Let's talk about PFF grades and the Miami Dolphins

[DOLFANMIKE]

As most of you know, I have a problem with PFF grades. They are fun to look at and it's always reassuring when we see them if they verify what we thought we saw in the game watching live. PFF is supposed to offer fair and consistent grades for players that we the fans can understand.
Because I always rewatch our games and do some grading myself most the time, I've noticed that the PFF grades frequently are hard to understand, and I've even see our coaches comment on player grades that seem to match up closely with where I had them, but PFF was way off with a much lower grade most the time.

So here is a short, prime example of my issue with PFF, and counting on their grading to evaluate our players.

Interesting case and point regarding PFF grading, which recently has improved for Eich.

Todays PFF OL Run blocking grades vs Washington:
RT Austin Jackson – 91.1

RG Robert Hunt – 90.4

OL Liam Eichenberg – 81.0

C Connor Williams – 67.1

OT Kion Smith – 62.8

LT Terron Armstead – 58.6

OL Lester Cotton – 42.5

Todays PFF OL Pass Pro Grades vs Washington
LT Terron Armstead – 85.4

OT Kion Smith – 83.6

OL Lester Cotton – 77.7

C Connor Williams – 68.3

RG Robert Hunt – 67.4

RT Austin Jackson – 65.6

OL Liam Eichenberg – 64.2

So Eich gets an 81% run blocking grade from PFF, and a 64% Pass Block grade from PFF for the Washington game. Keep in mind over the past 3 games, we have played the Raiders, Jets, and Washington and our offense played ok scoring 20, 34, and 45 points. Offensive yardage has been solid.
So then how is this information below possible? This is also a PFF grade / ranking. Want it to get worse - Look at Eich's PFF grades from the last 3 games, then look again at this tweet. How does Eich get a 64% pass pro grade and yet only allow 2 pressures, no sacks, no QB hits in over 200 reps?

 

[djphinfan]

I’ll try to explain coach as I went through their whole process of evaluating players and working for them

First they do not take into account the caliber of the opponent when making the grades

It’s total isolation on the players grades for each play and then they assess the grade.

So if Connor Williams stalemates Chris jone one on one he doesn’t get extra credit for it being Chris jones, just the normal grade for a stalemate

So in this case with Liam having lower grades than one would think when not giving up any pressures or sacks doesn’t mean he won on all those plays, he could of lost a rep in pass protection but the defender didn’t get to the Qb in time for the hit or pressure or sack, so, it goes down as a negative grade on the overall grade because he still lost his individual battle

If Tua make a poor location throw but the catch ends in a 80 yard td, the grade ends for Tua at the location of the throw, nothing after, and it wouldn’t be a positive grade because the location wasn’t good

 

[E30M3]

Eich is hard to judge by PFF standards. He is having to be cross referenced across the Center position and Guard. In the past 3 games he has only played guard, all prior were center except week 9 which was only 14 snaps. @brumdog44 has mentioned this in his weekly PFF ratings threads.

 

[PhinFan1968]

I think their grades on OL, specifically, are unreliable since there's no possible way they can take into account assignments, schemes, the impact some other player's blown assignment has on another lineman, etc.

I don't positively know that to be the case, just the feel I get when seeing their grades sometimes.

 

[DOLFANMIKE]

I think their grades on OL, specifically, are unreliable since there's no possible way they can take into account assignments, schemes, the impact some other player's blown assignment has on another lineman, etc.

I don't positively know that to be the case, just the feel I get when seeing their grades sometimes.

I agree

 

[DOLFANMIKE]

So in this case with Liam having lower grades than one would think when not giving up any pressures or sacks doesn’t mean he won on all those plays, he could of lost a rep in pass protection but the defender didn’t get to the Qb in time for the hit or pressure or sack, so, it goes down as a negative grade on the overall grade because he still lost his individual battle

I'll start by saying having graded more film than the average bear, that I'm not sure how anyone at PFF knows who Eich or anyone else is supposed to block a good amount of snaps. That said at least they can try to guess and then evaluate how the graded player did on their assumption. I'm ok with that, but again there's some weakness in that in terms of evaluation. If they even understand all the potential blocking schemes we use how do they know which we are using on that play, Etc. Do we have special adjustments for a player or scheme? It gets murky at best.

I understand that WHO they are blocking isn't a factor in their grading and it's easy to understand why that can't really be involved in the grading. I'm ok with that. That is something that coaches can better decide on how things went and why they may have not been able to accomplish an unrealistic blocking task on a stud like Chris Jones or a guy like Wilkins.

However, I'm not sure a Pro OL can do NOTHING to block a NFL defender and that player not have some type of an impact on a play. The DL for example is aligned within 3-6 yards of the QB depending on whether or not they are under Center or in the gun. How can an OL not do enough to win and still have the defender not impact the play in a way that is gradeable? Maybe once in while. Maybe a few times per game. But look again at the PFF grading from Washington on Eich's pass pro at 64% and then they claim that in 201 reps he had ZERO sacks allowed, ZERO QB hits (contact after throw), and only 2 QB pressures in 3 games. Yet Eich is at 64% Pass Pro for the Washington game.

I'm calling Bullshit.

It gets worse if you look over the Raiders and Jets games grades and do the same as what I just did with Washington. Those game grades, just like the Washington grades, do not reflect a player giving up ZERO Sacks, ZERO QB Hits, and only 2 QB Pressures in 3 games. Think about that!

Now do it over the season.

Now do it over Eich's career.

It's all the same problem I have been pointing out for quite awhile when folks try to use PFF as their "proof" that a guy graded out poorly or not. Now imagine a person wanting to argue about player performance without watching any film. Counting entirely on watching a game live and looking at PFF grades to base their point of view on player performance.

Last thing, I've graded my own film for over 33 years, and attended Coaching clinics with some of the best coaches teaching on how to do things.
I've been paid by other teams, groups, coaches to grade film on the side, and evaluate things related to film to discern keys or habits.

I also have a problem with specific plays that PFF grades and determines success or failure on. When McDaniel and I both have Eich graded within 3-4% points of each other around 70-75% and then PFF comes out with their evaluation around 50%-60% or worse that's a problem. This is what I've been saying for years. I see a massive inconsistency in evaluation going on here.

 

[djphinfan]

I'll start by saying having graded more film than the average bear, that I'm not sure how anyone at PFF knows who Eich or anyone else is supposed to block a good amount of snaps. That said at least they can try to guess and then evaluate how the graded player did on their assumption. I'm ok with that, but again there's some weakness in that in terms of evaluation. If they even understand all the potential blocking schemes we use how do they know which we are using on that play, Etc. Do we have special adjustments for a player or scheme? It gets murky at best.

I understand that WHO they are blocking isn't a factor in their grading and it's easy to understand why that can't really be involved in the grading. I'm ok with that. That is something that coaches can better decide on how things went and why they may have not been able to accomplish an unrealistic blocking task on a stud like Chris Jones or a guy like Wilkins.

However, I'm not sure a Pro OL can do NOTHING to block a NFL defender and that player not have some type of an impact on a play. The DL for example is aligned within 3-6 yards of the QB depending on whether or not they are under Center or in the gun. How can an OL not do enough to win and still have the defender not impact the play in a way that is gradeable? Maybe once in while. Maybe a few times per game. But look again at the PFF grading from Washington on Eich's pass pro at 64% and then they claim that in 201 reps he had ZERO sacks allowed, ZERO QB hits (contact after throw), and only 2 QB pressures in 3 games. Yet Eich is at 64% Pass Pro for the Washington game.

I'm calling Bullshit.

It gets worse if you look over the Raiders and Jets games grades and do the same as what I just did with Washington. Those game grades, just like the Washington grades, do not reflect a player giving up ZERO Sacks, ZERO QB Hits, and only 2 QB Pressures in 3 games. Think about that!

Now do it over the season.

Now do it over Eich's career.

It's all the same problem I have been pointing out for quite awhile when folks try to use PFF as their "proof" that a guy graded out poorly or not. Now imagine a person wanting to argue about player performance without watching any film. Counting entirely on watching a game live and looking at PFF grades to base their point of view on player performance.

Last thing, I've graded my own film for over 33 years, and attended Coaching clinics with some of the best coaches teaching on how to do things.
I've been paid by other teams, groups, coaches to grade film on the side, and evaluate things related to film to discern keys or habits.

I also have a problem with specific plays that PFF grades and determines success or failure on. When McDaniel and I both have Eich graded within 3-4% points of each other around 70-75% and then PFF comes out with their evaluation around 50%-60% or worse that's a problem. This is what I've been saying for years. I see a massive inconsistency in evaluation going on here.

lot to unfold here..

Obviously one of our lineman can outright lose their individual matchup and not have that defender touch Tua.

Like the receiver who houses a badly thrown location pass, the grade of the Oline stops at the loss, nothing after matters

It’s like when I evaluate players I totally watch in isolation, I don’t care what happens after the fact.

It is this process that keeps their process simple.

Of course there is some variation to the grades when the evaluator may not recognize the total responsibility schematically of each player on each play but imo most of the time you can get it right

 

[DOLFANMIKE]

Obviously one of our lineman can outright lose their individual matchup and not have that defender touch Tua.

I'm not sure that can happen to the tune of 64% success rate and still be the case you are trying to make with 201 reps and only 2 pressures. Sorry, I'm not buying it.

 

[djphinfan]

I'm not sure that can happen to the tune of 64% success rate and still be the case you are trying to make with 201 reps and only 2 pressures. Sorry, I'm not buying it.

With Tuas speed to throw I can totally see our Oline losing their individual battles off the snap and the defender not affecting Tua.

 

[E30M3]

With Tuas speed to throw I can totally see our Oline losing their individual battles off the snap and the defender not affecting Tua.

It happens quite often if you focus on them. The ball is just already starting to fly so the camera pans out.

 

[djphinfan]

It happens quite often if you focus on them. The ball is just already starting to fly so the camera pans out.

agreed..

No platform is perfect but at least they do the work on an individual basis per player, not sure what other platform can be better..

 

[E30M3]

agreed..

No platform is perfect but at least they do the work on an individual basis per player, not sure what other platform can be better..

Yeah there's not really any other data sets to use. You would figure this is a prime target for some company to come in and compile numbers a little better, but most of these things are nuanced in scheme, formation, etc. Its a really tough thing to do.

Amazon Nextgen should be ready in a few years, but at the moment they are still compiling data and only putting out the cool stats like fastest x or probability stuff.

 

[djphinfan]

Yeah there's not really any other data sets to use. You would figure this is a prime target for some company to come in and compile numbers a little better, but most of these things are nuanced in scheme, formation, etc. Its a really tough thing to do.

Amazon Nextgen should be ready in a few years, but at the moment they are still compiling data and only putting out the cool stats like fastest x or probability stuff.

I prefer the human grade each play, just have to make sure they are qualified to a certain extent

 

[PhinFan1968]

With Tuas speed to throw I can totally see our Oline losing their individual battles off the snap and the defender not affecting Tua.

Been happening since week 1.

 

[NBP81]

Yeah there's not really any other data sets to use. You would figure this is a prime target for some company to come in and compile numbers a little better, but most of these things are nuanced in scheme, formation, etc. Its a really tough thing to do.

Amazon Nextgen should be ready in a few years, but at the moment they are still compiling data and only putting out the cool stats like fastest x or probability stuff.

I prefer nextgen stats also because it's 100% human error proof.