Let's give WJ3 some jungle love

[omgdrdoom]

11 minutes ago, fredtoast said:

In 2015 Dre played 98% of the snaps on a defense ranked #5 against the pass (efficiency rating) and he ranked in the top 30 CBs in most objective stats (completion percentage allowed, success rate, yards allowed per target), yet PFF had him ranked as the 112th CB behind a bunch of scrubs who barely even played. 

 

He was the perfect example of how screwed up PFFs individual player rankings can be.

I think he was rated so low in 2015 because of a couple of factors.

 

1. He did have a couple of games where he got wrecked in coverage

2. T-3 most penalized player (any position) in the league

3. T-10 most missed tackles in the league (again, any position)

 

PFF will hammer you for that shit. I remember seeing some guys play a decent game but get negative scores for the game due to a couple of small, inconsequential penalties.

 

Not saying I agree with their grades or anything, but I think that could explain why he was so low.

 

Dre gave up 50+ yards and a TD in consecutive weeks in games 2 and 3 that year, and then he followed that up with being targeted 10 times against KC and gave up 7 catches for 122 yards in that game where Santos kicked like a bijillion FGs against us. He had a defensive holding penalty that game too so he just got fuckin' whacked a few weeks in a row and started off that year as a trainwreck of a CB. He cleaned it up by the end of the season though for sure.

[SlantNGo]

PFF weighs each play equally. For example, Muhsin Mohammad often ended up being one of their highest rated WRs because he dominated so much better than other WRs at run blocking. They have separate splits for run blocking and pass receiving for WRs, for example, but they usually just end up looking at the overall score.

 

Dre has got to be one of the worst tackling starting CBs in the league, and CBs make tackles (5+ times a game) way more often than they deflect passes (maybe 1 or 2 a game). My main criticism of PFF is not in their play-by-play scoring, but their poor aggregation. They are well behind FBO in this regard.

 

PFF's new scoring system (0 to 100) is even worse. I think they're just transforming the old grades into a scaled sigmoid function (https://en.wikipedia.org/wiki/Sigmoid_function) so that they end up between 0 and 100.

[fredtoast]

As I have said before PFF keeps track of a lot of good stats, but the way they formulate their final scores is a bit of a problem.

 

Player A can play fifty snaps and mess up twice (4% of the plays) to get a score of -2.  Player B can play only 4 snaps and mess up once (25% of his plays) yet his score of -1 will place him ahead of player A.