Writers of Pro Football Prospectus 2008

19 Sep 2014

Atlanta's Three-Quarter Ass Whooping

Well, this couldn't wait for Tuesday's DVOA commentary. Wow, did Atlanta kick Tampa Bay's ass tonight. It was 56-0 when third-string running back Antone Smith scored with two minutes left in the third quarter. At that point, the Falcons were on pace to break the all-time record of 73 points set by the Chicago Bears in 1940. From our FO perspective, we had to wonder: was this going to set a new record as the best DVOA game of all-time?

Well, it didn't. In fact, the Falcons didn't even end up with one of the top 40 all-time best games by DVOA. Tampa Bay ended up tied for the 20th-worst game in DVOA history... for now. Of course, these ratings will change once we have opponent adjustments for the full season. But for now, here's what we have:

ATL-TB VOA, Full Game
Team Offense Defense ST Total
ATL 33.2% -55.1% 20.4% 108.6%
TB -76.9% 33.1% -23.2% -133.3%

How did Atlanta not end up with one of the best DVOA ratings in history after scoring 56 points? Well, turnovers have something to do with it. Remember that the Falcons fumbled four times tonight, including a muffed punt by Devin Hester. The Falcons don't get any credit at all for the blown snaps by Evan Dietrich-Smith, although Tampa Bay does get penalized. And the Falcons also take a very tiny hit for Matt Bryant missing a 59-yard field goal at the end of the first half. (In the dome, the expected point value of this field-goal attempt is something like 0.67 points.)

However, the biggest reason Atlanta didn't end up with one of the best games in DVOA history is the fourth quarter, where Tampa Bay had a touchdown drive and then snagged a T.J. Yates pick-six. If we only look at the ratings through the end of the third quarter when the Falcons were up 56-0, the Falcons would in fact have the best game in DVOA history -- at least until we eventually apply opponent adjustments -- and the Bucs would have the worst game.

ATL-TB VOA, Through 3rd Quarter
Team Offense Defense ST Total
ATL 57.5% -80.1% 19.8% 157.5%
TB -91.0% 58.3% -22.9% -172.1%

This is where I'm sure there will be complaints about how our DVOA system handles "garbage time." Obviously, the Falcons didn't really care about the Bucs scoring in the fourth quarter, so you might wonder why the Bucs are getting positive value there. Well, the Bucs are still performing better than most teams would when losing by a large amount in the fourth quarter. Now, that's not based on games where a team is losing by 50, but that's because in the NFL there are almost no games where a team is losing by 50. The NFL has much, much less garbage time than college football. For the most part, every play counts and really does tell you something about those teams. This is one of the rare occasions where the game was so out of hand in the fourth quarter that it probably didn't mean anything. But we don't build the whole system around 56-0 games that come around once a decade. It's an issue where hard cases make bad law.

So, no single-game record for Atlanta or Tampa Bay. But Atlanta is sure going to rank a lot higher when the full ratings come out on Tuesday.

Here, for your enjoyment, are the top ten best and worst single-game DVOA ratings going back to 1989. (For those who are curious, Seattle had the 20th best game in DVOA history in last year's Super Bowl..)

Best Single-Game DVOA, 1989-2013
Rank Year Team DVOA Week vs. Score Opp DVOA
Rank
1 1991 WAS 149.6% 1 DET 45-0 17
2 1989 CLE1 149.2% 1 PIT 51-0 18
3 1999 PIT 145.4% 1 CLE 43-0 30
4 1993 SF 145.0% 20 NYG 44-3 7
5 2012 SEA 141.1% 14 ARI 58-0 27
6 2008 NE 139.0% 16 ARI 47-7 21
7 1994 PHI 138.2% 5 SF 40-8 3
8 1989 SF 135.2% 19 LARM 30-3 3
9 2002 ATL 135.1% 12 CAR 41-0 25
10 2005 CAR 131.6% 18 NYG 23-0 9
Worst Single-Game DVOA, 1989-2013
Rank Year Team DVOA Week vs. Score Opp DVOA
Rank
1 2002 ARI -167.9% 13 KC 49-0 4
2 2000 CLE -163.8% 14 JAC 48-0 15
3 2005 SF -160.7% 7 WAS 52-17 7
4 1989 PIT -158.5% 1 CLE1 51-0 2
5 2007 KC -153.6% 14 DEN 41-7 17
6 1999 CLE -153.2% 1 PIT 43-0 20
7 2009 TEN -147.8% 6 NE 59-0 4
8 2000 ARI -145.8% 8 DAL 48-7 23
9 2005 SF -145.6% 2 PHI 42-3 18
10 1991 TB -144.3% 9 GB 27-0 23

Posted by: Aaron Schatz on 19 Sep 2014

41 comments, Last at 25 Sep 2014, 4:15am by Theo

Comments

1
by theslothook :: Fri, 09/19/2014 - 2:35am

Before the season, I actually thought Tampa Bay would resemble something along the lines of what Houston has resembled so far. Namely, a team with ostensibly strong receivers, potential to run the ball, and huge question marks at qb. On the other side, the defenses had some real talent spattered across the roster where coaching stability would bring out the best in those units.

Houston has done well, TB not so much. But why? Just because Mccoy was injured? Schedule? I guess that makes sense, but after three weeks, I really can't explain what it was that made houston do what I expected and TB not to.

2
by Vincent Verhei :: Fri, 09/19/2014 - 2:51am

Well, they led the Rams going into the fourth, and came within a Hail mary of beating Carolina, so they weren't that bad then.

3
by theslothook :: Fri, 09/19/2014 - 4:00am

Both teams were playing 2nd and 3rd string qbs. If you can't beat the panthers with anderson and the rams with austin davis at home, who can you beat really?

Honestly - given their schedule, I fully expect them to end up with the number 1 overall pick.

6
by Laverneus Dinglefoot :: Fri, 09/19/2014 - 8:14am

I know the issue is debatable, but if it ended up snagging them a franchise QB, I think that would be worth one terrible season. Whether or not Lovie Smith can develop a young passer is another story.

4
by IntoTheVoid :: Fri, 09/19/2014 - 5:12am

I would love to know how these three quarters compare to the first three quarters of that game where Michael Vick mauled the Redskins a few years back.

5
by nat :: Fri, 09/19/2014 - 7:17am

This is where I'm sure there will be complaints about how our DVOA system handles "garbage time."
Since you ask...
The fundamental problem is that you don't have a way to separate "hard situations" from "situations that bad teams get into" when setting the baselines.
It could be done. But it would be difficult. You would calculate the average DVOA of all teams that get into each bucket, and use that to adjust the baselines that you calculate now.
The idea is to have the baseline represent an average team, not an average warped by the biased set of teams that get into each situation.
Other than that, publishing DVOA after three quarters is a good alternative.

7
by Peregrine :: Fri, 09/19/2014 - 8:51am

Falcons porn. Wheeeee!

8
by bubqr :: Fri, 09/19/2014 - 9:13am

What a shameful performance by TB. You can't be whupped like that without looking bad, but boy was that ugly.

9
by panthersnbraves :: Fri, 09/19/2014 - 9:53am

I stepped out of the room for a minute and my wife was yelling "Wow! Oh My!" kinds of things. I yelled back "English, please!" so she could describe what was going on - I missed the Hester return. At that point, she asked me if there was a Mercy Rule like in High School, and I told her no - but I couldn't watch any more after that. It was too sad.

10
by Aaron Schatz :: Fri, 09/19/2014 - 10:44am

Apologies for error, fixed above. I had switched the 1989 49ers games. The one in the top ten is actually the NFC Championship Game. The Super Bowl is further down, No. 18.

12
by The Hypno-Toad :: Fri, 09/19/2014 - 11:09am

1989 Super Bowl is clearly ranked too low because it still gives me nightmares. My own emotional scarring is way better than this. The Broncos ar teh gratest team in history in terms of getting utterly destroyed in Super Bowls.

11
by MilkmanDanimal :: Fri, 09/19/2014 - 10:56am

So, last night, I decided I couldn't watch anymore, went to play Destiny instead. My Xbox One, of course, has the NFL app installed, and it is kind enough to provide you a notification pop-up whenever anything involving your chosen favorite team happens. This means I kept hearing a little "doink", followed by an NFL logo in a small box stating, "Your Team Highlight--Atlanta TD".

THANKS XBOX TOTALLY NEEDED TO KNOW THAT.

I have no idea what the hell happened last night. Honestly, it was so comic it's not that bad. Just an unreal butt-kicking. I would love to see somebody dissect Tampa's pass coverage from these first few weeks and relate exactly how WRs are getting so incredibly open all the time.

13
by JoeyHarringtonsPiano :: Fri, 09/19/2014 - 11:36am

I still can't comprehend what the Tampa secondary was doing. Especially Dashon Goldson...there were several plays where it looked like he had never played football before.

And Evan Dietrich-Smith long-snapping the ball into his own taint will join the Buttfumble in the pantheon of comedic NFL gifs.

I feel a great amount of empathy for Bucs fans. I remember a similar experience at the dawn of the Millen era. For week 3 in 2001, the Charlie Batch Lions hosted the Kurt Warner Rams on Monday Night Football and got shellacked 35-0 (could have been worse, the Rams clearly eased up by the 3rd quarter). The fans in the Silverdome started chanting "Let's go Red Wings!" because they knew football season was clearly over. The darkness....the despair....the horror....

14
by BJR :: Fri, 09/19/2014 - 11:55am

That's a really annoying modern phenomenon. You want to avoid a sporting event because you fear the worst and would prefer simply checking the final scoreline to have your fears confirmed, than having to agonizingly live through it. But being unable to because your phone/laptop/whatever is persistently buzzing updates through. 21st century Western man has got it tough I tell ya!

15
by JoeyHarringtonsPiano :: Fri, 09/19/2014 - 12:14pm

#FirstWorldProblems

28
by MilkmanDanimal :: Fri, 09/19/2014 - 3:00pm

I see your point. To compensate, for the next 13 Sundays, I'm becoming Amish.

37
by Theo :: Sat, 09/20/2014 - 2:52pm

of course, there's no way possible to turn those things off or change the settings...

39
by CaffeineMan :: Sat, 09/20/2014 - 9:16pm

I have no TV service and have a friend recording the games for us to watch later. In order to avoid seeing the result of the recorded game while watching another game, he actually put blue painter's tape on the bottom of his TV screen to cover the ticker...

41
by Theo :: Thu, 09/25/2014 - 4:15am

I"ve always wondered who reads that info that scolls at the bottom.
Nobody, that's who.

NBC has cleaned up their broadcast, which is a good thing.

16
by tomdrees :: Fri, 09/19/2014 - 12:59pm

Yeah, let's continue to glorify the concept of an "ass whooping" in light of recent events. That nomenclature is not indicative of any pervasive unhealthy cultural assumptions among football and its fans or anything. Good call, guys. Call it a drubbing, I mean a beating, I mean... oh wow. Weird. "Atlanta's Three-Quarters of Glory?" Sorry to get all linguistics professor, but if there's an outlet that might actually respond to that, it's F.O.

17
by tuluse :: Fri, 09/19/2014 - 1:04pm

Given the embarrasement level, I'd go with "pantsing". Really though, you're never going to get rid of violent metaphors with sports. A large part of the enjoyment of football is that it's a controlled violence.

18
by tomdrees :: Fri, 09/19/2014 - 1:18pm

Hey, I get it. I'm just trolling up a little awareness here. Pointing some fingers at the mirror. You know. All that.

29
by dbostedo :: Fri, 09/19/2014 - 4:59pm

Are you suggesting that "a drubbing" or "a beating" is preferable to "an ass whooping"? I really can't tell for sure the way you have it written there.

Also, tying those terms beign used together with recent domestic violence events seems like a very weak and tenuous connection. I don't see what one has to do with the other. I don't believe that referring to games like this one with violent terms has anything to do with domestic violence, socially or psychologically, directly or indirectly. But of course, I could be wrong.

32
by RickD :: Fri, 09/19/2014 - 10:30pm

This.

36
by Bright Blue Shorts :: Sat, 09/20/2014 - 2:36am

Not so much an issue with domestic violence with wives or girlfriends ... but the Adrian Peterson news seem a little close on this one.

33
by RickD :: Fri, 09/19/2014 - 10:32pm

By all means, let's address the serious problems of domestic violence by attacking the usage of figurative language.

Because that's, you know, a serious approach to problem-solving.

35
by dbostedo :: Fri, 09/19/2014 - 11:31pm

And this.

38
by Theo :: Sat, 09/20/2014 - 4:21pm

don't be so butthurt

19
by Bowl Game Anomaly :: Fri, 09/19/2014 - 1:19pm

The reason DVOA is more predictive when including garbage time plays is obvious once you realize it. A team playing in garbage time is getting blown out (or blowing someone out) and whatever their true talent level is, they have been playing far below it (or far above it). A few plays at the end of the game may be meaningless from a competitive standpoint, but they do slightly mitigate the extent of the blowout which brings the final DVOA values a little closer to the teams' true talent levels, i.e. gives a more accurate impression of how the teams can be expected to perform in the future. QED, garbage time should be counted at 100% if you value DVOA as a predictive stat.

23
by nat :: Fri, 09/19/2014 - 2:16pm

I suspect you're wrong on this. It isn't that including garbage time "corrects" the DVOA from earlier in the game, as you seem to think. That would be adequately done by all the other non-garbage time in the rest of the season.

The problem with excluding garbage time is that we would be selectively excluding data based on how well the team played each day. That would bias the data, thus screwing up predictive power.

Garbage time DVOA data is still weak. It's just better than intentionally biased data.

24
by theslothook :: Fri, 09/19/2014 - 2:28pm

I don't think it's a bias issue. I think what's happening here is - what is the discount factor you use once you pass some threshold that defines garbage time. People complain that Fo's discount factor is too low - meaning garbage time dvoa should be worth far less than it is, but then FO arrived at that discount factor presumably based on predictive power.

I personally think it is a non issue. Most blowouts themselves are likely not representative of true team quality anyways - even if the game flowed conventionally with nothing fluky.

25
by theslothook :: Fri, 09/19/2014 - 2:28pm

I don't think it's a bias issue. I think what's happening here is - what is the discount factor you use once you pass some threshold that defines garbage time. People complain that Fo's discount factor is too low - meaning garbage time dvoa should be worth far less than it is, but then FO arrived at that discount factor presumably based on predictive power.

I personally think it is a non issue. Most blowouts themselves are likely not representative of true team quality anyways - even if the game flowed conventionally with nothing fluky.

20
by Bright Blue Shorts :: Fri, 09/19/2014 - 1:41pm

All those turnovers and back-to-back scores reminded me of the Cowboys-Bills Super Bowl. Wondering how that stacked up for DVOA?

21
by theslothook :: Fri, 09/19/2014 - 2:09pm

I am personally a bit surprised that the saints 62-7 thumping of the colts didn't manage to get in there, but the 23-0 crushing by the Panthers over the giants did.

It's why I can relate to Tb fans today. And it of course was the last game people saw so it was all they talked about the next day, not like just another blowout lost among the more interesting games.

22
by theslothook :: Fri, 09/19/2014 - 2:11pm

I actually want to pose a question to the FO folks - when you're favorite team experiences a complete annihilation, is it more frustrating that your team's offense was putrid or that your defense was terrible?

For me, its offense until the score gets into the high 40s.

26
by nat :: Fri, 09/19/2014 - 2:32pm

That would depend on how many of those points came off turnovers, short fields, and safeties, and how many were on kick or punt returns.

27
by MilkmanDanimal :: Fri, 09/19/2014 - 2:59pm

Being my favorite team is the subject, I'm saying the defense; the offense was clearly full of question marks, but the defense isn't supposed to suck like this. Granted, Gerald McCoy is the anchor and he was out, but Michael Johnson was supposed to, you know, have a pulse, right? The secondary is completely lost, there's no pass rush, and I feel like I need to sent Lavonte David a sympathy wreath for being stuck on this team.

Zero points is understandable; it happens sometimes. You have bad days, it's a good defense, whatever. There is never any way a team should be down 56-0.

30
by intel_chris :: Fri, 09/19/2014 - 9:20pm

As to garbage time, there is an argument for DVOA to attempt to measure and quantify that. DVOA is already a "quantum" system, in the sense that it doesn't count yards exactly but instead uses success points rather than yards--that is the yards are quantized into buckets. (Worth noting is that even conventional stats are quantized, you don't hear that the rusher gained 3.7651 years, but instead 3 1/2 yards or perhaps 3.7 yards). That means DVOA (or even conventional stats) still is (are) adjusting things. Moreover, most of the adjustments are downwards, long passes (and long runs) are all capped. Therefore, it is easy to extend this to capping the stats even more when the game is lopsided. The only difficulty is such capping (even the caps DVOA places now) make it hard to judge best performances. Every quantization loses some of the nuances of the actual event. Of course, the win-loss record or the scoreboard of any particular game is also a quantization and a much more crude one, so we still have a finer grain of analysis with DVOA.)

I'd like to note that this is actually intentional. DVOA was originally developed to argue that one of the early Patriots teams, was actually doing better than their record suggested and a metric was needed to prove that fact. The team in question used a variation on the West Coast offense which resulted in small but consistent gains on each play that slowly marched the games down the field in a field position style of game. They didn't necessarily score on any given drive, but slowly got better field position over the other team over a series of drives until they were able to score (all the while preventing the other team from doing so). Thus, the capping system does what it is supposed to and minimizes fluky plays that occasionally (but not "predictably") generate big changes in field position. In that sense DVOA is the antidote to "highlight reel" sports where only the big plays are reprised. (Another quantization, but one which emphasizes the unusual plays over the normal ones.)

If one accepts the premise of DVOA, then one could argue that capping the points that get allocated for successful plays when the game is "out of reach" would achieve a similar effect, and would better show the teams that are able to consistently beat opponents rather than occasionally blow out an opponent, but otherwise don't do as well.

You could follow the guidelines by the guts and stomps to get a rough estimate of the metric. Teams playing within one score of the other team get full credit for their successes. Teams playing between one and two scores away get diminished credit (say 90%) for their successes. Teams playing at two to three scores away get 80% credit. Beyond three scores, 70% credit. Beyond four scores, 60% credit, and so forth.

However, if one were to contemplate such a scheme, one would need to redo the guts and stomps analysis and see how predictive large wins and losses really are. Remembering, anecdotally unfortunately, a year where we had one (perhaps more) team(s) getting a lopsided victory one week only to have a lopsided loss on a subsequent week. It is possible that capping is appropriate. Of course, we already have variance, so perhaps that is enough.

31
by MJK :: Fri, 09/19/2014 - 9:44pm

I know Tampa Bay's problem. The Falcons ran out of bubblegum.

34
by RickD :: Fri, 09/19/2014 - 10:35pm

Maybe someday Tom Brady will be able to guide the Patriots to a performance that meets the high standard set by Matt Cassel in the 2008 game that made the Top Ten list.

40
by Plucky :: Tue, 09/23/2014 - 10:42pm

Suggestion:

IF (point differential) > (minutes remaining + 9) THEN GarbageTime = 1

Thoughts?