Normalization

The recent update to our grading normalization explained.

| 1 year ago
icon-close

Normalization


icon-closeNormalization.

We’re not fans of words that make it appear that what we do at PFF is based on statistics. However, in this case, nothing else can be used to better describe the process that helps turn our raw player grades into what you see on our Premium Site.

Why do we do it?

Well if we wish to keep the average grade for all positions as zero (which we do) then we need to accommodate those which only accrue negative or positive grades in certain circumstances. For example, if an offensive tackle (player A) gives up 30 quarterback disruptions on 400 pass plays that’s almost always a lot better than (Player B) doing the same thing on 175. So, if the raw grade for Player A was -27.5 and that grade for Player B -25.0 we need to normalize those figures based on what we would expect to happen by adding a small positive number for every pass play where no pressure is allowed.

It can be significantly more complex than that, but this is not a treatise on our full grading system – it’s an explanation of why some things on the site are changing.

So, this week we’ve updated our normalization factors across positions to account for two things:

1) An increased sample size more reflective of the modern NFL.

2) A change in how we graded penalties. Previously, any penalty went into the ‘penalties’ grade, whereas now penalties born about by bad play in coverage, pass blocking, run blocking etc. land in those categories, leaving penalties reserved for discipline-related issues like personal fouls and pre-snap infringements.

So, if you’re wondering why our grades have changed this explains that. And if you clicked on this page hoping to see a little more, then please enjoy some of these grade snippets from our seven years of grading:

•  Behold Robert Quinn who in 2013 amassed a +62.3 pass rushing grade. To put that in perspective the next best ever achieved was a +54.5 grade by Tamba Hali in 2010. His breakout season was truly a season for the ages.

•  While J.J. Watt lost the first triple digit PFF grade, he only did so by a fraction with a +99.8 grade in 2013 to go with his +94.2 effort a year earlier. The two-time Number One in the PFF Top 101 still has a sizeable lead over the rest of the competition with the next best effort a +39.2 showing from Justin Smith in 2011.

•  Only three guards have ever topped the +40.0 mark with Logan Mankins, Jahri Evans and Evan Mathis part of a very select club. What makes Mathis so impressive is that he’s done in each of his last three years. In 2014 can anyone else join? Larry Warford finished his rookie season with a +24.4 grade so could be a guy to watch.

This is just a snippet of some of our standout grades that we have at every position. If you want to see them all for 2007 through to 2013 while getting access for the 2014 season then sign up for a PFF Premium package today.

 

 

Follow Khaled on Twitter: @PFF_Khaled

  • J.d. Keim

    I understand the change in grading of penalties, and I think it’s great. However, I don’t quite understand the update based on ‘increased sample size.’ How does that tie into, for example, Brent Celek’s 2013 run blocking grade increasing by 8 or 9?

  • Ben Fitzgerald

    I don’t understand how a DE’s grade could improve due to normalization. Could you explain how all the different positions are affected by normalization?

    • Shizzle Dawg

      The average grade for ILB’s was -4.4 last year. So if they’re shifting the average to 0, most of the ILB’s grades would increase, but I’m not sure about that.

  • Nunya Beezwax

    Took away Watt’s triple-digit benchmarks… boo!!!

  • jtruff

    Now if you would only adjust grades to be on a per play basis…

    A player who earns a +10 grade on 400 snaps is a higher quality player than one who earns a +10 grade on 600 snaps. The corollary is that a guy who earns a -10 grade on 400 snaps is a worse player than one who earns -10 on 600 snaps. As is, those players are judged to be identical when, in reality, they are not.

  • Brett Vaccaro

    If you’re willing to praise Jahri Evans so much why are you grading him at merely “good starter”? He is a 5 time Pro Bowler and 5 time All-Pro from 2009-2013. That is 5 years of sustained excellence leading into the 2014 season. What more does he have to do to be seen as “high quality”?

    • jtruff

      They didn’t praise him “so much.” They made one comment that he had a season among the best they’ve graded among guards. That season was 2009. He’s graded out as the 30th, 9th, 8th, and 17th best guard from 2010-2014. PFF now has four more recent years of him as, on average, the 16th best guard in the NFL. That’s good, but certainly not “sustained excellence.” It’s more “sustained pretty good-ness”.

      As for All-Pro’s and Pro Bowls…we all know how those work–especially when it comes to linemen. Clueless voters just pencil in whoever made it the year before. Once a lineman is in he gets a lifetime scholarship (right Maurkice Pouncey?). That’s how Jeff Saturday made the 2013 Pro Bowl despite playing so poorly that he was benched before the season was over. A lineman’s actual performance lags far, far behind their name recognition when it comes to post-season awards.