Rams Up to No. 3 in Weighted DVOA

L.A. Rams celebrating
L.A. Rams celebrating
Photo: USA Today Sports Images

NFL Wild Card - The Los Angeles Rams put up over 100% DVOA for their huge blowout win over the Arizona Cardinals last night and move up to third in weighted DVOA this week. The Rams now come out as the highest-rated remaining team in the NFC. However, they would be essentially tied with the Green Bay Packers if we removed the games where Green Bay had to use a backup quarterback. Also, the Rams are high in part because of weighted special teams being at No. 1, and as we know, special teams is the least consistent of the three phases of the game.

One important note for the Rams is that Tampa Bay falls to sixth in weighted DVOA despite the big win over the Eagles. Tampa Bay is the only remaining team in the postseason with a weighted DVOA lower than its total DVOA (including the playoffs).

A reminder again on how playoff DVOA works:

  • All 32 teams are ranked, whether they made the playoffs or not.
  • Teams are ranked in order of weighted DVOA, not total season DVOA. Since weighted DVOA is meant to lower the strength of older games, these ratings do not include Weeks 1-3, and Weeks 4-11 are somewhat discounted.
  • The ratings listed do not include the adjustments used in the ratings for our playoff odds report. For example, Green Bay moves up significantly in those ratings because I removed Week 18 as well as the offense from Week 9 (Jordan Love).
  • Only weighted DVOA is listed for offense, defense, and special teams. Total DVOA is also listed, but one game doesn't change much in a 17-game sample so these ratings will be similar to those from the end of the season.
  • Teams which did not play in the wild-card round are treated as if they had a bye week. (That includes both the 18 non-playoff teams and the two teams with byes.)

The numbers below are slightly different from the numbers we ran for weighted DVOA yesterday because we discovered an error in the calculation of weighted special teams including the playoffs. The changes are small but they do result in Buffalo coming out No. 1 in weighted DVOA, slightly ahead of the eliminated Dallas Cowboys.

Yesterday, I ran a list of the top playoff games in DVOA history. In the comments, reader "andrew" asked where the Patriots came out on the opposite list, a list of the worst playoff games in DVOA history. For various reasons -- special teams, penalties, opponent adjustments -- one team's DVOA is not a mirror image of the other team's DVOA. So for comparison purposes, here is the list of the worst playoff games by DVOA since 1983. The Patriots come out seventh.

Worst Playoff Games by Total DVOA, 1983-2021
Team Year Week Opp. Total
DVOA
Off
DVOA
Def
DVOA
ST
DVOA
Score Opp
Rank
IND 2002 WC NYJ -128.8% -54.8% 47.5% -26.6% 41-0 6
MIN 2000 CCG NYG -124.1% -78.2% 39.2% -6.8% 41-0 11
NYG 1993 DIV SF -109.1% -77.9% 29.2% -2.0% 44-3 2
NO 1987 WC MIN -107.3% -69.4% 15.4% -22.5% 44-10 11
MIA 1999 DIV JAX -105.2% -76.0% 30.4% 1.2% 62-7 2
DEN 2011 DIV NE -103.9% -59.9% 37.4% -6.7% 45-10 4
NE 2021 WC BUF -98.3% -8.3% 84.6% -5.5% 47-17 2
NYG 2005 WC CAR -97.8% -67.1% 25.7% -5.0% 23-0 9
LARM 1983 DIV WAS -97.2% -45.1% 28.6% -23.5% 51-7 1
SD 1992 DIV MIA -95.9% -71.1% 14.1% -10.7% 31-0 11
NYG 2000 SB BAL -95.2% -96.2% -11.1% -10.0% 34-7 3

It's kind of amazing that the Giants end up with negative special teams in the worst Super Bowl performance ever, despite the Ron Dixon 97-yard kickoff return touchdown. But Dixon also fumbled away a kickoff return, and the Giants allowed their own kickoff return touchdown to Jermaine Lewis.

* * * * *

To save people some time, we remind everyone to put their angry troll hatred into the official zlionsfan angry troll hatred Mad Libs form:

<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>

If you are new to our website, you can read the explanation of how DVOA is figured here. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.

There are no adjustments here for sitting starters in Week 18, although we do adjust the ratings that we use in the playoff odds report.

Teams in yellow are still alive in the playoffs. Teams in gray lost this past weekend.

RK TEAM WEI.
DVOA
LAST
WEEK
W-L WEI OFF
DVOA
RK WEI DEF
DVOA
RK WEI S.T.
DVOA
RK TOTAL
DVOA
RK
1 BUF 34.2% 3 12-6 18.9% 4 -14.9% 5 0.4% 18 33.3% 1
2 DAL 34.1% 1 12-6 7.6% 9 -21.4% 1 5.1% 4 29.6% 2
3 LAR 29.0% 7 13-5 3.9% 13 -16.4% 3 8.7% 1 25.3% 4
4 SF 27.6% 5 11-7 14.0% 7 -16.3% 4 -2.6% 27 20.7% 5
5 KC 26.3% 6 13-5 19.1% 3 -2.3% 14 4.8% 6 18.4% 6
6 TB 26.1% 4 14-4 21.2% 2 -6.5% 9 -1.6% 26 28.5% 3
7 NE 18.2% 2 10-8 14.1% 6 -5.4% 10 -1.3% 25 16.9% 7
8 SEA 15.7% 9 7-10 10.6% 8 -1.1% 17 4.0% 8 11.7% 9
9 GB 15.1% 8 13-4 22.4% 1 2.2% 22 -5.1% 31 11.8% 8
10 IND 12.5% 10 9-8 1.9% 14 -10.2% 6 0.4% 17 9.8% 10
11 LAC 10.1% 11 9-8 14.6% 5 5.7% 26 1.2% 13 8.7% 11
12 TEN 4.9% 13 12-5 -4.7% 20 -9.6% 7 0.0% 21 -3.0% 20
13 CIN 4.4% 14 11-7 5.6% 11 4.6% 25 3.3% 10 1.5% 17
14 DEN 2.4% 15 7-10 5.7% 10 -0.6% 18 -3.9% 29 -1.5% 18
15 PHI 1.5% 12 9-9 4.4% 12 3.1% 23 0.1% 20 2.3% 16
16 NO -0.1% 19 9-8 -19.8% 30 -19.0% 2 0.7% 15 4.4% 14
17 MIN -0.5% 18 8-9 -1.2% 15 0.3% 21 1.0% 14 2.6% 15
18 CLE -0.7% 17 8-9 -1.5% 16 -4.9% 11 -4.1% 30 4.8% 13
19 LV -2.9% 20 10-8 -3.9% 17 -0.5% 20 0.5% 16 -5.9% 21
20 MIA -4.5% 21 9-8 -9.1% 22 -7.4% 8 -2.8% 28 -10.9% 24
21 CHI -5.7% 22 6-11 -12.3% 24 -2.6% 13 4.0% 7 -10.1% 23
22 ARI -5.9% 16 11-6 -8.8% 21 -4.2% 12 -1.3% 24 7.8% 12
23 WAS -9.6% 23 7-10 -10.9% 23 -1.6% 16 -0.3% 22 -10.1% 22
24 NYJ -10.7% 25 4-13 -4.0% 18 14.9% 32 8.1% 2 -18.5% 26
25 BAL -12.1% 24 8-9 -4.2% 19 13.0% 31 5.1% 3 -2.4% 19
26 PIT -13.8% 26 9-8-1 -16.4% 26 -2.3% 15 0.3% 19 -11.4% 25
27 ATL -19.8% 27 7-10 -15.5% 25 8.1% 29 3.8% 9 -29.5% 30
28 HOU -20.0% 28 4-13 -18.8% 27 6.1% 27 4.9% 5 -24.3% 28
29 DET -27.0% 29 3-13-1 -19.4% 29 8.7% 30 1.2% 12 -27.2% 29
30 JAX -31.4% 30 3-14 -18.9% 28 7.0% 28 -5.5% 32 -31.1% 32
31 CAR -33.3% 31 5-12 -29.5% 31 3.4% 24 -0.5% 23 -24.0% 27
32 NYG -43.1% 32 4-13 -45.7% 32 -0.6% 19 2.1% 11 -29.7% 31

Click here for a look at full-season DVOA with offensive and defensive splits.

Next, let's look at all the single-game ratings from wild-card weekend.

DVOA (with opponent adjustments)
TEAM TOT OFF DEF ST
LV -22% 1% 23% 0%
CIN 30% 24% 6% 13%
NE -98% -8% 85% -5%
BUF 138% 108% -25% 5%
PHI -33% -37% -10% -5%
TB 47% 9% -30% 7%
SF 35% -13% -45% 2%
DAL 13% -27% -26% 14%
PIT -30% -15% 9% -5%
KC 39% 38% 8% 8%
ARI -65% -63% 9% 7%
LAR 104% 28% -61% 16%
VOA (no opponent adjustments)
TEAM TOT OFF DEF ST
LV -20% 0% 20% 0%
CIN 30% 22% 5% 13%
NE -127% -30% 91% -5%
BUF 122% 102% -15% 5%
PHI -57% -41% 11% -5%
TB 42% 13% -22% 7%
SF 11% -23% -32% 2%
DAL -7% -34% -13% 14%
PIT -50% -15% 30% -5%
KC 46% 34% -4% 8%
ARI -83% -75% 15% 7%
LAR 87% 19% -53% 16%

Comments

76 comments, Last at 20 Jan 2022, 1:33pm

2 2000 GIants

It's kind of amazing that the Giants end up with negative special teams in the worst Super Bowl performance ever, despite the Ron Dixon 97-yard kickoff return touchdown. But Dixon also fumbled away a kickoff return, and the Giants allowed their own kickoff return touchdown to Jermaine Lewis.

The Giants also lost yardage on every first-quarter punt exchange when neither team could move the ball, so the Ravens were running plays in Giants' territory before they ever got a first down.

3 Wait

Why doesn't this chart match the one on the Team DVOA page, which still shows DAL/NE 1 and 2 for weighted DVOA?

4 Playoffs vs Regular Season

In reply to by Treima

Make sure you have the drop downs on the top set to "Playoffs with Regular Season" instead of just "Regular Season."

6 Thank you

I had thought the Vikings 41-0 loss would be up there.   Travis noted that part of their (and the Colts) rankings were lack of garbage time making up some dvoa, which the Patriots seem to have benefited from somewhat.

Anyway, until we get around to having time agents to get us the 1940 season DVOA, I guess the answer is the Colts.....

7 I thought the first Buffalo…

In reply to by andrew

I thought the first Buffalo/Dallas Super Bowl would be there too, considering we fumbled eight times and the only positive play from that was Beebe saving a touchdown off one of them.

9 NE ran up the score in 4th…

In reply to by andrew

NE ran up the score in 4th qtr against NYJ/CLE/TEN/JAX when leading by 3+ possessions and the other team's offense unable to do anything. Seems to be one of the reasons why DVOA believes NE and DAL are better than public opinion (DAL ran up the score against NYG/ATL/WAS/PHI's backups)

62 Lol

If you define a blowout as a 3+ score margin of victory (i.e. 17+ points), every remaining playoff team has at least 3 such games in their portfolio. The Bengals have 5 such games. The Bills have 8. You'll note, I hope, that the Bills were the team that defeated the Patriots, which would suggest that DVOA does in fact correctly note that blowout victories do actually correlate with team quality. FYI, the Patriots had 7 such victories, which would suggest that the Bills were the somewhat better team and should have been favored in the playoff match up.

8 Issue with the DVOA tables

The Single Game DVOA by Team/Week table lists Weeks 1-3 and 18 twice with the second incorrectly listing the opponent. For example: https://www.footballoutsiders.com/stats/nfl/single-game-dvoa-by-team/2021/NE Week 1 is listed twice, the second time being against "WAS".

11 RE: NFC teams

I do think the four best NFC teams are still playing.

 

I am not here to throw any shade at all this hard work. Just don't believe DVOA is able to incorporate intangibles that are in Dallas' case legitimate and significant.

13 Specifically

In reply to by big10freak

Specifically, I believe DVOA ignores most penalties.  And Dallas commits far more penalties than most of the top teams. 

23 The funny thing about this…

In reply to by RickD

The funny thing about this is that Mike Tanier's recent article pointed out that Dallas's opponents get more penalties called on them, too. He said something like "Dallas's games just happen to have more penalties" - and then I realized, why does that have to be true?

The officials are human. It's totally possible that when they see one team (Dallas) being stupid-sloppy all over the place they end up calling things tighter on the other side, too.

I mean, yes, Dallas had the most offensive holding calls - 13 above league average - and sure, those are judgements. But they also picked up ~10 penalties over league average from neutral zone infractions and defensive offsides, and those aren't judgements. Those would get called by any crew.

36 To the question of whether…

To the question of whether DVOA should consider penalties: could it be that penalties don’t add predictiveness to the model for most teams, but for teams who commit a lot of penalties it does? Could that logic be applied to any number of areas of the game - where there’s a general predictive trend, but when introduced to a unit that does things totally differently than the rest of the NFL, DVOA may not adequately understand the extrapolation needed for the outlier? If so, is that fixable or is that just the reason why we still need to consider factors beyond just the DVOA number? 

Specific to Dallas in this game: they committed 5 penalties and the 49ers committed 7 in the 4th quarter. 6 were presnap. And I don’t recall every penalty, but there was a lot of stuff that was egregious, and there were plenty of boneheaded plays that weren’t penalty related.  Maybe they should quit complaining about being targeted and instead work on some fundamentals so they don’t lose their minds when it matters most. 

49 I've been wondering about…

I've been wondering about this too. It's another thing that brings me back to the "two models" the VOA/DVOA that describes what happens and should include everything and the PDVOA the model that throws out data and changes weights to give better predictive results. 

Some penalties may not be predictive so they are thrown out. I kinda feel that turnovers can fit this too. Turnovers are hugely descriptive. They have a massive impact on the game, but are past turnover predictive of future turnovers? Aaron Rodgers has played in 213 games, started 206. He has 17 games with multiple INT. He has 14 more games where had multiple INT+Lost Fumbles. He threw 2 INT against NO. GB had a horrendous rating for that game (deservedly so). He threw 2 INT over his next 14.5 games (missed one game sat 2nd half of another). So how much predictive value did those 2 INT against NO have? They don't really matter now because of the weighting formula, but what if that game happened in week 12. About 8% of games he starts will have multiple INT (15% he will have 2 or more turnovers).

Should those have more predictive weight than DAL averaging 10 penalties a game (or whatever the number was)? Right now they do. By a lot. Those 2 INT should absolutely be there when looking at the value over average that tells us what happened in the game. I know they've run the models a few times. But I keep thinking there is more we can capture from the play by play data. I also think they may have the backend data in better shape now which should allow for faster modeling and testing than what they could do in the past making some of these questions easier to answer.

I mention turnovers just because of the massive leverage and my recollection that year to year for sure and possibly game to game they are not very stable. But that recollection could be wrong. I know with my health issues my memory is slipping faster. So turnovers may be valued completely right when you describe why one team beat another. But that value may be very different when you look at the chances of one team beating another because they might be more random than the formula says. Penalties might be more descriptive and more predictive than they currently are.

As you ponder it may also be that they are more predictive and/or descriptive for the outliers and hence should only apply to outlier game games.

It does feel a bit arbitrary that some penalties aren't included and others are, especially if teams that commit penalties are likely to commit similar amounts game over game. As a fan they feel like they have a stronger game to game correlation than turnovers. Though some of that with turnovers might be because they are less frequent now than when the system was built. When even the best teams averaged a turnover or more a game and going 3 games without one was a big deal they may have been more strongly correlated. Now when it seems like a team going 4 games without a turnover isn't that big a deal it feels like the correlation is a lot lower even if the impact on game outcome is as high or higher than ever.

 

Also I still dream of having a "running AGL" that uses some snap count formula. Like you can figure out starters and key back-ups based on average snap % played over the last 3 years. You can then use that weight results if there are a lot of injuries, probably with some positional value attached. Possibly treating the offensive line as a single player with 5 inputs so that missing one key player replaced by a back-up with a a lot of previous snaps doesn't bring it down as much but when you now have 2 back-ups and one of them hasn't played at all the value drops even more. You'd have to keep rolling the current season into it. The Packers may not have wanted to play Yosh Nijman at LT but after he had gotten 8 starts in the last 14 games his replacement status on the line wasn't as impactful. I get that would be trickier and might even delay things too, but I do think it's more data that is on the same level as the play-by-play, that we have multiple years of data for, that could help the system. But that is a much bigger ask. Though it should save the manual adjustments to playoff odds.

 

 

51 In general, I am excited…

In general, I am excited about the thought of having something like that. However, when I think about how to actually build it, it gives me pause. In short, I think it might be too complex to capture a ton of it. I do think there are probably other items that could be improved, and I suspect penalties fit into that category. 

As an example, here are some situations that make me worry:

1. Frequently young engineers try to quantify minute inputs. They’ll try to capture and include preload, assembly stress, inclusions in a casting, etc. Though well-intentioned, they haven’t considered that those minute details are already captured by safety factors and validation standards - expressly because it’s too difficult and expensive to capture everything for every design. Focusing on too many of those inputs can lead to an overall worse model or a model where teams are being evaluated based on different criteria, where the model may start feeling arbitrary and the analysis article becomes impossible to write. 
2. Can we accurately predict outliers? By definition, they don’t match a linear model. So how do we quantify that it’s an outlier and not just the last data point, without arbitrary or purely subjective means. And since, by definition, outliers don’t match the usual model, how can we adjust for them when we have such a small sample size to rely on? 
3. Can we do this through prediction, or is it only possible through post-processing? For example, when the next Lamar Jackson comes around, is there anything within the data set that could be used to realize that he’s different and that an offensive DVOA adjustment is necessary to capture it? Some of these things might just be so specific to a single player that they can never be used to improve a model. 

I think about these things… good luck Aaron! 

53 I agree with all that. I…

I agree with all that. I also have concerns outside football about blindly accepting models without understanding the inputs and what they are measuring. The understanding about why that is an issue is widely understood by people that post here which is something that I like about discussions on this site. Having it all be about football where it doesn't really matter, so is low stakes, is even better.

I do think penalties are the easiest thing to check out again. It would be less effort than many of the things I dream about* and it would be "backward compatible" with all the play-by-play data they have. It also can be done even if their data conversions aren't done yet. Checking game to game correlation on penalties is not hard adding their yardage value into the current formulas shouldn't be hard, seeing if that helps the model is not hard. Time consuming depending on the tools sure, but it's very in the wheelhouse of what they have been doing with the model all along so shouldn't be hard. I completely understand why they haven't done it yet, and may still not do it anytime soon going forward because time is hugely valuable and limited.

But I know they have been working on back end data storage and formatting. I get excited about that because a lot of my career has been doing just that, taking data and making it easier to analyze and seeing that you could now actually try and answer questions you couldn't before. I know work is work but you don't make a career of something for as long as I did if you really hate what you work on. Of course with being able to ask new questions you may find that the data still doesn't help anything. But you now know you either need different data, that the current data already covers it in ways you didn't realize, that you are asking the wrong question, etc. You still get some value, negative results are not worthless. Some of things I've been wondering about were questions that were simply too hard for them to even try and analyze with the tools and they had. Remember this all basically started with play by play and spreadsheets (more to it than that but it conveys the idea). There are a lot of tools that simply don't work well or at all when your data is just sitting in spreadsheets. But it's not an easy task to convert it. If it was I wouldn't have been worth the money multiple companies paid me to do the jobs I did and I was sometimes underpaid. Sometimes that was rectified by a bonus based on the increased profitability that was easily traced to my work sometimes it was rectified by me moving to another job, etc.

Anyway yeah it's mostly just geeking out over data analytics.

As to the injury adjustments. I'd still just love to have that based on snap counts in some way for them to be able to quickly provide a running total of AGL. I think even without it being worked into a formula it would still be valuable additional data. Just having a quantified set of numbers to help your analysis of what the model is saying is nice.

 

* What I really dream about is what you could do with the data that AWS is collecting and giving us some glimpses of what it can do. But with all the player tracking data they have, in a format that they can quickly convert to graphics, you can actually see routes and stuff. You can way more easily build a model to rate how a cornerback is doing in coverage and get a better value on a shutdown corner because you could calculate if the receiver was even an option or not with some confidence because you will know how fast they are and how much space is open etc. You could suss out some level of play calls and mistakes, etc. Some teams already do some level of this with game planning, and they always have with manual film study, but that data is now in forms that can simplify and/or enhance that process. The commercials AWS runs tout using it to help analyze player safety which is awesome too but there is so much more you can do with that.

68 It's not that exactly that…

It's not that exactly that penalties aren't predictive. It's more that differences in penalty rates aren't enough to be predictive, mostly because they're just so infrequent. And then players with high penalty rates are, well, bad, and so they aren't kept around typically. It's tough to include penalties because you can't just do it like other plays: the majority of it is just bad luck.

The dominant penalties in the league (in order) are offensive holding, false start, DPI, unnecessary roughness, and delay of game. Of those, only holding and unnecessary roughness aren't included in DVOA already.

With holding, the worst player on the worst team (Connor Williams) averaged around 4 holding calls/yr-equiv before this year. He had 11 this year. Going from an average of 4 to 11 in one year isn't ridiculous - if his true rate was around 6, it's an outlier, but definitely not significant. Second is Udoh, a guard for the Vikings in his first starting year with 9. Third is Simpson for the Raiders, in his first starting year also with 9. Then Taylor for the Jaguars, in his third year for the Jaguars, with 8 - he averages around 6, so obviously 8 is totally normal.

But the thing is that the typical holding rate in the league is around 3-4/yr anyway. For everyone. So even if the worst players in the league are like 6/yr, it's just not enough to be able to tell. And usually the worst players in the league either disappear, or get better.

The league average was 21 holding calls. You'd expect a standard deviation of like ~4.6 with that just due to counting statistics. Standard deviation among all the teams was 5.6. Even if that excess was entirely skill, it means counting statistics dominates over skill pretty noticeably. And because teams in general get rid of bad players, even the 'skill' part's not consistent. Now couple in that there are referee rate variations and there's very little room left for skill. 

The difference with interceptions is just that there are far, far more of them, per player, and because QBs aren't as replaceable, high interception rates don't in general go away from year to year. But I will say that interceptions are interesting because when FO was founded they were far more common. 

67 Penalties

The biggest issue with modeling penalties is figuring out how much penalization reflects the team getting penalized and how it reflects the referee crew. Obviously, penalties have a negative effect on outcome, but variance among referee crews means that the season-long effect of penalties is essentially random (i.e. non-predictive). The only thing I can suggest is that pre-snap penalties probably have some predictive negative correlation with outcome, but it's hard to say whether that should be considered by rate of incidence or at a threshold amount (i.e. teams that hit a certain number are probably poorly coached).

69 The only real judgement call…

In reply to by coltsandrew

The only real judgement call penalty that matters (from this perspective) is holding - unnecessary roughness, while a huge effect, just happens too infrequently. It's less than 1/game. And DPI is already included. 

With holding, the crew variation on a game-by-game basis is much less than just counting statistics. You get ~2.5 holding calls/game, and the crews range from like 3 to 1.5. So in general the effect on a game basis is very small. On a season-long basis it can add up, and it's probably responsible for the majority of the variation outside of counting statistics. Which leaves very little room for skill.

 The only thing I can suggest is that pre-snap penalties

False starts are already in DVOA.

26 They were a bully team like…

In reply to by big10freak

They were a bully team like New England. NE and Buffalo were high variance teams where a blowout by one side seemed abnormally high. A bad Josh Allen game could lead to a ball control 24-10 loss where NE had the ball for 40 minutes and Allen threw a bunch of picks. SF was a very good team with low variance. The division/non-division split on the SF-Dallas game heavily favored SF. SF reminded me of the 80's Giants teams that would surprise the weak division dominating 49ers team during some of their early playoff exits in the dynasty. SF thoroughly dominated the trenches in that game and probably would have won by multiple scores without making multiple huge mistakes by coaches and players in the second half. The 49ers just got done beating the Rams and Cowboys on the road while playing sloppy games where they had a couple really bad halves of football. That stretch from the beginning of the second half against LA and the first half against Dallas was some of the most dominant football any team has played this year. If they put 4 good quarters together in GB and they have good chance of blowing them out. They do another one of these sloppy bipolar games and it will be another nail biter that favors GB because they have a much better QB and coaching staff than the Cowboys. If the worst case scenario complete bad game happens, we will see GB resting starters in the 4th quarter up 28-3 while Jimmy throws his 3rd interception. I have been watching the 49ers all year and I am feeling another one possession game that is won by the last team with the ball. If Bosa plays and is > 90% Aaron is going to have a very tough time dealing with the SF pass rush. I actually think there is a decent chance this game is completely dominated by the SF lines. They just pounded 2 of the best O lines in the entire NFL and with Moseley back from and injury and the rookie Thomas ascending at CB this is far and away the best the defense has been all year. These teams are going to look nothing like they did in their first game. Rodgers is going to have to dominate the 49ers secondary deep and outside the numbers to win. If they can block SF I think he will do that. If they get whooped like the Rams and Cowboys did, they are going to need to decisively win the TO margin to win.

12 Packers week 18

You removed the Packers week 18 game against the Lions when making adjustments for playoff odds. 

The Packers only rested their starters for the second half of their game against the Lions.  They lost the first half 17-13. Seems like that should count for something.

I can’t post a comment on POR so I’m doing it here…

15 Maybe something

In reply to by d3s3rtf0x

But then again they werent playing for anything and were probably vanilla (looked like it). Not like they'll go on the road like that. 

17 It’s a real data point…

In reply to by ImNewAroundThe…

It’s a real data point unlike the second half and the offense for the Jordan Love game.

 

the Lions?! scored 17 points on their defensive starters in a half and DVOA says their defense is weak anyway.

when they play the Bills or the Chiefs it will be their defense that matters

22 Didn't disagree

But I wouldn't put too much into it. Still weren't likely "trying hard" and that was without Aaron Jones (rbsdm) and 1st team all pro Devondre on defense.  Along with what looks like Zada and Jaire coming back (former all pros).

I might weigh it but not heavily like any other non KC half imo.

37 Agreed. The data should lead…

In reply to by ImNewAroundThe…

Agreed. The data should lead. The belief that the Packers defense is bad is not a valid reason to include the first half of that game. 

for those who watched that game, it looked like the Packers were treating it like a preseason game. They got Bahktiari some reps and ran plays specifically to test some things and then they put in backups. 

39 It’s not a belief, their…

It’s not a belief, their defense is bad, for instance DVOA rank == 22.

Their starters played the first half not one series.
 

You’re talking about their offense (Baktiari).  

Also, ignoring data to fuel confirmation bias is not a valid reason to throw it out.

if your team performance model is based on real data there is bound to be variability in the dataset like having a backup qb start a game.  It is certainly conceivable Rodgers will need to be spelled for 3 minutes (1/17th of a game) in a game due to minor injury or other random event.

It makes sense to base your statistical model on what actually happened rather than what would happen in a perfect world where players never get injured and accidents don’t happen.

 

 

48 I guess you think that the…

I guess you think that the only difference between that game and every other game was in the 2nd half when players sat. If you watched the game compared to others, you would realize that everything from the preparation to the play calls was different. 

Garbage data is so much worse than no data. Using week 18 for the Packers is equivalent to using the preseason games in the DVOA calculation early in the year. It’s just not good. 

43 Yeah.  It’s hard to account…

In reply to by ImNewAroundThe…

Yeah.  It’s hard to account for atypical events.

 

AR is certainly at risk of missing a game from COVID as well so maybe there’s point for including JL’s game.

 

ATBE if everyone’s healthy they have home field and would be a tough out

45 12 doesn't get tested til like next month

Since it's 60 or whatever days after testing positive the first time.

But most teams lose their starting QB they're screwed. Having the one seed locked up prior to w18 despite Adams missing a game, Rodgers missing a game, Zada missing all but 1 game, Bak all season, Jaire but a few games, bodes quite well for them being much stronger than any other week of the year. Good sign of depth.

Include it, just realize whatever it's telling you isn't quite what it'll look come GameDay. Barring any freak occurrences of course. Zada, Bak, Jaire, Campbell, Jones will play the entire game likely come Saturday unlike at Detroit.

14 Patriots' worst playoff game

Having watched both, I'll still go with Super Bowl XX.  A bit surprised it didn't make the list. Or any of the other late '80s/early '90s Super Bowl blowouts, for that matter. 

21 Opponent Strength

I guess the opponents strength was strong enough to bump them up just enough.

 

But yeah, I'm curious to see the Pats DVOA in that game, or Denver's in the 55-10 loss to the Niners.

32 I think it was Saturday night

I also watched both. Super Bowl XX was a blowout, but the '85 Bears were legendary, and the '85 Patriots were a bit of an afterthought and considered lucky to be there. They were 7th in the league at 17.7% DVOA that year, but their opponents were 1st at around 53%, which is insane. This year's team was much higher in the relative ranks, was much higher entering the game, and was playing a team that was ... well, not on the same level as the '85 Bears. Also, this team was coached by the famed Bill Belichick, defensive genius! And they got absolutely destroyed in historic fashion.

Anyway, there's no right answer here. It's like trying to decide between 51-3 and Saturday night's game. They were both very, very similar.

52 depends on the question

I think the Pats played worse in XX.  That's a different question than "should their DVOA be higher or lower?"  Certainly the '85 Bears were a better team than the '21 Bills (no offense intended).  But I felt that only the defense was terrible on Saturday, even though they were remarkably bad for a playoff team.  I'd give the offense a 'C' grade, which is higher than either the offense or defense deserved in Super Bowl XX.

56 I agree. I commented in the…

I agree. I commented in the game day thread that part of what I was enjoying so much about that game was that NE wasn't playing that badly. As you point out that was even more true for the offense. Take away the to INT and I'm pretty sure the DVOA is getting very close to 0% which is not quite saying that they did about what you would expect them to against that defense based on what how both units had played in the past but it's not completely wrong to think way. Also it's not hard to take the INT away from them. The Hyde play on that first one was amazing and while it's possible Jones should have been able to see that potentially happening it didn't feel like a bad decision or throw. The 2nd one was a bad throw and decision he should have read that coverage and the throw needed to lead the receiver even more than he was, but any tipped pass being caught for an INT is still a bit of a fluke. Change them both to incompletes or one to a TD and the other to an incomplete and then consider what the offense did.

Even on defense they weren't completely killing themselves. Buffalo just didn't really screw up at all to give them a chance to leverage anything. Even the best defense in history still tends to need the offense to mess up a bit in order to take advantage of it. They defense could have been better on Saturday, there were obvious mistakes. But I'm not sure there were more than any other NE game. I didn't see them much this year I think just 4 times including Saturday but they didn't look much worse than I remembered. Buffalo just looked better, making the right call for the defense, executing the calls about as well as you can hope. Compare it to any other game this weekend and despite the scores the Pats played better by eyeball test than pretty much any other losing team. The Bills just played the best game by any winning team. Sucks for NE but it happens.

60 Right, depends on the question

Which team played worse? Probably the SBXX Patriots, judging by the eye test. Which loss was more embarrassing? Probably the one a lot of people expected them to win, only to see them pull an emotional no-show.

And yeah, no offense taken ;-)

16 and, wow

That 2000 NY Giants was involved in two of the top ten games on the list!  (Not losing both, obviously.)  

18 SB20

I'm surprised this beatdown by the Bears didn't make the list.  It certainly included the worst QB performance in any Super Bowl, though Tony Eason only played in the 1st half: 0-for-6 passing and 3 sacks, one a strip-sack.  Maybe the Steve Grogan-led garbage time TD (to make the score 44-10) pushed the game down the awfulness list?

19 I remember watching that NYG…

I remember watching that NYG-SF game in 1993. That is right up there with the SF-Dallas NFCCG in 1994 and 2012 SF-GB divisional where Kaep went off and destroyed the Packers. The Giants have been a villain for generations of 49ers fans in the playoffs. It was definitely the end of the line for the stars of the NYG dynasty and the 49ers were reloaded for another 6 year run of dominance. But ending Phil Simms' career after his surprisingly good PB come back tour was awesome. If SF some how pulls off road wins against GB and LAR to get to the SB I wouldn't even care if we got blown out. That would be the best NFC playoff run of my life taking out all the historical rivals in one playoff run on the road. This SF-GB game should be really good. All 4 of these match ups look like they are going to be great this weekend.

20 worst quarterback performance in a playoff game

While seeing some really awful quarterback play these playoffs, we debated (on the unofficial FO discord) on what was the worst quarterbacking performance in a playoff game.     While we don't have dvoa numbers on older games, we went over a lot of truly awful games.   I think the 1979 Tampa Bay Buccaneers loss to the Rams in the NFC Championship game deserves special mention:

starting QB Doug Williams went  2 of 13, for 12  yards, 1 interception.   He didn't have any rushes.  An amazing 0.9 ypa.

Backup QB Mike Rae, also went 2 of 13, but for 42 yards, 1 fumble.  Also 1 sack.   He had 1 rush for 0 yards.  3.2 ypa.

But RB Jerry Eckwood was 1/1 for 42 yards, their biggest play of the game by far.   (he also had 2 rushes for 5 yards).  42 ypa.   Combined their quarterbacks went 4 of 26 for 54 yards and a pick.

25 Frank Filchock deserves some…

Frank Filchock deserves some attention, for his determination to hit the wide open DB in both 1940 and 1946. That 1940 game was a doozy, where Sammy Baugh was bad, but was still an entire universe ahead of Filchock and Zimmerman.

The 1955 title game, where the Rams went 11-28, 1 TD, and 7 INTs.

Williams also stunk up the joint in 1982.

 

Combined? That 2018 Bills-Jags game might be it.

30 I know you're focusing on…

I know you're focusing on individual game performances, but Scott Mitchell deserves special mention for looking somewhat competent in the regular season (look up his DVOA/DYAR in 1995!), but then laying stinkbombs in the playoffs.

Against PHI in '95 WC game: 13/29 155 yards, 1 TD 4 INT, 2 sacks

Against TB in '97 WC game: 10/25 78 yards 0 TD 1 INT (benched for 36 year old Frank Reich, who immediately looked way better).

In two games, 42% cmp, 4.3ypa, 0.02 TD%, 1.0 INT%, 2 sacks, and his team got crushed in both games, primarily due to his suckitude.

40 Before my time (I was 7, and…

Before my time (I was 7, and not interested in football yet).  My dad tells me that Danielson was dreadful, but because of Billy Sims awesomeness, the Lions still came within a choked FG away from knocking off a strong 49ers team.  

Looking at the PFR boxscore, 5 interceptions and 2 sacks is bad, and 6.2 Y/A is solidly below average, but a lot of quarterbacks have had far worse playoff games.  Jake Delhomme's 2008 Divisional Round implosion is worse, and his team got blown out.

24   One important note for the…

 

One important note for the Rams is that Tampa Bay falls to sixth in weighted DVOA despite the big win over the Eagles. Tampa Bay is the only remaining team in the postseason with a weighted DVOA lower than its total DVOA

I don't think that's that surprising, because I think Tampa's struggles have mostly been injuries, and those mostly happened later in the season. They were pretty solid early on.

After two days remove from the Eagles game, I'm more of the opinion that that game said more about Tampa's defense than Philly's offense. Getting Barrett and David back seriously changes the face of that defense.

27 The injuries to TB make…

The injuries to TB make their match up with the Rams a lot worse on paper. Gronk and the RBs are going to have to go off to win. Donald is a lot better than Hargrave on the Eagles and he started wrecking that game. In a game against a competent offense TB would have been in serious trouble last week. There is a decent chance the Rams dominate TB because of some serious personnel advantages. They are going to have to pressure Stafford early and often or he could dice them up. It will be interesting to see how hard TB leans on the quick passing game. LA is soft in the middle so that is where you have to live. Evans is probably going to do nothing on Sunday. 

28 In a game against a…

In a game against a competent offense TB would have been in serious trouble last week. 

Yeah, I'm not so sure - Philly didn't look terrible all on its own. Half the reason Hurts had so many issues is just that there wasn't really any place for him to go. The announcers kept focusing on DeVonta Smith being left "one on one" but a lot of those were disguised coverages and Hurts just avoided them because they weren't clear. Tampa Bay's defense was much more complicated than it had been the past few weeks.

That being said, the Rams passing game is almost a mirror of the Eagles, so I basically agree with the end result of the statement - this just isn't a good matchup for Tampa Bay on defense. Definitely nowhere near as good as the Eagles were with David coming back.

Donald is a lot better than Hargrave on the Eagles and he started wrecking that game.

Yeah, that was a lot more about the unsettled offensive line situation. That basically depends on how Wirfs and Jensen heal up.

29 It could just be the Rams…

It could just be the Rams are a bad matchup for TB (like the Saints are). The Brady Bucs are 29-9 in the last two seasons, but are 1-5 against the Saints and Rams. (28-4 against all others)

34 Yeah it seems like teams…

Yeah it seems like teams with good interior pass rushes that excel in man coverage have always been his problem. I think the 49ers would be a bad match up for them as well for that reason. The NFC West teams have really adapted to it because they see it all the time. But Donald could completely take over that game if TB has center/guard issues again due to injuries.

72 Eh. Moreso for a guy like…

In reply to by reddwarf

Eh. Moreso for a guy like Brady who's like the ultimate structure QB. Sandlot-y type QBs like Wentz have less trouble. But "in trouble" Brady is still better than "less trouble" QBs anyway.

31 I'm surprised the Chiefs…

I'm surprised the Chiefs defensive DVOA came in at a whopping +8% in that game.  The starting defense only gave up a single score and, even including the final two drives with primarily backups in they held PIT to 3.9 yards per play (though is was significantly less than that heading into garbage time).  On the other side of things Pit gave up 42 points and 7.5 yards per play and were at +9%. 

Is the opponent adjustment for playing Pittsburgh's offense that severe?

55 It should be

Is the opponent adjustment for playing Pittsburgh's offense that severe?

It should be. The Chiefs' defense could only play the opponent they were given, but my goodness that was easymode.

Pittsburgh has one of the most comprehensively un-good playoff offenses I think I've ever seen personally...and yes, I remember the 2000 Ravens.

70 So many great defenses let…

So many great defenses let down by abysmal offenses in that era of Texans football. That Connor Cook vs. Brock Osweiler playoff game was one of the worst playoff games I have ever witnessed. Those Texans teams were like a dollar store version of the Buddy Ryan Eagles. Watt and White anchoring defenses that have to fight to make up for their awful offenses. Luckily for the Texans they were in a bad division.

33 My experience with Matt…

My experience with Matt Stafford is limited to watching games on t.v., of course, so for all I know he could be a jerk, but I've always thought he's played well and honorably while being chained to a crappy organization& coaching (4 yrs. of Caldwell and this year of McVay excepted). I'd really like to see the Rams keep winning playoff games.

47 I'm curious where the Cards…

I'm curious where the Cards performance lands. Maybe not as bad as NE but 2 epicly bad performances in the same playoffs is interesting. 

54 These lines are looking…

These lines are looking pretty good right now. I cleaned up in the WC round betting Buffalo against the spread and SF outright as well as the under. That more than made up for losing on an inexplicable bet on the KC vs. Steelers under for some reason.

Ten -3.5 vs. Cin: Gotta take Cincinnati with the hook there because of how even the match up is and the amount of back door cover ability the Bengals seem to just ooze with the Burrow/Chase connection. Question is how good is this Titans team with a full complement of starters for the first time all year? Either way this looks like a really good game with minimal opportunity for a blow out in either direction. That over on the O/U 46.5 looks pretty enticing too.

KC -2.5 vs. Buffalo: If the O/U hits 55 I have to drop a pile on the under. Saw it hit 54.5 at one point. Buffalo has a really good pass defense that can definitely bog a game down for KC. I envision Andy Reid channeling his SB winning plan against the 49ers to run a bunch out of shotgun into light boxes to soften up the Bills pass defense. The other thing to look at for the Chiefs is can Chris Jones take over and keep Allen in check. Both these teams are high variance with super human QBs that can just straight up disappear for long stretches of games. Feels like the perfect under game after last week. I would just disregard the spread.

GB -6 vs. SF: The first game was a 2 point last second win in SF for GB where Aaron was only pressured on 11% of his drop backs, GB held Trey Sermon and the 49ers 4th string running game to 67 yards on 21 carries, they won the TO differential 2-0, and abused the 49ers O-line. GB as a 6 point favorite in this game makes absolutely no sense. SF just finished playing the Rams and Cowboys on the road where they gave up 2 total sacks to two of the best pass rushing units in the NFL. If Bosa is back at full speed they could dominate the GB O-line. This 49ers defensive line has been incredibly destructive over the last half of the season and they have one of the best rushing defenses in the NFL. This seems like a worse match up than SF vs. Dallas in a ton of key areas. The public looks like they are going to bet this thing up a lot for GB even though it is basically going to require a Jimmy dumpster game and an Aaron super-human performance to get over that spread. Just back up a truck and drop it on the 49ers +6. There seem to be very few scenarios where this is a blow out in favor of GB with most likely outcomes being a final possession game around 3 in either direction. This game gives me serious 2011 GB-NYG vibes with this insane SF pass rush.

TB -3 vs. LAR: The Rams dominated the first match up against a much better TB roster at home and have swapped Desean for OBJ. Seems like something to stay away from unless TB went to -3.5 then grab the Rams.

Generally speaking the AFC match ups seem a lot better. The NFC ones look like opportunities to get on dogs against big time public teams. 

 

59 I actually agree that the…

I actually agree that the line on GB SF is nuts. No way should GB be favored by that much.

I would keep the unknowns around the SF pass rush and GB pass pro in mind though. We don't know about Bosa and GB hasn't played their normal offense with the line they will have at all this year. They had the line for the first 2 series of the DET game but they weren't running their normal offense they were running "always look to Adams first to get him the record". Actually it's possible that Turner will be back so that the line they start in playoffs is yet another combo (it would be the 8th with 10 changes because the week 5 combo was reused in week 14). That increases the variability in a few ways. In theory this line combo will be the best they have had all year. But it's also less tested. SF faced Nijman (LT3), Patrick (LG2), Meyers (C1), Newman (RG2), Turner (RT1) in week3. This should be Bakhtiari (LT1), Runyan (LG3), Meyers (C1), Patrick (RG1), Turner or Kelly (RT1/RT2). The line has never played a game with more than 2 preferred starters in their preferred position they might have 4 for this game only missing LG Jenkins. So even with Bosa back the SF pass rush and run defense might be less effective than they would have been against any other line combo.

But that is still just one area and I still worry that SF will do exactly what they have done to other teams with that front 7.

I also don't think GB will do anything special with their pass rush, even if Z Smith and Mercilus can play significant snaps. Sure they should help the pass rush rotation they are better than the guys who have been playing even if they aren't 100% but I don't think it matters against the SF line so while that might be swaying the public I think you are spot on with thinking it's false.

I don't think this is quite the same deal as 2011 either. The GB offense is not as one dimensional as it was then. The SF pass rush is very good but it's not quite the same level as that Tuck, Canty, Joseph, JPP line was. Also while a great 4 man rush is a great way to be Rodgers (and Brady for that matter). The LaFleur/Hackett offense is more adaptable than the McCarthy/Philbin. I also never expect GB to have 4 turnovers in a game. The GB special teams could easily give up 14 points of value though even given the issues SF has had with ST itself. GB doesn't need the other team to do anything to screw up on special teams.

That being said SF is still a very tough match-up and I see multiple paths for GB to lose this game and multiple to win. Like you said I think most of those, either way, are likely to lead to a one possession game. I tend to favor GB simply because of Rodgers vs Garoppolo and the extra rest time but it should not be a strong favorite, that line you quoted is definitely shifting because of public perception vs what it should be.

61 I agree with pretty much…

I agree with pretty much everything including the odds of 4 TOs for GB being extremely low. It reminds me of that game in style more than anything. I also can't see Jimmy putting up numbers like the ones Eli did in that game. For teams that already played each other earlier this year it is crazy just how different the line ups will be when they play. Adams could very easily put up 200 yards receiving and I could see Deebo going off for 100 yards rushing. It's crazy how much firepower there is in the NFC this year. I could easily see any of the 4 teams going to the SB and winning. The big wildcard in this SF-GB game has to be the nearly equally terrible ST units. I just hope the game isn't decided by muffed punts or some other ST quagmire. A clean game between the 2 should end up being really good and competitive.

65 Re: "Buffalo has a really…

Re: "Buffalo has a really good pass defense that can definitely bog a game down for KC"

Not everyone agrees with that:

https://fivethirtyeight.com/features/the-buffalo-bills-got-the-least-bang-for-their-schedules-buck-this-season/

71 That is a great take. I have…

That is a great take. I have really had trouble buying the greatness of the Bills pass defense and just could not figure out why. They have the best safety tandem in the NFL, which is a big contributor, but the pass rush looks overly manufactured with the best rusher being an old Mario Addison. It is a flawed pass defense kind of like the 2nd place Cowboys with their overrated secondary that relied heavily on a fantastic pass rush to cover up the gambling DBs mistakes. I might have to jump on KC -2 after reading that. The Bills and Cowboys are giving me some serious 2019 Patriots vibes where the defense got overrated by dominating terrible opposition without a dominant pass rush. I wonder how big of an impact back on the inside Chris Jones will have on this game. Should be really interesting.

73 I want to go Packers homer a…

I want to go Packers homer a little bit here and push back on the idea that the SF defense is a bad matchup for the GB offense. It might actually be just as true the other way around. The SF front four is excellent and they were obviously great against the run all year, but their pass defense as a unit has been pretty middling. And as good as their pass rushers are, they actually haven't registered a great pressure rate this season (20th in the NFL per PFR charting). Receivers will get open behind the SF front four and you can't get pressure if the ball is out.

Meanwhile, the Packers have one of the most efficient quick passing games in the league. Rodgers is very decisive in terms of identifying when he has the coverage/matchup he wants downfield vs. when to get the ball into the flat. He used to be the poster child for holding onto the ball and extending the play, but these last two seasons he's been getting the ball out very quickly, even on the plays where they attack down the field. It's going to be a different experience for the 49er defense than playing Dak Prescott, who likes to stand in and cycle through his progressions. If the 49ers jump out ahead of GB, Rodgers can definitely start to press and fall back into his old tendencies, but I think when the Packers passing game is at its best, they like to play exactly the way you would want to attack the 49er pass defense.

I am a lot more worried about the other side of the ball. Even setting PTSD from 2019 aside, the 49ers have really evolved what they're doing in the run game over the course of this season (will be a lot different from what we saw Week 3) and should have a big edge there. The Packers are also still bad at covering TEs. Hello, George Kittle. I guess the wild card is still Jimmy and how he's going to play with his bad shoulder and bad hand in the cold weather, and whether we'll see any of Lance.

74 Rodgers should be able to…

Rodgers should be able to feast on the weak 49er secondary, especially when it comes to deep passes.

I don't agree that Dak necessary likes to go through his progressions. He's at his best in the no huddle offense when he can confidently fire off the ball without thinking and when a presnap read determines where to immediately go with the ball. Against better defenses, he appears to hold the ball too long, either because he becomes locked on a receiver or goes through his progressions too fast because he doesn't trust himself or the receiver to be open. Romo pointed out numerous receivers downfield that Dak simply didn't throw to in the wildcard game, and there's a Kurt Warner analysis out on YouTube that I've heard points out the same.

76 The difference on the 49ers…

The difference on the 49ers getting talked about the least is just how much better their D-line is against the run since the first game. Ever since they moved Armstead inside next to DJ Jones the rushing defense has completely changed. The biggest problem for SF in the first game was how effective the GB running game was at getting positive yards running inside. Jones and Armstead inside are dramatically better than Kinlaw and Jones were in the first game. Dre Greenlaw is also back and much better at stopping the run than Harris or Al-Shaair. If GB can't run effectively against our nickel defense it is going to make passing quickly a lot harder since the safeties won't have to come closer to the line. Adams is going to feast no matter what in this game. If SF minimizes the other receivers that exploited them in the first game this game is going to be significantly harder offensively for GB. Lenoir and Norman are not in our secondary anymore either for Aaron to feast on. The 49ers with Jimmy in his last 3 starts have gone 2-1 on the road against the Titans, Rams, and Cowboys while he threw 2 TDs and 5 Ints. They lost the Titans game by 3 points even though SF gained 100 more yards. Unless GB decisively wins the TO battle, 2+, they have almost no chance of making this more than a one score game. If SF wins the TO battle, extremely big IF, I see very few ways GB can win at all. If Bosa is 100% this game is going to be very evenly matched and tough to win. The 49ers defense since week 8 compares closest to that Saints defense from week 1 due to the incredible rushing defense and interior pressure they bring. Because Jimmy is banged up and making too many mistakes I don't see any way the outcome would look anything like that blow out. But if the GB rush defense looks anything like it did when they beat Cleveland by 2 in week 16 they better hope Jimmy throws 4 picks. This is going to be a very difficult game for GB.