Advanced Metrics

[Dantés]

The Athletic's Sheil Kapadia recently put out an article that gave some interesting metrics. I'm not going to share the full results, but here are the Packers totals:


Explosive Rate: These are 20+ yd plays adjusted for total numbers of drop backs.

• Aaron Rodgers, 7.81% (25th)

Deep Throwing Rate: These are pass attempts that travel 20+ yards past the LOS, adjusted for total drop backs, with pass attempts inside the opposing RZ removed.

• Aaron Rodges: 17.19% (8th)

Pressure Rate: Sacks and hits adjusted for drop backs.

• Aaron Rodgers: 31.94% (6th)

Pressure Rate (Def): Sacks and hits adjusted for opposing drop backs.

• Packers: 24.14% (10th)

Pressure Rate (Def/Ind): Sacks and hits adjusted for opposing drop backs, by player.

• Preston Smith: 10% (21st- Tie)
• Za'Darius Smith: 6.94% (48th)

Some of this matches observation, and some is a little surprising. We all know the offense has sputtered so far and that Rodgers is taking too many sacks/hits. The Packers have also played two really good pass rushing teams so far.

But to my eye test, the Smith's have looked a lot better than 21st and 48th.

But that's kind of the point. I bring this up to illustrate that advanced metrics, depending on how they're sliced, can contradict other metrics as well as the "eye test."

My understanding is that PFF has the Packers at #1 or #2 in pressure? And I know they feel very good about our protection. I'm not saying they're right and Kapadia's wrong, or even vice versa.

What I am saying is that metrics like this, regardless of source, are informative but not conclusive. I see PFF metrics cited all the time as though they settle a question. That's just misleading. Often, those guys prove they aren't very good at analyzing the game, and furthermore, how set the perameters of a stat will have a big influence on the results.

For example, PFF can decide for itself what it considers a "pressure." By their definition, Z. Smith is the best in the league. But when Kapadia restricts the perameters to the more concrete "sacks and hits," Smith drops to 48.

Who is right? Both. And neither. Our eyes should be telling us that Z is balling out. The metrics should just help inform what we are seeing.

 

[HardRightEdge]

The Athletic's Sheil Kapadia recently put out an article that gave some interesting metrics. I'm not going to share the full results, but here are the Packers totals:


Explosive Rate: These are 20+ yd plays adjusted for total numbers of drop backs.

• Aaron Rodgers, 7.81% (25th)

Deep Throwing Rate: These are pass attempts that travel 20+ yards past the LOS, adjusted for total drop backs, with pass attempts inside the opposing RZ removed.

• Aaron Rodges: 17.19% (8th)

Pressure Rate: Sacks and hits adjusted for drop backs.

• Aaron Rodgers: 31.94% (6th)

Pressure Rate (Def): Sacks and hits adjusted for opposing drop backs.

• Packers: 24.14% (10th)

Pressure Rate (Def/Ind): Sacks and hits adjusted for opposing drop backs, by player.

• Preston Smith: 10% (21st- Tie)
• Za'Darius Smith: 6.94% (48th)

Some of this matches observation, and some is a little surprising. We all know the offense has sputtered so far and that Rodgers is taking too many sacks/hits. The Packers have also played two really good pass rushing teams so far.

But to my eye test, the Smith's have looked a lot better than 21st and 48th.

But that's kind of the point. I bring this up to illustrate that advanced metrics, depending on how they're sliced, can contradict other metrics as well as the "eye test."

My understanding is that PFF has the Packers at #1 or #2 in pressure? And I know they feel very good about our protection. I'm not saying they're right and Kapadia's wrong, or even vice versa.

What I am saying is that metrics like this, regardless of source, are informative but not conclusive. I see PFF metrics cited all the time as though they settle a question. That's just misleading. Often, those guys prove they aren't very good at analyzing the game, and furthermore, how set the perameters of a stat will have a big influence on the results.

For example, PFF can decide for itself what it considers a "pressure." By their definition, Z. Smith is the best in the league. But when Kapadia restricts the perameters to the more concrete "sacks and hits," Smith drops to 48.

Who is right? Both. And neither. Our eyes should be telling us that Z is balling out. The metrics should just help inform what we are seeing.

The first thing to none about all of Kapadia's data you cite is that it can be objectively determined if one assumes Elias's crediting of sacks and hits is objective. Whether a pass attempt actually travelled 19 or 20 yards might require a subjective determination, but those would be uncommon. Sacks and hits are official and semi-official numbers at any rate.

PFF includes "hurries" in their pressure stats, which are subjective determinations, whereas Kapadia evidently does not. It would not be unusual for a pass rusher to be credited with more hurries than sacks+hits. To illustrate, see last last season's PFF pressure leaders (sacks and hits taken from pro-football-reference.com):

Donald: 106 = 20.5 sacks + 41 hits, with an implied hurry count of 44.5
Cox: 95 = 10.5 sacks + 34 hits, with an implied hurry count of 50.4
Ford: 78 = 13 sacks + 29 hits, with an implied hurry count of 36

It's not hard to imagine a guy racking up a bunch of hurries per snap without a like number of sacks and hits over a small sample size of two games as is evidently the case with Z.

Why hurries? I can't find the PFF claim at the moment but according to their data the league-wide completion rates and QB ratings drop meaningfully when hurried, which stands to reason. Hurries have value.

A few years back PFF produced tables where they scored players on their pressures (as the distinct from their every-play grades), allocating 1.0 point for a sack, 0.75 points for a hit and 0.75 points for a hurry. I found this to be odd in that hurries seemed to be grossly overvalued. I haven't seen this table produced for public consumption in quite a while, but there is circumstantial evidence that they do value hurries fairly high in their grades.

As the #2 pressure guy in the league through Week 2 according to PFF, with a relatively high hurry count with comparative low sacks+hits, it's worth noting their overall grade for Z. is a high 85.0, a Pro Bowl level number:

https://www.pff.com/nfl/players/zadarius-smith/9555

While we don't get a pass/run breakdown in that link, it is fair to assume the pass rush grade is very high.

So, if the eye test tells you Z. is playing darn good football despite pedestrian sack+hit count (a total of 5), then you would find yourself in agreement with PFF's subjective grading elements that Kapadia does not attempt.

 

[HardRightEdge]

The Athletic's Sheil Kapadia recently put out an article that gave some interesting metrics. I'm not going to share the full results, but here are the Packers totals:
Pressure Rate: Sacks and hits adjusted for drop backs.

• Aaron Rodgers: 31.94% (6th)

On the offensive side of the ball, we have some relative agreement between Kapadia and PFF.

I don't know about anybody else, but I'm pretty disappointed with the O-Line play to date. Taylor and Graham have been the favored whipping boys, but whether you like it or not, the other guys have not done that great a job either. I called out Bakhtiari for a sub-par game in Week 1, including 2 holding calls, which in some minds is heresy; we now know he's playing with a bad back.

Here are the overall PFF grades for the O-Linemen through Week 2, and they meet my eye test:

Bakhtiari: 58.0

In a separate write-up they noted Minnesota got a lot of pressure (meaning "hurries" in this case) bull rushing him, while his run blocking was sub-par. Playing does not equate to playing injured.

Taylor: (114 snaps): 58.6
Jenkins (18 snaps): 75.7
Linsley: 57.3
Turner: 64.8
Bulaga: 76.4

With modified run and pass schemes this year, we can hope the kinks are still being worked out.

 

[HardRightEdge]

Here's a PFF curiosity, with grades through Week 2:

They give King an overall grade of 73.2 on 3-7 passing, 1 pick and 5 solo tackles.

https://www.pff.com/nfl/players/kevin-king/11788

They give Alexander an overall grade of 60.4 on 6-16 passing, no picks, 6 solo tackles:

https://www.pff.com/nfl/players/jaire-alexander/49350

Now, teams have so far picked on King less and he did get that INT. This data does not reflect what a player is doing on plays where the ball is not thrown his way which we often do not see without an all-22 view. PFF grades those plays. Perhaps King is giving up separation less often?

I would be curious to know how PFF graded the MIN TD where the ball went through Alexander's hands. King blew that coverage, beaten like a rug. By all rights Alexander had no business being at the ball in the first place. Getting to it in itself was a good play even if he didn't catch the ball. Would PFF dock Alexander on that play, whereas another corner just tending to his guy would be unscathed?

Personally, I don't see any reason to give King a better overall grade to date. Teams do like to throw at Alexander, going back to last season, probably because of his size even though he does not give up much separation. He had two good opportunities for picks against MIN and missed on both. If he starts making those plays then opponents might change their approach.

 

[weeds]

I'm just a simple guy ... I thought "Hey...a thread on socket sets and ratchets.

Not to be.

 

[Dantés]

Here's a PFF curiosity, with grades through Week 2:

They give King an overall grade of 73.2 on 3-7 passing, 1 pick and 5 solo tackles.

https://www.pff.com/nfl/players/kevin-king/11788

They give Alexander an overall grade of 60.4 on 6-16 passing, no picks, 6 solo tackles:

https://www.pff.com/nfl/players/jaire-alexander/49350

Now, teams have so par picked on King less and he did get that INT. This data does not reflect what the players are doing on plays where the ball is not thrown his way which we often do not see without an all-22 view. PFF grades those plays. Perhaps King is giving up separation less often?

I would be curious to know how PFF graded the MIN TD where the ball went through Alexander's hands. King blew that coverage, beaten like a rug. By all rights Alexander had no business being at the ball in the first place. Getting to it in itself was a good play even if he didn't catch the ball. Would PFF dock Alexander on that play, whereas another corner just tending to his guy would be unscathed?

Personally, I don't see any reason to give King a better overall grade to date. Teams do like to throw at Alexander, going back to last season, probably because of his size even though he does not give up much separation. He had two good opportunities for picks against MIN and missed on both. If he starts making those plays then opponents might change their approach.

As far as I know, Alexander was graded negatively for that play. Which demonstrates a laughable degree of ignorance by whoever they had grading that game.

I've never been a big PFF fan, but their analysis has gotten worse as they've gotten bigger.

 

[HardRightEdge]

As far as I know, Alexander was graded negatively for that play. Which demonstrates a laughable degree of ignorance by whoever they had grading that game.

I've never been a big PFF fan, but their analysis has gotten worse as they've gotten bigger.

Something just dawned on me that sounds a bit fishy:

1) PFF says they grade every player on every play on a -2 to +2 scale.
2) They have two guys producing preliminary grades.
3) Then a 3rd. senior guy resolves any discrepancies.
4) They take into account stuff like, "A four-yard run that gains a first down after two broken tackles will receive a better grade than a four-yard run on 3rd & 5, where the ball carrier does nothing more than expected. A quarterback who makes a good pass that a receiver tips into the arms of a defendper will not negatively affect the quarterback's grade on that play, despite the overall negative result for the team."
5) They say they produce the grade for paid subscribers within 24 hours of the conclusion of the game.

Let's say a particular game has 65 offensive plays by each team. So, 130 plays x 22 players = 2,850 player grades to produce. Lets say those first two guys spend 12 hours reviewing a game, during the game taking notes (for their first take "Refocus" notes) and then tape review after the fact. 12 hours x 60 minutes = 720 minutes.

That breaks down to 15 seconds per player-play grade. At that pace, these guys are going to be bleary eyed before they hit the two hour mark of tape review making snap judgements at breakneck speed. This does not account for the numerous plays where a considered judgement would require multiple rewinds. This does not appear to be reliable process.