DVOA: #25

[GoBills808]

1 minute ago, TwistofFate said:

DVOA (Defense-adjusted Value Over Average) calculates a team's success based on the down-and-distance of each play during the season, then calculates how much more or less successful each team is compared to the league average. According to Football Outsiders, DVOA "breaks down every single play of the NFL season to see how much success offensive players achieved in each specific situation compared to the league average in that situation, adjusted for the strength of the opponent.

 

Not that complex to understand.  We've faced terrible teams all year, we haven't had any convincing or dominant wins over any of them. 

 

Its no wonder we are ranked where we are. 

 

Multiple ints, fumbles and incompletions against the weakest defenses in the league will help drop your DVOA score in a hurry. 

Yes, that is the copy and pasted definition. If you say it's not complex to understand you definitely don't understand what FO is trying to define. I submit that their attempts to calculate individual football plays on the basis of 'over (or under as the case may be) average' ie determining the value of a particular play's relative success relative itself to a similar 'average' NFL play is WAY too ambitious...the game simply has too many moving parts to comfortably assess every single play vs your average-adjusted median.

 

I like Football Outsiders. I think they have a unique way of looking at the game and I appreciate their statistical modelings. But the fact that it's derived entirely from NFL play-by-play stats means for me it will always be limited. They aren't watching the games, and they certainly aren't watching the individual plays. So there's no WAY I'm going to buy that the Titans, Bears, and Broncos are better teams than the Bills. The Jaguars aren't the 11th best team in the league. DVOA is just another stat to add context, I don't think it's anywhere close to definitive and doesn't deserve (no one metric does imo) to be treated as such.

[jrober38]

1 hour ago, TwistofFate said:

 

Who did we beat, how did we beat them and was it convincing? 

 

You all just keep looking at 5-2, 5-2, 5-2...its great...but you fail to take into consideration the quality of teams we played and how we performed in the win.  DVOA actually looks at those metrics.  Its not just looking at wins and losses. 

 

This is a middle of the pack team. 

 

 

 

Yeah, we barely beat the Jets, struggled against the Dolphins until Tre White took over the game, needed the Titans kicker to miss 3 field goals, and left it late against the Bengals. 

 

We're 5-2, but there's not a lot of real proof we're actually any good as a football team. We appear to have a really bad offense, and a defense that is a bit unknown because they've mostly played terrible competition. 

Edited November 1, 2019 by jrober38

[GoBills808]

^^so you guys are pretty much in agreement then, that the Bills are the 25th best team in the league? 

[jrober38]

6 minutes ago, GoBills808 said:

^^so you guys are pretty much in agreement then, that the Bills are the 25th best team in the league? 

 

No.  I don't think DVAO is the be all end all.

 

I do however think our schedule is the biggest reason we're 5-2. We've played teams with a collective record of 19 and 34, with a win percentage of .358 which is abysmally low.

 

I think our defense and special teams are pretty good, but our offense is terrible. In 5 of our 7 games we've played defenses ranked 22nd or lower and are still averaging less than 20 points per game in scoring. Ultimately I think our offense is our achilles heel. 

[GoBills808]

16 minutes ago, jrober38 said:

 

No.  I don't think DVAO is the be all end all.

 

I do however think our schedule is the biggest reason we're 5-2. We've played teams with a collective record of 19 and 34, with a win percentage of .358 which is abysmally low.

 

I think our defense and special teams are pretty good, but our offense is terrible. In 5 of our 7 games we've played defenses ranked 22nd or lower and are still averaging less than 20 points per game in scoring. Ultimately I think our offense is our achilles heel. 

*4 out of 7

 

also played the #1 and #4 fwiw

[jrober38]

17 minutes ago, GoBills808 said:

*4 out of 7

 

also played the #1 and #4 fwiw

 

Sorry - was a typo saying 22nd or worse. Should have been 21st or worse (Eagles, Bengals, Jets, Bengals, Dolphins). 

 

Aside from maybe the Giants, we didn't look great on offense against any of those teams. 

 

I think our offense, and particularly the passing game is one of the worst in the league. 

Edited November 1, 2019 by jrober38

[vincec]

I mean, what difference does it really make? They will be judged by their W-L record and their performance in the playoffs, not their DVOA.

[Thurman#1]

23 hours ago, KzooMike said:

So much debate about how good we really are. It's indicative of every hot start we have had since the start of the playoff drought. Expectations rise and crash just as fast.

 

DVOA is disgustingly accurate KPI brought to you by Football Outsiders. It is to the NFL what KenPom is to College Basketball.

 

 

DVOA is a really good number. But disgustingly accurate goes beyond what it actually is.

 

More, you have to look carefully at the specifics of it all.

 

Quick example, what's the Bills defensive DVOA when NOT adjusted for strength of opponent? The answer is ... 3rd in the league.

 

So, it's purely adjustment for opponent strength that has placed us so low. In other words, our schedule has not let us demonstrate (for DVOA anyway) strength. We've done very very well defensively this year - and unadjusted DVOA agrees with that. But we were only allowed to play the teams on our schedule. Basically the Bills have proven as much as they can against the weak slate they've faced.

 

Unadjusted offensive DVOA is 24th.

 

Special teams are also part of it, but when your offense is 24th and your defense is 3rd, that doesn't make you the 25th best team. It only means that though you're somewhere around 12th before they start adjusting for opponent.

 

We've proven a lot against the schedule we've faced. But the adjusted DVOA scores penalize you for a weak schedule. Fair enough, but you have to keep that in mind as you look at the rankings.

 

 

[Mikie2times]

24 minutes ago, Thurman#1 said:

 

 

DVOA is a really good number. But disgustingly accurate goes beyond what it actually is.

 

More, you have to look carefully at the specifics of it all.

 

Quick example, what's the Bills defensive DVOA when NOT adjusted for strength of opponent? The answer is ... 3rd in the league.

 

So, it's purely adjustment for opponent strength that has placed us so low. In other words, our schedule has not let us demonstrate (for DVOA anyway) strength. We've done very very well defensively this year - and unadjusted DVOA agrees with that. But we were only allowed to play the teams on our schedule. Basically the Bills have proven as much as they can against the weak slate they've faced.

 

Unadjusted offensive DVOA is 24th.

 

Special teams are also part of it, but when your offense is 24th and your defense is 3rd, that doesn't make you the 25th best team. It only means that though you're somewhere around 12th before they start adjusting for opponent.

 

We've proven a lot against the schedule we've faced. But the adjusted DVOA scores penalize you for a weak schedule. Fair enough, but you have to keep that in mind as you look at the rankings.

 

 

Fair, I could argue strength of schedule, within the world of DVOA is very accurate and more of a 20,000 foot view. But I can also appreciate the breakout. No question we have performed well in DVOA against who we have played. Dismissal of DVOA in either form makes me want hit my head against the wall. Pythagorean theorem is a very simple statistical tool to project future success. That is much more accurate then simply looking at W/L. Looking for unemotional data points that predict future success is a very logical pursuit. 

 

My focus once we over perform or under perform is trying to understand why and if that element is sustainable. As an example in our playoff season we ranked 21st in DVOA, so I looked at why? Turnover margin. I considered if it was sustainable. With somebody like Tyrod and our defense that year I felt it was. Generally speaking very little correlation exists in turnover rates over a full season so you see regression as the year progresses in teams with large + turnover differential early = regression to mean. I didn't see it in that team and I knew if that played out it had more weight then what DVOA could handle. I also see elements in this team I really tried to focus on in the original post as reasons why I think we will outperform our DVOA as we did in 2017.      

Edited November 1, 2019 by KzooMike

[Nextmanup]

2 hours ago, vincec said:

I mean, what difference does it really make? They will be judged by their W-L record and their performance in the playoffs, not their DVOA.

Of course, but DVOA is a metric used to try and predict future performance.

 

If your attitude is to wait and watch the games, see what happens, and have no interest in predicting future performance, you don't really need to focus on DVOA.

 

 

[SJDK]

17 hours ago, Seven-N-Nine said:

 

Yeah that's it, bury your head in the sand, typical...

I’m with him, 5-2 is not a bad place to be. I hope you’ll keep posting when we are in the playoffs 

Edited November 1, 2019 by SJDK

[GoBills808]

27 minutes ago, KzooMike said:

Fair, I could argue strength of schedule, within the world of DVOA is very accurate and more of a 20,000 foot view. But I can also appreciate the breakout. No question we have performed well in DVOA against who we have played. Dismissal of DVOA in either form makes me want hit my head against the wall. Pythagorean theorem is a very simple statistical tool to project future success. That is much more accurate then simply looking at W/L. Looking for unemotional data points that predict future success is a very logical pursuit. 

 

My focus once we over perform or under perform is trying to understand why and if that element is sustainable. As an example in our playoff season we ranked 21st in DVOA, so I looked at why? Turnover margin. I considered if it was sustainable. With somebody like Tyrod and our defense that year I felt it was. Generally speaking very little correlation exists in turnover rates over a full season so you see regression as the year progresses in teams with large + turnover differential early = regression to mean. I didn't see it in that team and I knew if that played out it had more weight then what DVOA could handle. I also see elements in this team I really tried to focus on in the original post as reasons why I think we will outperform our DVOA as we did in 2017.      

Do you see the justification for the discrepancy between Buffalo and New England's defensive DVOA when we've played the tougher schedule?

[Seven-N-Nine]

6 hours ago, SJDK said:

I’m with him, 5-2 is not a bad place to be. I hope you’ll keep posting when we are in the playoffs 

Some of you really seem to be on a gif-kick...

[eanyills]

I mean, the best record this team will finish with is 9-7. DVOA is a really good predictor, like kenpom and Corsi. Corsi had the Sabres pegged as awful from day one last year and nobody wanted to believe it.

 

Data analysis can be wrong but DVOA has proven to be fairly accurate in my reflection. The eye test also confirms what they’re saying a bit. Heaven forbid I say that multiple aspects of the Bills look bad though..... I must not be a “real fan,” whatever that means.

[SoTier]

23 hours ago, cwater10 said:

Are you effing serious right now?  Did you just equate data collection and application methods used by astronomers and physicists to the recreational junk science known as Football Outsiders?  Queue perspective any time you feel inspired for your next revelation.  Data does not show anything.  Data DOES inform our interpretation!  You have yours.  I find it depressing and self flagellating as a fan.

 

Statistical analysis is statistical analysis whether it's applied to astronomy or to football.   The specific algorithms and variables are different but the fundamental principles are the same.

[SoTier]

22 hours ago, HappyDays said:

 

I'll be surprised if we get blown out. Since McDermott got here we've always played New England close. The NFL is a matchup league. DVOA does a very good job, probably as good as any statistic can, at measuring the quality of NFL teams. But it doesn't properly account for matchups. I had the Eagles game pegged as a loss because their weakness - pass defense - didn't match up with our team's strength. We couldnt take advantage of their weakness. The Patriots are the opposite. Their strength on defense is pass defense. That doesn't hurt our team as bad as it does other teams. Their strength on offense is passing the ball. Our defense matches up well with that. I expect we will lose that game by a TD or less.

 

The bolded statement is simply untrue.   Belichick has beaten the Bills by 20, 21, and 19 points in 5 games with McDermott as HC.  If not for a garbage time TD in the 2nd 2018 loss,  that would have been an 18 point drubbing instead of just a 12 point loss.

 -  In 2017, the Bills lost to NE 3-23 and 16-37.   

 -  In 2018, the Bills lost to NE 6-25 and 12-24.

 -  Their 2019 10-16 loss this season is the only close game.

 

 

[dave mcbride]

15 hours ago, Joe in Winslow said:

 

They may lose. But I FULLY expect them to beat the likes of Miami, NYJ, Cleveland and washington at a MINIMUM, and likely shittsburgh too.

Also, those same Jets beat the Cowboys, so...there's that.

 

 

They should lose to Dallas, but there is always the any-given-Sunday factor: turnovers, fluke plays, injuries. You never know.

[BananaB]

The fact that we are winning games against teams we should beat is a step forward for this organization. Anyone who has watched us play for the past 2 decades should know this. I don’t care that we ain’t blowing teams out, you learn more from those games then you do blowouts. Players learn more, coaches learn more, management learns more. We still went toe to toe with the best team in the league and if Allen had not got knocked out who knows how that game would of ended. We are almost half way into the season, we’re 5-2 and we let one game get away from us. This is an improvement, the strength of schedule does not matter. 
 

Over the years I seen plenty of teams have weak schedules get a lot of Ws and score a lot of points, then when the games toughen up late in the season they look lost.  The way we are going atleast we know we have ***** to improve on, that way we may have it figured out for when the game really matters. 

Edited November 1, 2019 by BananaB

[Big Turk]

On 10/31/2019 at 5:55 AM, JoPar_v2 said:

I mean, why is DVOA “disgustingly accurate?” Apparently it’s not if it predicts the bills at 3-4 when, in real life, they’re 5-2. I am not anti-analyics. Far from it. But these numbers need to, at some point, be predictive and they dont seem to be. Call it an outlier maybe, but there’s only so many outliers until the model starts breaking down

 

Well, its more like what they "should" be based on the statistics...

Honestly its not hard to see that considering they easily could have lost to the Jets, Bengals and Titans and possibly should have lost to at least one of them.

[LSHMEAB]

On 10/31/2019 at 5:42 AM, dave mcbride said:

 

Who is “most”? These numbers have been raised in other threads to relatively little criticism. I also think you’re looking at it incorrectly. Like fielding independent pitching (FIP), which isn’t an accurate indicator of past ERA but a great one for future ERA, DVOA offers you an idea of where the team is headed next. The issue is that the Bills schedule is SO easy — like, historically easy, as in the next 4 games are against bottom tier DVOA teams — that the win loss record won’t show you how mediocre the team really by week 11. If they get blown out by Dallas, Baltimore, and NE while getting to 11-5/10-6,  you’ll have a sense of who they really are.

Excellent post, but BABIP would have been a better baseball reference.

 

I don't find this surprising. We really haven't beaten a good team and have looked suspect in some of the wins.

 

Stats can't take away the 5 wins the Bills have accumulated, but if we're being honest, I think everyone knows this is not a VERY good team. Not sure why that's controversial.

 

As a fan, I just want to see them win the games they need to win.