Wednesday, August 17, 2011

Explaining ESPN's New Quarterback Rating System (Or Trying)

If you watch ESPN at all, you're going to be hearing about a new Quarterback Rating System (QBR) that they have developed.  You're going to hear about it a LOT.  Trust me, they will ram it down our throats until we simply accept it as gospel.

I still think that virtually every NFL game will use the old quarterback rating system (which is equally as impossible to understand), but Monday Night Football will use the QBR.  I could see where Ron Jaworski will rave and rave about it.

So, in an effort to embrace and try to understand this insanely complicated new system, I am going to attempt to explain it to you here.  We'll be breaking it down into identifiable sub-parts to come up with the whole and then discuss some criticisms.  

Here goes:
  • There are three main parts to the QBR (by the way, this is all computerized):
    • Expected Points Added
      •  Every single play a quarterback is involved in is scored based on how much they add to a win.  
      • ESPN will determine "expected" point totals for each situation and assign a certain point total to a quarterback based on the results of each individual play
    • Dividing Credit  
      • ESPN will use their own analysis to assign credit for a certain play; such as, a quarterback throws a 3-yard vertical pass to a receiver who then goes 40 yards for a touchdown
      • Within that particular play, there are dozens of possibilities: pass protection, whether the receiver ran the right route, and even whether the throw caught the receiver in stride or was a little behind.
      • Every single one of these factors will go into calculating how important a quarterback was to a specific play outcome
    • Clutch Index
      • This element looks at how "clutch" the play was in producing expected points and is based on analyzing different plays in determining how important they were to the overall win probability  
      • Therefore, a 15 yard pass for a touchdown in the first quarter will not be worth as much as one in the fourth quarter to tie the game.
  • Total scores will range from 0-100 for each individual game, with 50 being an average score.  It is apparently possible for a quarterback to have an individually great game and get a 90, but ESPN expects a Pro-Bowl level quarterback to have a rating between 65-70 for the season 
  • Obviously, ESPN will control the analysis and do the scoring themselves, as their computations and analysis will be proprietary
  • For instance, ESPN applied the QBR to the 2010 season and named the top QBs:
    • Tom Brady - 76
    • Peyton Manning - 69.5
    • Matt Ryan - 68.6
    • Aaron Rodgers - 67.9
    • Michael Vick - 66.6
    • Drew Brees - 65.9
    • Eli Manning - 64.3
    • Josh Freeman - 63.5
    • Philip Rivers - 63.2
Does that look right that Rivers is ranked below Eli Manning or Josh Freeman? Well, it is in the new system!

I understand that it's supposed to be able to tell a quarterback who is truly making clutch plays in crucial times versus a quarterback who has a good game but always fails in the clutch.  That way, a quarterback's true performance is measured and not simply their statistics.

Brian Burke over at Advanced NFL Stats has taken a good look at this new system, and has a few concerns that we'll note here:
  • He doesn't like the fact that ESPN isn't sharing exactly how they've come up with the formulas for how much different situations count.  Methodology should be open so that it can evaluated by outside sources for objectivity and correctness.
  • It mainly is a ranking system; a way of determining which quarterback is truly playing better.  However, it doesn't tell you how much added value that they give to their team.
  • Ultimately Burke likes it and thinks it's a step forward
I contend it is yet simply another thing that ESPN can gloat about, and that will ultimately force others to refer to in the future.  How much the NFL will use it will depend on how reasonable it seems once it's in use for a full season.  Having to rely on ESPN's analysis for scoring probably won't sit well with the NFL.

No comments: