm not sure how much we learned from it even way after the fact. Tom Brady's fans were certainly vindicated when he started putting up elite number when the talent improved dramatically, but then I also suspect Brady himself vastly improved even when he was winning his second and third superbowl.
Meanwhile, Manning's production remained relatively stable and his time in Denver only complicates the matter. Harrison is a HOF and Wayne a near one, but none of the cast mates in Denver appear to be anything near hall of fame worthy and the coaches kept changing.
edit - Welker may get in, but by the time he joined Denver, he was on the fading end of his career.
#65 by jacobk // Nov 13, 2019 - 11:18am
Is the bottom quartile of the league noticeably worse than usual? The race for the #1 pick feels more competitive than most years.
I ask because I recall one of the problems with doing DVOA for college is that extreme talent differences make it hard to extract information. Even if LSU and Alabama both played the same set of cupcakes, we just can't get a lot of data from analyzing how ferociously they stomp on the Eastern Oregon Agricultural Institute.
I wonder if there's a threshold where that kind of effect starts to pop up. I also wonder how close Pats-Fins comes to the threshold.
#89 by jimbojonessmith // Nov 13, 2019 - 2:53pm
The Ravens D was #30 after Week 4 with a 17.3% DVOA. Since then they've climbed to #14 (-1.2%). That corresponded with wholesale changes including remaking the LB corps, acquiring Marcus Peters, Jimmy Smith returning healthy, and taking the green dot from first Onwuasor then Tony Jefferson and finally putting it with Clark who now plays 100 % of defensive snaps). Just curious what their D ranking would be if you counted only what they've done beginning in Week 5. I know you can't cherry pick, but would also argue their horrible back-to-back 500-yard performances in Weeks 3-4 are obscuring how much better they've been since.
#118 by Swilson1472 // Nov 15, 2019 - 12:18am
Mahomes is once again the Passing DYAR leader for QBs on the season even though he's played in 1.5 games less than his closest competitors in Prescott and Wilson. He still has a DVOA of over 40% for the tops in the league amongst those with the qualifying number of dropbacks. Andy Reid definitely deserves a lot of credit, but Matt Moore only had a DVOA of 13% which is pretty average.