Do you prefer Yards or points allowed as the metric of a team’s unit ranking?

[Non-Issue]

Of the two, points. Because that is really all that matters. If you added yards per play I would lean that way.

I didn't make it past the 3rd grade. So DVOA scares me and I have a feeling it probably hates america.

[JAF-N72EX]

8 hours ago, Broncofan said:

I will heart every post in this thread that says DVOA.   Or at least until my 20 likes are done.   

Giving up meaningless yards or points when the game is out of reach (but making team use up clock while scoring) or when the O puts you in a short field is so context specific.   Why I prefer DVOA by a mile.  It’s not perfect but it tries to address context and game situation (6 yards given up on 3rd and 3 not the same as 3rd and 10).   

DVOA FTW.  

And just to add to this. The reason as to why no available methology is going to ever be perfect is because the metrics are based on analytics. Not scouting (which is also not perfect)! Scouting and Analytics are totally different. 

I think people really need to learn the difference between the two and learn how to properly decipher them a little better.

Also, just to get this off my chest. I see people all the time saying this team was ranked X DVOA that year and so they were better than Z team. While ignoring that Z team were more consistent(VAR), finished the season at a better rate (Weighted), the Def/Off schedule were tougher, and did it with worse starting field position (Drive stats).

 

  

[Broncofan]

3 hours ago, JustAnotherFan said:

And just to add to this. The reason as to why no available methology is going to ever be perfect is because the metrics are based on analytics. Not scouting (which is also not perfect)! Scouting and Analytics are totally different. 

I think people really need to learn the difference between the two and learn how to properly decipher them a little better.

Also, just to get this off my chest. I see people all the time saying this team was ranked X DVOA that year and so they were better than Z team. While ignoring that Z team were more consistent(VAR), finished the season at a better rate (Weighted), the Def/Off schedule were tougher, and did it with worse starting field position (Drive stats).

 

  

Excellent points.  In a perfect world the best teams combine scouting with analytics.   Gettleman’s ridicule of analytics is really worrisome for NYG inasmuch as CLE’s former reliance on metrics as the main driver shows the extremes that should get avoided (although if you are going to lean one way the market inefficiency to exploit is that teams not being metric-aware).  

The more nuanced discussion within metrics is to recognize the weaknesses of each methodology.  Most of us here know PFF rating is just a starting point and not the end all and be all.   It’s helpful to know where the flaws lie - my main beef with PFF whenever the number seems wrong for what the game film shows is the assumption all opponents faced are the same.  It’s one reason why Matt Paradis’ 2017 season looks deceiving - 8th best C sounds great but when facing the soft run D of OAK/LAC/KC for 6 games and 4 others against bottom 10 run D’s, 8th is much poorer than his previous 2015 heights where he was top 2-3 against great completion.   

On the other hand as a general indicator metrics are going to be more reliable in comparing individuals within a group - just not fooproof.   It’s a tool like scouting is.  And just as flawed as it is strong - but in different ways.  Why the best teams use both.   

 

 

[SBLIII]

DVOA

[FinSting]

7 hours ago, JustAnotherFan said:

And just to add to this. The reason as to why no available methology is going to ever be perfect is because the metrics are based on analytics. Not scouting (which is also not perfect)! Scouting and Analytics are totally different. 

 

 

  

How bout the Patriots last year? What a horribly weak defense, yet they made it to the Super Bowl anyway. 

[childofpudding]

Whichever has a higher correlation with DVOA over the last 5 years or so (probably points, but that's just a guess)

[JAF-N72EX]

6 hours ago, FinSting said:

How bout the Patriots last year? What a horribly weak defense, yet they made it to the Super Bowl anyway. 

I'm not quite sure where your getting at here. Are you saying the metrics are wrong because the Patriots went to the SB with a defense that data says were 31st worst in the league and you disagree with it? 

If that is the case, then this kind of goes back to what I was saying earlier about looking at the data properly in order to find out why. In this particular case - why was NE's defense ranked 31st but still made it to the SB? Metrics may answer this question as well if you really look it.

(Remember, when it comes to DVOA negatives ratings are positive on offense and negatives are positives fon defense because the formula starts at zero.)

NE's defense ranked 31st compared to the league average with a DVOA rating of +10.9% but finished with a Weighted DVOA of 3.9% rating --- a 7% increase. This tells us that the defense got better as the season progressed. The question now is...just how much better? And the answer to that is - significantly better. And the data reflects this as NE's 7% increase was the 2nd highest in the league last year behind only Oakland's 7.20%.

This is also further confirmed if you lok at the DVOA for each half of the season you will see that the defense rated poorly in the first 8 games (an avg of +20.9%) in which they posted a negative DVOA rating only one time (NYJ) but in the last 8 games (+0.41%), they did so in four of them. 

All of this data tells us that the defense was not as bad as they appeared to be heading into the playoffs as shows were progressively getting better. 

Because I like visual representation and I'm not sure if I'm explaining myself well enoguh, here is a graph that I hope will help in understanding what I'm saying.

 

 

[TheKillerNacho]

On 8/31/2018 at 9:37 AM, Danger said:

I mean between the two, I'll take points allowed, but neither really accurately measures the level of play of a defense. Too much subjectivity in those stats.

I don't care if you give up 500 yards a game if you somehow manage to only give up 10 points a game.

I don't think subjectivity means what you think it means; neither points allowed nor yards allowed are subjective in any way (although someone's conclusions using these stats may be subjective).... But I do agree with your general notion that these stats don't tell the whole story. And points is by far better, although also flawed.

[FinSting]

15 hours ago, JustAnotherFan said:

NE's defense ranked 31st compared to the league average with a DVOA rating of +10.9% but finished with a Weighted DVOA of 3.9% rating --- a 7% increase. This tells us that the defense got better as the season progressed. The question now is...just how much better? And the answer to that is - significantly better. And the data reflects this as NE's 7% increase was the 2nd highest in the league last year behind only Oakland's 7.20%.

 

Both NE and Oakland had crappy defenses to start, and they finished with crappy defenses. Being less crappy was still crappy. 

Stats are just stats as you say, they can be interpreted six ways from Sunday. A D can give up 79 yards on one drive, then recover a fumble or ****** an INT or the opposing team misses a short FG and suddenly those 79 yards earned zero points. Yet it makes the D look awful, giving up so many yards on one drive.