Week 1 DVOA Ratings
by Aaron Schatz
You love them when your team is high! You hate them when your team is low! Once again, the famous Football Outsiders DVOA and DAVE ratings return for 2017.
Most Football Outsiders readers are familiar with DVOA, which we use all year round. Not as many Football Outsiders readers are familiar with DAVE, which we only use for teams during the first two months of the season. DAVE is our rating that combines the preseason projection with the results of early games to give us a better prediction of how each team will rank at the end of the year. For those who don't know the story, this metric is called DAVE as a reaction to criticism that our stats are too much alphabet soup. I mean, who can argue with a guy named Dave? (Technically, it stands for DVOA Adjusted for Variation Early.) In this week's DAVE ratings, the preseason projection counts for 90 percent, and the current DVOA counts for 10 percent. The value of the preseason projection changes each week until we are only using current-year data after Week 8.
Of course, though I'm calling our main metric DVOA here, it is actually VOA because there are no opponent adjustments right now. We do not apply opponent adjustments until after Week 4, so in Weeks 1-3 DVOA and VOA are the same thing. Please don't get all nitpicky about it. Most readers know what's up, and if you don't, I just told you!
Also, special for 2017, please note that the DAVE listed for Miami and Tampa Bay is actually just the preseason projection. The Dolphins and Bucs will be "a week behind" in the ratio of current performance to preseason projection during the two months we use DAVE. Also, the projection part of DAVE has been adjusted for the biggest injuries and personnel changes after Week 1. The Dallas offensive projection no longer incorporates an Ezekiel Elliott suspension; the Kansas City defensive projection is worse because of the injury to Eric Berry; and the Arizona offensive projection is dropped for the injury to David Johnson, although the playoff odds report also tries to simulate the chance that the Cardinals will get Johnson back late in the season.
Last year's Week 1 had 11 different games decided by one score, which meant a lot of games where our ratings didn't quite sync up with the final result. This was not an issue in 2017. The winning team had a higher DVOA than the losing team in every single game except the last one, as the Los Angeles Chargers end up narrowly ranked higher than the Denver Broncos.
The fact that DAVE currently is still counting our preseason projection as 90 percent of its estimate of team quality is another element of the "don't overreact" mentality that led Football Outsiders to coin the phrase "National Jump to Conclusions Week" to describe the period after the NFL's first official weekend of games. If we only look at Week 1 performance, the best team in football right now is the Los Angeles Rams. No, seriously. So don't jump to conclusions after one week. Just ask last year's No. 1 team after Week 1, the San Francisco 49ers.
Then again, the Rams aren't just high in the DVOA ratings after one week. Because Football Outsiders' preseason projections had the Rams higher than pretty much anyone else's projections, the Rams are also really high in DAVE. When you add 10 percent "best performance of the week" to 90 percent "great defense and special teams, awful offense" instead of 90 percent "awful everything," you end up with the Rams fifth in DAVE. That means our playoff odds report now gives them a stronger chance to win the NFC West than it gives the Seattle Seahawks. This is a little extreme. I don't think anyone around here thinks the Rams are suddenly among the Super Bowl favorites.
[ad placeholder 3]
Part of the issue in decoding the Rams' huge Week 1 win is this: just because we don't do opponent adjustments in Weeks 1-3 doesn't mean you can't add common-sense opponent adjustments in your brain when looking at the tables. We know Scott Tolzien was horrible. The Colts defense was missing the top cornerback and isn't usually that good anyway. The Bills are No. 3 this week while the Chiefs are No. 4, but the Bills played the Jets while the Chiefs beat the Patriots. Do a little common sense in your head and you know who had the better Week 1.
Speaking of the Patriots, we still have them near the top of our playoff odds report and No. 2 in DAVE, behind the Pittsburgh Steelers. When your projection is so far ahead of the rest of the league, you can absorb a loss in the first week of the season. It also helps when the rest of your division is trash. Out of curiosity, I went back to look at which other teams had higher playoff odds after a Week 1 loss. We only have records of our playoff odds after Week 1 going back eight years, but even in that short period -- surprise! -- the Patriots do not have the highest playoff odds of any 0-1 team. Instead, that title belongs to the 2012 Green Bay Packers, who were coming off a 15-1 season but lost to San Francisco 30-22 at home. However, the Patriots do have the highest Super Bowl odds of any team in the last eight years to start the season 0-1. Interestingly, we give this year's Patriots much better odds of rebounding than we gave to the Patriots in 2014. We listed that team with playoff odds of just 42.8 percent and Super Bowl odds of just 3.8 percent after a 33-20 loss on the road to Miami. They went on to win the Super Bowl anyway.
| Best Playoff Odds at 0-1, 2010-2017 | x | Best Super Bowl Win Odds at 0-1, 2010-2017 | ||||
| Year | Team | Odds | x | Year | Team | Odds |
| 2012 | GB | 69.4% | x | 2017 | NE | 9.5% |
| 2017 | NE | 68.9% | x | 2015 | SEA | 8.3% |
| 2010 | ATL | 64.1% | x | 2012 | GB | 8.2% |
| 2011 | PIT | 62.1% | x | 2010 | ATL | 7.5% |
| 2015 | SEA | 57.6% | x | 2016 | ARI | 7.0% |
| 2010 | IND | 56.9% | x | 2011 | PIT | 5.4% |
| 2012 | PIT | 56.4% | x | 2017 | SEA | 4.6% |
| 2011 | NO | 54.7%* | x | 2010 | IND | 4.2% |
| 2011 | ATL | 54.3%* | x | 2011 | ATL | 3.8% |
| 2013 | CIN | 53.9%* | x | 2014 | GB | 3.8% |
| 2016 | ARI | 53.5% | x | 2014 | NE | 3.8% |
| 2017 | SEA | 50.8% | x | 2011 | NO | 3.7% |
| * Odds are high in part because entire division began the season 0-1. | ||||||
* * * * *
A week ago on Twitter, I promised to unveil the long-awaited 1986 DVOA ratings and commentary on the first Tuesday after the season started. That was before I remembered that we run four different articles every Tuesday, more than any other day. Yikes! The 1986 ratings are interesting and come with a really cool companion interview, so I don't want them to get lost among the other articles. They'll now run either Thursday or Friday of this week.
* * * * *
Once again this season, we have teamed up with EA Sports to bring Football Outsiders-branded player content to Madden 18. This year, our content for Madden Ultimate Team on consoles comes monthly, while our content for Madden Mobile comes weekly. Come back to each Tuesday's DVOA commentary article for a list of players who stood out during the previous weekend's games. Those players will get special Madden Mobile items branded as "Powerline, powered by Football Outsiders," beginning at 11am Eastern on Friday. We will also tweet out images of these players from the @fboutsiders Twitter account on most Fridays.
The Football Outsiders stars for Week 1 are:
- C Mitch Morse, KC (FO HERO): Helped lead way for Kareem Hunt's 17-carry, 148-yard performance.
- ROLB Thomas Davis, CAR: Four tackles to prevent conversions on third or fourth down, including a sack.
- SS Karl Joseph, OAK: Led Raiders with 9 combined tackles; played a major role in holding Titans slot receiver Eric Decker to 3-of-8 for 10 yards.
- LOLB Josh Martin, NYJ: Half sack, two run TFL, pass reception TFL.
- LG Kelechi Osemele, OAK: Controlled Jurrell Casey, giving up no sacks.
* * * * *
[ad placeholder 4]
All stats pages are now updated with 2017 data except for OFFENSIVE LINE and DEFENSIVE LINE, which will be updated after Week 2, and FO snap counts. This year, for the first time, we are planning to update Football Outsiders Premium data after Week 1 instead of waiting for Week 2. However, that will end up delayed a bit because new tech director Dave Bernreuther lives in Miami and has lost power and Internet access due to Hurricane Irma. That's also the reason snap counts aren't updated; we'll take care of all these updates as soon as we can.
* * * * *
These are the Football Outsiders team efficiency ratings through one week of 2017, 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.
Please note that there are no opponent adjustments in DVOA until after Week 4. (It's still listed as DVOA instead of VOA because I don't feel like going through and changing all the tables manually.) In addition, our second weekly table which includes schedule strength, variation, and Estimated Wins will appear beginning after Week 4.
DAVE is a formula which combines our preseason projection with current DVOA to get a more accurate forecast of how a team will play the rest of the season. Right now, the preseason projection makes up 90 percent of DAVE. For Miami and Tampa Bay, only preseason projection is listed.
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>
| TEAM | TOTAL DVOA |
TOTAL DAVE |
RANK | W-L | OFFENSE DVOA |
OFF. RANK |
DEFENSE DVOA |
DEF. RANK |
S.T. DVOA |
S.T. RANK |
|
| 1 | LARM | 99.6% | 12.0% | 5 | 1-0 | 20.0% | 9 | -67.7% | 2 | 12.0% | 1 |
| 2 | BAL | 82.9% | 11.5% | 7 | 1-0 | -12.5% | 22 | -94.6% | 1 | 0.8% | 15 |
| 3 | BUF | 68.7% | -3.8% | 19 | 1-0 | 27.0% | 6 | -48.2% | 3 | -6.4% | 27 |
| 4 | KC | 64.2% | 7.9% | 10 | 1-0 | 74.3% | 1 | 5.4% | 17 | -4.6% | 26 |
| 5 | CAR | 55.9% | 8.4% | 8 | 1-0 | 8.2% | 17 | -44.8% | 4 | 2.9% | 9 |
| 6 | DAL | 48.1% | 16.4% | 3 | 1-0 | 22.3% | 8 | -16.5% | 10 | 9.2% | 3 |
| 7 | PIT | 47.9% | 25.7% | 1 | 1-0 | 25.2% | 7 | -18.6% | 9 | 4.1% | 8 |
| 8 | DET | 39.2% | 5.4% | 13 | 1-0 | 10.4% | 15 | -29.7% | 6 | -0.9% | 19 |
| 9 | ATL | 35.9% | 6.7% | 11 | 1-0 | 38.9% | 3 | 5.8% | 18 | 2.7% | 10 |
| 10 | GB | 34.7% | 15.8% | 4 | 1-0 | 13.4% | 13 | -23.8% | 7 | -2.6% | 24 |
| 11 | JAC | 33.2% | -6.0% | 21 | 1-0 | 11.9% | 14 | -36.7% | 5 | -15.4% | 30 |
| 12 | PHI | 24.1% | 5.7% | 12 | 1-0 | -3.0% | 20 | -20.1% | 8 | 7.0% | 4 |
| 13 | MIN | 16.0% | -1.0% | 16 | 1-0 | 38.6% | 4 | 20.7% | 24 | -1.9% | 23 |
| 14 | OAK | 15.2% | 8.3% | 9 | 1-0 | 39.4% | 2 | 29.1% | 30 | 4.9% | 6 |
| 15 | TEN | 5.2% | -0.8% | 15 | 0-1 | 37.0% | 5 | 28.5% | 29 | -3.3% | 25 |
| 16 | MIA | 0.0% | -4.7% | 20 | 0-0 | 0.0% | 18 | 0.0% | 13 | 0.0% | 16 |
| TEAM | TOTAL DVOA |
TOTAL DAVE |
RANK | W-L | OFFENSE DVOA |
OFF. RANK |
DEFENSE DVOA |
DEF. RANK |
S.T. DVOA |
S.T. RANK |
|
| 16 | TB | 0.0% | -7.4% | 24 | 0-0 | 0.0% | 18 | 0.0% | 13 | 0.0% | 16 |
| 18 | LACH | -1.0% | -2.0% | 18 | 0-1 | 9.7% | 16 | -9.7% | 11 | -20.4% | 31 |
| 19 | DEN | -4.6% | -6.4% | 22 | 1-0 | -3.5% | 21 | 3.3% | 15 | 2.2% | 12 |
| 20 | CHI | -9.6% | -6.6% | 23 | 0-1 | 16.2% | 10 | 27.1% | 27 | 1.3% | 13 |
| 21 | WAS | -23.3% | -1.8% | 17 | 0-1 | -17.4% | 24 | 4.7% | 16 | -1.2% | 21 |
| 22 | NO | -25.1% | -10.5% | 27 | 0-1 | 15.0% | 11 | 46.3% | 31 | 6.1% | 5 |
| 23 | SEA | -29.9% | 11.9% | 6 | 0-1 | -25.5% | 26 | 13.8% | 21 | 9.4% | 2 |
| 24 | NE | -45.7% | 19.3% | 2 | 0-1 | 13.9% | 12 | 63.9% | 32 | 4.4% | 7 |
| 25 | NYG | -49.4% | 0.9% | 14 | 0-1 | -20.1% | 25 | 28.5% | 28 | -0.8% | 18 |
| 26 | HOU | -52.9% | -14.2% | 28 | 0-1 | -40.1% | 28 | 15.1% | 22 | 2.3% | 11 |
| 27 | CLE | -55.0% | -21.9% | 31 | 0-1 | -13.8% | 23 | 18.5% | 23 | -22.8% | 32 |
| 28 | ARI | -61.2% | -10.3% | 26 | 0-1 | -35.0% | 27 | 13.5% | 20 | -12.8% | 28 |
| 29 | SF | -65.0% | -19.3% | 30 | 0-1 | -56.4% | 30 | 9.6% | 19 | 1.0% | 14 |
| 30 | NYJ | -68.0% | -23.6% | 32 | 0-1 | -45.7% | 29 | 21.0% | 25 | -1.3% | 22 |
| 31 | CIN | -100.1% | -9.3% | 25 | 0-1 | -105.1% | 32 | -6.1% | 12 | -1.1% | 20 |
| 32 | IND | -119.2% | -18.1% | 29 | 0-1 | -79.3% | 31 | 26.2% | 26 | -13.7% | 29 |
Comments
111 comments, Last at 21 Sep 2017, 7:22pm
#1 by deus01 // Sep 12, 2017 - 6:37pm
Interesting that the chargers have better DVOA than denver, especially since special teams DVOA is so low. I assume the low ST DVOA is due to the blocked field goal; I would think those are fairly random but I guess it may not be treated that way in DVOA.
#2 by theslothook // Sep 12, 2017 - 6:50pm
As a bayesian - I think the LARM prior is way too weak. It would take me at least a quarter of the season before I felt any kind of confidence about their future. Ok, maybe goff was all a function of Jeff Fisher incompetence - but I've seen that same excuse passed for every failed qb ever. Plus - I wonder if theres any team that gets blown out on the road more than the colts. Especially a colts team without andrew luck and playing outside the putrid AFC South/
#40 by zlionsfan // Sep 13, 2017 - 10:15am
Answer provided: in the last 20 years plus week 1 from this year, eight teams have lost more games by 28+ points on the road than the Colts, who've lost 10 such games. Most of the top ten teams will be no surprise, but a couple of them might be.
17 - San Francisco
14 - Arizona
13 - Oakland
12 - Buffalo, LA Rams
11 - Chicago, Detroit, Kansas City
10 - Indianapolis, Tampa Bay
Note: p-f-r currently tags the Rams prior to 2016 as STL and during 2016 as LAR, so watch your pivot tables carefully.
#50 by Eddo // Sep 13, 2017 - 11:53am
I feel like going back 20 years is not what the original poster intended; I can't imagine he was lumping in the Manning Colts with teams that often lay complete eggs on the road.
I did a quick playfinder from 2012 onward (the Pagano/Luck era), and the Colts do have the most road 28+ losses in the league, with 6:
IND 6
CHI 4
MIA 4
NYJ 4
OAK 4
PHI 4
SFO 4
TAM 4
GNB 3
JAX 3
KAN 3
NYG 3
No other team even has 5 such losses. It's also odd that Packers, Chiefs, and Giants are near the top of the league in this metric, given that they've been among the top half of the league pretty consistently during that time.
#55 by JoeyHarringtonsPiano // Sep 13, 2017 - 1:17pm
"It's also odd that Packers, Chiefs, and Giants are near the top of the league in this metric".
Giants tend to implode during a "bad Eli" game, so it's not that suprising. As for the Chiefs, how many of those losses were in the Romeo Crennell 2-14 year in 2012? The Packers having three such losses, is surpising. On the other hand, Rodgers missed most of the season in 2013, and I remember then having a couple of ugly losses when captained by the esteemed Scott Tolzien and Matt Flynn.
#75 by SuperDackMorph // Sep 14, 2017 - 1:30am
F.O was fairly high on the Rams this year. I believe their mean projection was 8 wins. And their projections are notoriously conservative. Schatz and company see Wade Phillips as defensive coordinator as a big reason for this year's jump. He has made even average defenses top 10.
#41 by Chip // Sep 13, 2017 - 10:37am
More perplexing is the DEF at 27th. They held ATL to 2.5 yd/rush, 23 points and forced 3 3-outs (ATL avg'd 1 per game last year).
Must have either been the 3 long 10+ play drives or the 88 yard TE bomb.
Either way, the DEF looked good and OPP adj. will likely improve this ranking.
#108 by Steve in WI // Sep 15, 2017 - 2:04pm
I had the same gut reaction to both rankings, but aside from small sample size and the usual caveats, I think my biases go a long way toward explaining my reaction. DVOA sees a defense that gave up the long drives and the bomb you mentioned; I see that a defense that had been godawful for most of the past 3 seasons didn't crap the bed against one of the better offenses in the league, so I'm probably overvaluing basic competence.
#5 by LionInAZ // Sep 12, 2017 - 7:04pm
It's amazing to me that a defense that looked so good against the Lions is punished so hard, and that the Lions offense gets so little credit. This is the ultimate failing of preseason projections.
In other words: preseason projections are worth less than the projectors want to admit.
Furthermore. minor weekly adjustments fail to adjust adequately, which the projectors don't really want to admit either.
#9 by Eddo // Sep 12, 2017 - 7:35pm
Huh? For one, there are no opponent adjustments yet, so Arizona's +13.5% defensive "D"VOA is not affected by the Lions' offensive ranking.
The Lions' offense averaged 5.3 yards per play and had 19 first downs and a turnover. They went 8/16 on third down. Having a "D"VOA of +10.4 might be about right, depending on how the individual plays went. They weirdly had five fewer first downs than their opponent despite running two more offensive plays.
I didn't see much of that game, but from what I was hearing, it sounds like the Lions dominated on defense, which "D"VOA shows; Detroit had -29.7%, ranked 6th overall this week (only behind a team that shut their opponent out, two others than had two defensive touchdowns each, and the teams that played the Jets and 49ers).
#8 by lokiwi // Sep 12, 2017 - 7:31pm
My jump to conclusions assumption for the week is that the DAVE projection for NEs defense was overly optimistic. They were projected to have the 2nd largest positive jump in the league (Non-Wade Phillips Division). Only larger jump was Det projected to go from dead last to middling, which seems like a projection for better health. But I'm not sure where the huge jump for NE was supposed to come from. Swapping Ryan for Gilmore doesn't look like that big a difference, and neither does adding two mid-round DEs in the draft. So I am jumping to the conclusion that they will settle into the 10th-15th range in defensive DVOA that they've been in for the past two years, rather than scraping the top 5. I'm sure they will prove me horribly wrong.
Also, I forgot how low DAVE is on the Chargers D. Dropping from top 7 to bottom 7 is harsh.
#11 by theslothook // Sep 12, 2017 - 8:18pm
Its pretty remarkable how awful the defense has remained for this many years. They've tried to fix it but its been stubbornly in the cellar. Just as its plenty hard to keep fielding great defenses every year, it seems equally difficult to suck this bad for this long. Poor Brees.
#12 by dmstorm22 // Sep 12, 2017 - 8:25pm
To me it is equally, if mot more, stunning how little flak Payton has gotten for having such consistently terrible defenses. I get that he's the offensive guy, but we've seen the reverse situation many times, where the defensive-minded coach has issues with the offensive side of the ball, and those coaches get pilloried and usually inevitably fired.
#38 by Aaron Brooks G… // Sep 13, 2017 - 9:57am
It might be okay. MIN has a good defense, but the Saints didn't exactly look vintage against them. Their line has quietly turned into hot trash.
Not the raging compost fire that is Seattle, Houston, Cincinnati, or NYG, but hot trash nonetheless.
I'm guessing NE will do okay against the corpse of the New Orleans Saints.
#46 by dmstorm22 // Sep 13, 2017 - 11:15am
I think you're selling MIN's defense short. Sure, the Saints offense is not nearly as consistent and explosive as it used to be, but (despite pre-season DVOAs love for NE D) going from Minnesota on the road in primetime to NE at home might be a good dropoff and allow the Saints to look like the Saints of old for a bit.
Let's not forget, Saints ended last season #2 in scoring.
#21 by Sleet // Sep 13, 2017 - 2:26am
Guess one should expect outliers and anomalies with one week's worth of data. I prefer to lump FO in with the haters.
Last year the Titans ended with FO's 5th ranked offense, including FO's 4th ranked rushing attack and 8th ranked passing attack (https://www.footballoutsiders.com/stats/teamoff). Week 1 this year Murray and Henry can't break 70 yards and even with 3 nice runs by Mariota, the Titans can't break 100 yards. The Raiders' DL, with 4 new main cogs, including two rookies (Edwards, Ellis, Vanderdoes, and Hester), stuffed the Titans' OL and its two All Pro OT (Lewen and Conklin). Vanderdoes had more pancakes this week than Osemele; must have been a joy watching those two battle in training camp. The Raiders' secondary gave up yards early, but held in the redzone and was solid in the 2nd half. At the end of the day, the Raiders' D gave up 16 points, even without any turnovers and only 1 sack. That's good run D. That's good coverage. That's good defense against FO's 5th ranked offense last year. Yet, somehow, the haters at FO compiled a DVOA that ranked the Raiders' D 30th after Week 1. If it's not a conspiracy, this ranking discredits FO's good work.
#Embarrassing
#70 by nat // Sep 13, 2017 - 5:37pm
1) Weren't some of those points from a fumble return and not involving the Houston defense at all?
2) Did you take a look at the drive stats? They're interesting when comparing Houston and Oakland.
a) Oakland faced fewer meaningful drives, giving them fewer chances to give up points than Houston had.
b) One such drive was curtailed by the game clock, leading to a missed FG attempt on 1st down. That first down "stop" had little to do with the defense's skill. But it sure helps the points per drive stat.
c) The Oakland defense faced much easier field position - enough to account for about 100 extra yards or about 7 points.
d) Neither team forced a fumble or made an interception.
e) Houston was better in terms of yards per drive.
f) Houston was better at preventing first downs. (drive success rate)
The main point in favor of Oakland's defensive drive stats is that they did a bit better in the red zone than Houston did. That can be important in a game, but is not usually considered predictive, and therefore doesn't factor into VOA. They also gave up fewer points per drive. But that's a consequence, not a reason. It's more than adequately explained by the field position and the clock running out on the final drive.
Mostly, the difference in VOA isn't very big and it's in a one game sample. It's quite understandable, really. Not very important. But reasonable.
Don't sweat Oakland's defensive VOA. Heck, you could be a Patriots fan!
#76 by Sleet // Sep 14, 2017 - 9:22am
You FO homers crack me up:
1) Houston's D played against a lesser offense with a bigger lead that played conservatively by design. How does your precious stat calculate that?
2) The Raiders' D had more meaningful stops, where it prevented Tenn from taking the lead or tying the game, which was not decided until late in the fourth quarter. Where does that get factored in?
3) The Raiders' D gave up 16 points, however you want to belittle it, and had as much to do with the victory (during its 29 minutes and 64 snaps on the field) as the Raiders' O, yet their relative rankings are 28 spots a part. 28 spots? Really? That's the issue. No one is saying that the Raiders' D deserves a top 10 ranking. But it does not deserve a bottom 3 ranking whatever math you use.
4) Playing against a better QB who runs well (which FO's beloved QBR rewards) and a strong running attack, with a bend-don't-break philosophy (Norton blitzes among the league low at 18% of the time), field position affects defensive coverages and calls. How does your stat calculate that?
5) Your generic view of drive stats also does not take into consideration (a) how Mariota had to convert multiple 3rd-and-long plays to score in the first half, (b) how the Raiders' DL consistently controlled the LOS throughout the game, stuffing Tenn's run game and pressuring Matiota, which is why they held in the redzone, and (c) how the Raiders' D improved and Mariota was unable to keep converting low percentage plays as the game wore on and mattered most, getting off the field and giving up only 6 points in the 2nd half.
6) Mack had a strip sack, fumble recovery nullified by a stupid penalty, which you don't mention and which obviously does not get factored in, but which is more reflective of the pressure Mariota was under than FO's #30 defensive ranking.
Point stands, no one watching the game would remotely view the Raiders' D as the #30 ranked defense after Week 1. You defend FO's ranking all you want, but it doesn't change that reality. FO can ignore the anamoly and say "shit happens" or tinker with its formula and how it weights what it considers important. If it chooses the former, perhaps an adjustment based on the quality of Tenn's O (as compared to Jags' O in your example) will bring the ranking more in line with reality, but I'd venture a guess not by much. At the end of the day, however the Raiders' D plays over the course of the season, this overly poor Week 1 ranking will negatively skew the Raiders' D's ranking going forward. And creating a ranking or list is the entire point of the exercise.
It should be more accurate.
#77 by Will Allen // Sep 14, 2017 - 9:56am
Please show us your statistical approach to describing a football game, so we can all better understand how things "should" be.
I don't think anybody has stated that DVOA is a perfect measurement of a football team's performance, so your rhetoric is a bit puzzling.
#92 by Sleet // Sep 14, 2017 - 8:43pm
Well, let's not be pedantic. I'll leave the process and trivial points to you. I'm not the person who publishes a list and ranks defenses by their performance. I'm the person who pointed out an anomaly. When a list does not match what people saw on Sunday, perhaps the problem is not in the criticism.
A lot of good points have been raised to defend and explain the ranking. But at the end of the day, the Raiders' D's performance on Sunday did not deserve a #30 ranking. It just didn't. The D did not play 28 spots worse than the Raiders' O, relatively speaking, in terms of contributing to that win, which often was the case last year. It wasn't on Sunday. Yet, this ranking will carry forward, at least to some degree, skewing the Raiders' D's ranking the rest of the season.
Everyone can get mad, be pejorative, or jump up and down b/c the formula produced a number/ranking that's off. Yes, I realize it's a problem associated with sample size and inherent in any formula. Sometimes a low point total and relatively few rushing yards and solid pressure on a QB might mean something different. Heck, the Raiders' D could fall on its face against the Jets. But for one week, FO's #30 opinion/support for the narrative that the Raiders' D sucks, well, is just wrong.
#101 by Will Allen // Sep 15, 2017 - 1:13am
You keep writing as if one week's ranking that doesn't meet your perception is significant, in terms of judging what a metric "should" be. It isn't, and the fact that you fervently believe otherwise is indicative of a misunderstanding of what DVOA is.
#111 by Sleet // Sep 21, 2017 - 7:22pm
You have it a little backwards. My objection is to what the ranking "should not" be (#30), as I don't think the metric (and the hard work put into it) is accurate. I don't know what that ranking should be. That is why I came to FO to see what the ranking "should" be, but was disappointed b/c FO published a ranking that was not accurate. Perhaps you don't see the difference. But time won't change why FO's metric viewed the Raiders' Week 1 performance as the 3rd worst in the NFL, when it wasn't.
#78 by nat // Sep 14, 2017 - 10:26am
Wow. Just Wow.
Point by point.
1) (D)VOA is not opponent adjusted at this time of the season. Get over it.
2) VOA has some adjustments for situation. In essence it compares a play's results to similar situations in many, many other games. But it doesn't assign extra weight to "meaningful" situations. Why not? Because defenses and offenses are trying to succeed on almost all plays - meaning almost all plays are equally valid data points for how well they are playing and predicting future success.
3) Oakland's offense was extremely successful on a per drive basis: 5th in yards per drive, no turnovers, and fourth in converting first downs (dsr). The defense was 25th in yards per drive, dead last in turnovers, and 20th in preventing first downs. There's really no reason to think they were equally good. It's not even close.
4) See (1) and (2).
5) Actually, VOA does a good job at combining the success at different downs. If a team consistently gives up first downs from third-and-long it is penalized for that. On the other hand, it will have been rewarded for the two plays before that. It tends to balance out. But since it results in a lot of first downs, it balances out against the defense, as it should.
6) Yes. A defense committing an offside penalty has made a bad play. VOA does not try to reward them for what looked like a good play, since the stats have no way to know whether the penalty contributed to the good play. On the other hand, we KNOW the defense made a bad (stupid) penalty. That tends to be predictive, and counts against the defense. Get over it.
My suggestion is that you ignore the rankings, especially after one game. Instead, focus on what the stats say and why: giving up lots of yards, no turnovers, bad at stopping the chains from moving, but using great field position to limit points tells us a lot about Oakland's defense (in this one game): it needed its offense and special teams to make it look good.
One last point: Oakland's defense gave up first downs on all downs, not just third downs. In fact, it gave up more first downs on second down than third, and five on first down. That's not good.
#79 by dmstorm22 // Sep 14, 2017 - 10:37am
This is good stuff all around.
Want to add as a note for Sleet - focus more on their actual DVOA, not the ordinal ranking.
Someone else mentioned this, but they are closer to ARZ in 20th place than NO in 31st place. It just so happens this was a low-offense week in the NFL, so a lot of defenses graded out fairly well.
Additional notes:
- The 3rd & long conversions only happened a couple of times, mostly in the first half. It was often them giving up 1st downs on 2nd down
- DVOA is not grading things like pressure faced. Even if there was massive pressure, and Mariota got hit right after he threw, if he completed a 12-yard pass for a 1st down that will go down as a negative play; You can view pressure rate and things like that to see those impacts. No one is saying DVOA is the be-all and end-all stat for judging a defense
- The point that this is still VOA and not DVOA is really important. This performance may not look all that bad by the end of the year if the Titans are a Top-10 offense week-in and week-out
#80 by Poolboy // Sep 14, 2017 - 12:49pm
Longtime lurker, first time poster. I have to say well done to the standard posting community of FO here for your diplomacy, tolerance and patience with Sleet. I mean, I might have, long ago, said "clearly you don't agree and want to crybaby about your Waiders defensive wanking - so maybe this site isn't your thing?" But that's just me. For my piece of it I will say its not much fun to read your same (whiny) points repeatedly with slightly different wording 6 or 7 times. 2 or 3 times is plenty. Thanks.
-Poolboy
#81 by dmstorm22 // Sep 14, 2017 - 12:57pm
I'll say this. Sleet (and others in the past) seems to actually care about what FO is trying to accomplish, and is bringing up questions and points that while potentially (and in many case literally) invalid and wrong, are more interesting than the standard.
Also, I think he is not wrong in that for people who watched the Oakland-Tennessee game, the Raiders defense did not look on the whole awful. I too was surprised they graded out poorly. I expect when the D and DVOA becomes more impactful it won't look as bad, but still did not think this is one of the worst defensive performances of the week.
I actually like these types of discussions because it allows for a magnifying lense to be shined a bit on DVOA. The legendary example was the NYJ@NE game in 2011 when the game graded out that the Jets outplayed New England. I think it became so heated a discussion Aaron (or someone at FO) released a detailed excel scoring on the game to show that, yes, in fact the case can be made play-by-play that it was an evenly played game.
#84 by Bright Blue Shorts // Sep 14, 2017 - 1:46pm
Given that Sleet absolutely slated the analysis of Derek Carr at the beginning of last year and then pops up on this thread questioning the stats about the Raiders defense, I find it ironic that he refers to others as "FO homers".
#88 by nat // Sep 14, 2017 - 4:33pm
Thanks. Though I did come close to being snippier than I wanted to be.
I am reminded of the discussions of the Patriots' #1 points allowed defense last year, with its 8th place ranking in (unadjusted) VOA and 16th in DVOA. If you looked closely, you could see that #1 PA ranking was a mirage.
As Aaron noted:
A lot of what people think is Patriots defense is actually field position: strong kickoffs, strong punts, and a lack of Brady turnovers means the defense almost never faced a short field this year.
Coupled with a weak set of opponents (offense, anyway), it meant the "best" defense in the league (by naive standards) was close to average and properly ranked 16th out of 32 teams by DVOA.
#87 by MilkmanDanimal // Sep 14, 2017 - 3:53pm
I just roll my eyes at this point, as I've never understood why somebody who doesn't like stats would hang out on a website dedicated to statistical analysis.
To be fair, my standard for FO annoyance pretty much starts with, "Oh, God, he's talking about Tomlin again, isn't he?"
#94 by Sleet // Sep 14, 2017 - 8:55pm
Actually, I love stats, which is why I frequent FO. UCLA, unfortunately, didn't offer much of a curriculum in this regard when I went there, which I'm sure is many years after you went to college. But I took the course it did offer and enjoyed it. I was using analytics when I first started playing FF, perhaps before many here were born. I grew up playing Strat-O-Matic (the box game in the below link). Making generalities about people in an attempt to belittle them sometimes misses the mark.
http://www.ebay.com/itm/like/112554225837?dispItem=1&ul_ref=http%253A%252F%252Frover.ebay.com%252Frover%252F1%252F711-117182-37290-0%252F2%253Fmpre%253Dhttp%25253A%25252F%25252Fwww.ebay.com%25252Fitm%25252Flike%25252F112554225837%25253Fchn%25253Dps%252526dispItem%25253D1%2526itemid%253D112554225837%2526targetid%253D335785610296%2526device%253Dm%2526adtype%253Dpla%2526googleloc%253D9031151%2526poi%253D%2526campaignid%253D884782301%2526adgroupid%253D44756791576%2526rlsatarget%253Dpla-335785610296%2526abcId%253D1128996%2526merchantid%253D110817247%2526gclid%253DEAIaIQobChMIoLuq8Pyl1gIVSIJ-Ch2xdQQaEAQYASABEgI4QPD_BwE%2526srcrot%253D711-117182-37290-0%2526rvr_id%253D1310268282829&chn=ps&dispItem=1&ul_ref=http%253A%252F%252Frover.ebay.com%252Frover%252F1%252F711-117182-37290-0%252F2%253Fmpre%253Dhttp%25253A%25252F%25252Fwww.ebay.com%25252Fitm%25252Flike%25252F112554225837%25253Fchn%25253Dps%252526dispItem%25253D1%2526itemid%253D112554225837%2526targetid%253D335785610296%2526device%253Dm%2526adtype%253Dpla%2526googleloc%253D9031151%2526poi%253D%2526campaignid%253D884782301%2526adgroupid%253D44756791576%2526rlsatarget%253Dpla-335785610296%2526abcId%253D1128996%2526merchantid%253D110817247%2526gclid%253DEAIaIQobChMIoLuq8Pyl1gIVSIJ-Ch2xdQQaEAQYASABEgI4QPD_BwE%2526srcrot%253D711-117182-37290-0%2526rvr_id%253D1310268282829&ul_noapp=true
#91 by Sleet // Sep 14, 2017 - 8:41pm
And your didn't.
Yes, I am a fan of the Raiders, which always means I'm subject to being profiled.
Let's not act like FO's numbers are not infallible. Kacsmar has been predicting doom for Carr ever since Carr was a prospect. Now I wonder if Kacsmar doesn't play it up for clicks.
IMO, any fan that wouldn't ridicule an undeserved #30 ranking, well, is not much of a fan.
#93 by Sleet // Sep 14, 2017 - 8:46pm
Thanks. Nice post.
I appreciate that DVOA is not the end-all, be-all stat. Your pressure point is a good example.
But it is the stat FO publishes and uses to rank defenses. And, I assume my response is not a-typical of a fan. They don't want to see a ranking that they view is unfair or not reflective of performance, which your star purports to judge.
Bend-don't-break is a thing, again, not one I particularly like, but it is a thing. Low rushing totals and pressure rates are things too. And they seem to get washed out in the DVOA ranking.
#89 by Sleet // Sep 14, 2017 - 8:02pm
Wow, indeed.
You ignored the two most fundamental points in my post:
1. 16 points. Bend-don't-break is a defensive philosophy, one I particularly don't like, but that really good defenses, like KC's, seem to have perfected in recent years, and which the Raiders executed well on Sunday. They blitzed a near league low 18% of the time. They mostly played cover 2 and some quarters, keeping be ball in front of them. And it worked.
2. Less than 70 yards from Tenn's RBs (and less than 100 yards total when you add in Mariota, one of the better running QBs). Yes, those numbers are more impressive when you consider that it was Murray and Henrey getting stuffed behind one of the best OL, including two All Pro OTs. But even without adjusting for the opponent, when you consider it was a close, relatively low scoring game, where eating clock and keeping Carr & Co. off the field were high priorities, and yet the Titans only mustard 70 yards from its top 2 RB, one can deduce that the Raiders' DL controlled the trenches, and, in fact, it did. Yet, you make a big deal that the Raiders did not get a turnover when relying on the fickleness of turnovers, to me, is far less predictive than winning the battle up front. The Raiders' DL dominated the Titan's really good OL, both in terms of stuffing the run all game and generating pressure as the game wore on.
You also ignored the conclusion, instead getting lost in the weeds. No way a defense that gave up 16 points and dominated the trenches, as evidenced by giving up only 16 points and less than 70 yards rushing from RBs, should be rated #30. Again, I'm not saying where they should be. I'm just saying where they shouldn't.
#85 by Wikitorix // Sep 14, 2017 - 1:53pm
And creating a ranking or list is the entire point of the exercise.
No, creating a ranking is not the point of the exercise. What FO in general and DVOA in particular is attempting to do is gain a deeper understanding of the game of football.
At the end of the day, however the Raiders' D plays over the course of the season, this overly poor Week 1 ranking will negatively skew the Raiders' D's ranking going forward.
Rankings are meaningless in and of themselves. Stop complaining about the ranking. It doesn't factor into any calculation and it won't skew anything going forward. Worry about the DVOA - it's the number that matters. The DVOA isn't set in stone, either. The DVOA calculation is missing sixteen weeks of inputs.
If it chooses the former, perhaps an adjustment based on the quality of Tenn's O (as compared to Jags' O in your example) will bring the ranking more in line with reality, but I'd venture a guess not by much.
There will be an adjustment for this week's DVOA number once there is a better understanding of how good Tennessee's offense is (after week four). After week five, there will be an even better understanding of the Titans, so they'll come back and revise the numbers again. This week's DVOA will get revised every week thereafter. This adjustment might be larger than you think. Opponent adjustments can be worth 30 points or more in DVOA.
#98 by Sleet // Sep 14, 2017 - 9:08pm
I appreciate your passionate defense.
As a consumer (football fan), DVOA is a list. It is a ranking. That is what FO markets and sells. That is why FO gets clicks. Yes, FO goes to some length (perhaps more than others) to explain the purpose and limits of its undertakings, including DVOA, but at the end of the day, FO is in the business of making and selling new, creative lists/rankings.
This one struck a cord. That's all.
#25 by eggwasp // Sep 13, 2017 - 7:14am
Thanks v helpful.
None of that explains why 28 defenses are ranked above OAK, but thanks for your input. Oakland is 8th in points conceded, 19th in yards allowed, 23rd in yards/play, 23rd in 3rd Down % and 7 other teams didnt get any fumbles - including their opponents. Its therefore a point of discussion as to why they ended up ranked #29, as to what happened in the VOA system to result in this ranking. Given we only have a partial description of (D)VOA system/formula.
#28 by roguerouge // Sep 13, 2017 - 8:24am
As for why, according to your post, they gave up yards early, didn't get turnovers, and only got one sack. Points do matter, but consistently moving the chains and staying on schedule is something that DVOA values dearly.
#47 by Anon Ymous // Sep 13, 2017 - 11:21am
Because DVOA straddles the line between being descriptive and being predictive. When teams allow consistent success on 1st and 2nd down but have just enough 3rd down success to limit TDs, then that generally does not carry over to larger samples.
The error you are making is thinking that DVOA is making a commentary purely about how week one went in isolation.
#58 by Sleet // Sep 13, 2017 - 1:33pm
I love this point. It's why analytics are useful. I just don't understand it. Having watched the game, I think holding Murray and Henry below 70 bodes will for the future. MEJ, Ellis, Vanderdoes and Hester providing more interior pressure that the four main cogs on the Raiders' DL last year bodes well. Smith not getting shredded like he did early in the season last year bodes well too. Heck, Nelson and Joseph might have had good games at the same time for the first time. So, while I appreciate the general response, it still does not remotely reconcile the #30 ranking with what we saw on Sunday or can expect based on that performance against a top offense last year.
#31 by Sleet // Sep 13, 2017 - 8:50am
Unless you're KC, of course, which gives up yards between the 20's as much as any D, and seems to make plays only in the redzone. Even their ranking (#17) is a little silly, but KC's D did give up 27 points to last year's #2 offense.
Point is, the formula has produced a number (#30) that is out of touch with what everyone saw watching the game. If you watched the game, Tenn's OL got dominated in that tenches by Mack, Edwards, Ellis and Vanderdoes, who ended up with PFF's 3rd best pass-rushing productivity grade while Tenn doubled/tripled teamed Mack, who still ended up with PFF's highest overall grade among all defenders. So, stop being a FO homer and either be transparent and publish the formula, or call stupid, stupid.
BTW, in trying to come up with an explanation, myself, I wondered about time of possession and the Raiders' D's early inability to get off the field multiple times in the 1st half on 3rd and long. But 29 minutes, 64 plays and yards between the 20's should not get more wieght than holding one of last year's best running games to less than 100 yards, and neutralizing a top 10 passing team last year to 16 points without any turnovers and just 1 sack. That's good pressure and coverage, which is what we saw on Sunday, and what FO's hyperfocus on something is not picking up accurately (IMO).
#32 by dmstorm22 // Sep 13, 2017 - 8:55am
While I agree with the overall point that 30th seems low for a defense that seemed to play reasonably well, DVOA is not accounting for the fact that TEN was really good running the ball last year.
There are no opponent adjustments in yet. And when they start to be included, it will be from this season. I think TEN is fairly good on offense, so that Raiders performance may start to look better down the line.
#35 by Sleet // Sep 13, 2017 - 9:19am
I agree. That's where I started (one week anomaly) and your point about no adjustment yet is good. It's just not fun. Heck, FO included a "DAVE" to account to projections. Perhaps that collume should rely more on last year's opponent's actual rating.
#37 by dank067 // Sep 13, 2017 - 9:34am
Since you bring up that the Raiders finished 23rd last week in yards/play allowed—that's one particular stat that's known to correlate pretty well with DVOA. The team that finished 23rd in DVOA, Cleveland, finished with a defensive DVOA (18.5%) that I would say is pretty much in the same ballpark as the Raiders (29.1%), at least as far as a single game is concerned. In fact, the single game number that Oakland finished with is closer to Arizona at #20 (13.5%) than the next team down one ranking in DVOA, New Orleans at #31 (46.3%). New England is even more distant (63.9%) at #32.
So it looks to me like, while DVOA was not a huge fan of the Raiders performance last week (yds/play allowed helping explain why), the only two teams who who played truly terrible defense last week were New England and New Orleans. The Raiders defense didn't play much worse (if at all) than the mass of teams that finished from 20 to 30. We need more data before the rankings themselves really mean anything (and by then of course we'll actually have seen more on-field performance too).
#36 by Sleet // Sep 13, 2017 - 9:27am
So, the problem is my reading comprehension and not the secret formula? Really? My reading comprehension explains the #30 ranking for the Raiders' D, which did as much if not more to win the game Sunday than the Raiders' #2 ranked offense. Seems more like a narrative being pushed than anything we witnessed watching the game on Sunday? Heck, the Raiders' #6 ranked special teams (Giorgio, King and Patterson) did more to win the game than the a Raiders' offense (IMO). Stats deserve criticism when they don't reflect reality. Don't be an apologist. Perhaps consider improving the secret formula.
#49 by onemuddyshoe // Sep 13, 2017 - 11:44am
I don't think the sauce is secret at all.
I think FO is making a very simple but powerful argument: A defense is good (over the course of a season) when it keeps the opposing offense behind schedule consistently. Sure, in any given game, a defense can make a stop on 3rd and two. And sure, a good defense will make more stops on 3rd and 2 than a bad defense will. But over a full season, a better predictor of points allowed will be how often a defense prevents the other team from having favorable down, distance, and field position.
#54 by onemuddyshoe // Sep 13, 2017 - 1:15pm
Yes and no. Yes, sample sizes are an issue. But more than that, it's an issue with scale. At fine scales, a process can appear to be random (stochastic) but deterministic in the aggregate. I can't predict how many children my neighbor with have, but I can predict with a high degree of likelihood how many children will be born in the U.S.
#61 by Sleet // Sep 13, 2017 - 1:51pm
Another good point. But it, too, is too abstract. It also erroneously suggests, I'm guessing, that the Raiders' D's Week 1 performance will look better in the aggregate, and won't be weighted down by this poor #30 Week 1 score. Yet, the Raiders' D's poor #30 Week 1 rating happened for a reason. Will that magically change and not scue future weekly results?
What is predictive in the aggregate is owning the trenches on both sides of the ball. The Raiders did that against a really strong OL and front 7. Yet, while the Raiders' O got a nice #2 ranking (which I think is a little high), the D's #30 ranking is way too low regardless of how one tries to justify it (IMO). It just does not comport to what we saw on Sunday.
#66 by onemuddyshoe // Sep 13, 2017 - 3:04pm
DVOA isn't a Raiders-Titans Week 1 2017 game model. There is a game log for that, which will explain in great detail what happened, although it won't typically tell you why. If you watched the game and think you know why, you don't need this model at all. To criticize this or any model for being an imperfect representation of reality misses the point.
If you are interested in predicting how well a team will do in the future, I would think eventually, when more data accumulates, this model can bring something that looking at box scores, game logs, or even watching games with an intensity previously unknown to mankind can't.
#71 by Wikitorix // Sep 13, 2017 - 5:54pm
Well, first of all, the ordinal ranking doesn't factor into any calculation at all. The #30 ranking is irrelevant to everything. It's the (D)VOA that matters.
In answer to your question, yes, when the opponent adjustments are applied it will change these numbers. Not the numbers published in this article, but it will affect the numbers from this week used in DVOA calculations in the weeks to come.
#59 by Sleet // Sep 13, 2017 - 1:40pm
You apparently did not watch the Raiders/Titans game. In the first half, the Raiders' D's problem was not 1st and 2nd down. They got in numerous preferred down and distance situations on 3rd down. In fact, I'd guess that they had more 3rd and longs than defenses rated in the top 10 by FO after Week 1. Unlike other defenses, Norton refuses to blitz (a low 18% of the time) and the Raiders' LB struggle in coverage. But come the 2nd half, the Raiders got off the field more consistently. Heck, it would be interesting to see if any defense had more 3 and outs and redzone stops than the Raiders. My guess, far fewer than 29 other D's.
#64 by Anon Ymous // Sep 13, 2017 - 2:49pm
No, I didn't, I was speaking generically. Chris' comment below elaborates on how long drives - due to their higher number of plays - can outweigh drives where the defense is successful more than it appears at face value.
#60 by ChrisS // Sep 13, 2017 - 1:46pm
I did not watch the game. But looking at the drive stats (this is easy) Tenn had 9 drives, 4 scoring drives of 93, 75, 67 & 46 and a missed FG drive of 41 yards. These 5 drives consisted of 51 (mostly successful) plays, they also had 4 bad drives with 17 (mostly unsuccessful) plays. So about 75% (51/68) of the plays were successful that is likely bad defense. For more insightful analysis look at the play-by-play and see what sort of success rate the Titans offense had. Looking at ordinal rankings is sub-optimal they are closer to 22nd than to 31st.
#99 by Sleet // Sep 14, 2017 - 9:14pm
Two long scoring drives which totaled just 10 points in the first half, converting too many low percentage 3rd down plays.
Two seemingly shorter drives which totaled 6 points in the 2nd half after the Raiders' D made adjustments at halftime. I actually can't believe I just typed that considering the years of poor coaching in Oakland.
Not only does your analytic view (unaided by actually watching the game) divorce itself from this trend, but, given how the Raiders' DL was controlling the trenches in the 1st half, also ignores that the Raiders' D's far better statistical result in the 2nd half should have been expected by those not jaded by past Raiders' poor performances in the 3rd quarter. Instead, FO's ranking does not recognize how well Raiders' D played throughout the game, even though statistically it did not show up until they got in the redzone and the 2nd half.
That said, your post provides a good explanation of the low ranking. Thank you.
#107 by Richie // Sep 15, 2017 - 12:22pm
I keep hoping somebody at FO would chime in here to help explain Oakland's low rating.
I didn't see much of a game, but looking at the stats and the play-by-play it is hard to see why this would be a bottom-half performance. I suspect it has to do with a component of Oakland's defense. Such as an unusually high success rate on third downs which would lead to a win, but is not as reliable to sustain and predicts future struggles.
Though I trust in the system, because if the Raiders defense really is good, DVOA will do a better job reflecting it after a couple more games. Or, if the Raiders defense isn't as good as Sleet thought it was, the scoreboard will probably start reflecting it.
#109 by Raiderjoe // Sep 16, 2017 - 9:03am
sometimes DAVE and DVOA get together for a little druinking early in season. No way Raiders 30th best defense. I suually don't even notice the individual rankinsg. Usually only look at overall. aDjustment to happen when more games ar eplayed. Raiders defense ranking certainly will be moving on up[ to the eats side soon enough
#39 by Aaron Brooks G… // Sep 13, 2017 - 10:00am
How was the dumpster fire that was DET special teams only a DVOA of -0.9%?
Maybe the Lions should have Prater return kicks, too.
Incidentally -- what's the point of seeding scores with DAVE if DAVE isn't also used to seed DVOA?
#100 by LionInAZ // Sep 14, 2017 - 11:28pm
The coverage teams didn't allow much on returnable kicks, either, and they didn't incur the usual stupid penalties. The only screwups were the two in the first quarter.
I guess some credit goes to the offense for not getting pinned deep after the punter went down.
#51 by Mostly Anonymous // Sep 13, 2017 - 12:03pm
Looking forward to seeing how far the #1 LARM DVOA falls over the season. 99.6% is astonishingly high. Assuming same old sorry Rams, they could drop well into negative DVOA territory, meaning a drop north of 100%.
#65 by Anon Ymous // Sep 13, 2017 - 2:55pm
I was curious when the DVOA were so optimistic about NE's defense, but it's even more fascinating how much NE's projection drops on the heels of a single loss. Given how poor the division looks this year, I can honestly say I'd give NE a better-than-70% chance of making the playoffs if they start 0-4, let alone 0-1.
#67 by Bright Blue Shorts // Sep 13, 2017 - 3:42pm
Have to say that as poor as NE's defense looked in surrendering those big plays to an abducted-by-aliens Alex Smith imposter, they were still within minutes of winning the game against one of the AFC's better teams.
Compare that to say the Steelers scraping past the Browns, it's hard to see the Pats missing the playoffs.
#72 by dmstorm22 // Sep 13, 2017 - 5:55pm
DAVE thinks less highly of Buffalo, which makes sense.
Beating the Jets easily (by DVOA terms) may not look all that great a few weeks from now if the Jets are as bad as everyone thinks.
For Carolina, DAVE liked them pre-season, and why not? They weren't as bad as 6-10 last year, will likely be less injured. The young corners should get better. The team is a clear regression upwards candidate.
#73 by DezBailey // Sep 13, 2017 - 8:05pm
Week 1 BES Rankings are out - http://besreport.com/week-1-bes-rankings-first-issue-2017/
A bit more tempered by Week 1 results compared to DVOA except for in the case of the Jaguars. BES and DVOA, however, basically concur on the Chiefs and Ravens.
#104 by nat // Sep 15, 2017 - 10:43am
Sure, because eliminating garbage time skews the sample. Good teams get plays from good days removed from the sample. Bad teams get plays from bad days removed. That's the problem with manipulating data sets based on their content.
I've long wondered whether simply eliminating the fourth quarter would result in better predictive power for DVOA, since it would solve the garbage time problem without introducing a results-driven bias to the sample. That is, is eliminating the garbage time problem worth giving up 25% of your sample?
Of course, it would utterly fail as a descriptive stat: You can't correctly describe the KC-NE game without including the fourth quarter where New England got blown away. But it might be a good thing to try occasionally, when interesting teams are getting into a lot of blow outs one way or the other.
It would let you say things like:
"Team X might be playing better/worse than DVOA states, because their fourth quarter DVOA differs so much from their first-to-third quarter DVOA and might be affected by garbage time play calling."
#105 by dmstorm22 // Sep 15, 2017 - 11:16am
I do wonder if there is a lot of selection bias in how we think garbage time skews stats. We remember the games were QB X threw for 150 yards and 2 TDs when down by a bunch, but don't remember the times that QB X just continued to suck (The Colts last week, for example).
Given DVOA controls for situation and score, so 'garbage time' plays are judged against the average result in 'garbage time' plays, if the hypothesis that most defenses go into a bend-but-don't-break, soft cover-2 shell, then a defense that does that and gives up yards shouldn't actually get impacted that much.
This system would credit the teams that either stay aggressive or play that soft defense better than others and still keep teams from moving the ball or scoring.
#106 by nat // Sep 15, 2017 - 11:45am
I wonder the same things. Do teams really play differently in garbage time? If so, can this be handled by using garbage baselines for comparison, or does that just add more noise?
It's possible that the use of garbage time play averages as the baseline just makes things even more skewed. Or less skewed? Or just noisy?
I remember a Patriots game fairly early in a season where the score was completely lopsided, but both coaches seemed to decide that it was a good time to hold a realistic joint practice. Both teams continued to play real football with starters and play calls that would be appropriate for a close game. Excellent data for DVOA and the coaching staffs, except who knows what the baseline for VOA looked like?
They could as easily have decided to just kneel on both sides to avoid injury, or to field their backups for game experience. I have no idea what the baseline coaching decision is in that situation.
Anyway, it's not likely to be a big deal. My suggestion was simply to use fourth quarter DVOA vs Q1-3 DVOA as a way to check for garbage time issues when they are suspected. The answer could be
"Team X's fourth quarter DVOA is in line with their other quarters, so we don't see any reason to worry about their high number of plays in so-called 'garbage time'."