Rethinking Quarterback Stats in the Wall Street Journal

Mike Sielski has written the following article for the Wall Street Journal: Rethinking Quarterback Stats: Vick’s Heroics Show How Traditional QB Ratings Short-Change Runners; Dr. Berri’s Theorem

Yes, Mike’s artice reference the research on NFL quarterbacks presented in The Wages of Wins and Stumbling on Wins.  Along the way, he also details the history of the NFL’s quarterback rating.  All in all, a very good read (and I am being completely objective J.

Mike’s article also focuses on Michael Vick (a popular topic at the Wall Street Journal).  Let me add to this discussion by reporting the Wins Produced rankings after Week 10.

Michael Vick is currently producing 1.036 wins per 100 plays.  In Stumbling on Wins we presented Relative Wins Produced per 100 plays, or WP100 adjusted for both current and historical averages.  Such an adjustment allows one to compare performance across time.  Vick’s current RWP100 is 0.906.  And if you have your copy of Stumbling on Wins handy, you can see that Vick’s performance would rank 8th among all quarterbacks who have played since 1970 (you can also see exactly how RWP100 is calculated).  Who are the other seven quarterbacks? Well, go look in the book :)

Let me close by noting once again the “outstanding” Detroit Lions. Yes, they are 2-7.  But as I have noted, Pro Football’s SRS measure (based on point differential and strength of schedule) ranks the Lions as the 4th best team in the NFC.  Across the entire league the Lions rank 11th in SRS.  Michael David Smith – of the Wall Street Journal – has labeled the Lions the unluckiest team in the NFL.  He also notes that teams with losing records and positive point differentials tend to post much better records in the future.

And today I noticed that Wayne Winston has an NFL ranking that says the Lions are the 14th best team in the NFL.  So the evidence the Lions are “good” keeps rolling in.

Of course, critics of this research note that this is driven by the Lions 44-6 victory over the 4-5 St. Louis Rams.  If one throws out this game, the Lions no longer have a positive differential.

Let me respond by noting…

  • I am a Lions fan, so I am simply picking out whatever evidence I can find the says the Lions are “good”.
  • Furthermore, the sample size we are talking about is nine games.  And that sample is only going to 16 games.  One can throw out stuff from any team’s “sample” and change the conclusions you reach.  That is the nature of NFL analysis (in other words, one can devise a different ranking that emphasizes other stuff and reach a conclusion different from Pro Football Reference and Wayne Winston).
  • And remember, in basketball teams play 82 games.  After nine games (or even after 16 games) we tend to treat all conclusions as very tentative (because we know stuff can change across the remaining 80% of the schedule).

I do think that if the Lions got to play 82 games against a schedule that was more representative of the entire league (they have played the 3rd hardest schedule in the NFC), the Lions record would approach the 0.500 mark. 

But if you don’t like that conclusion, you can reach any other conclusion you like.  Again, the sample we have is quite small.  And small samples don’t allow us to reach definitive conclusions (although I am going to keep saying the Lions are “good” as long as I can find any ranking or shred of evidence that allows this conclusion).

– DJ

P.S. By the way…if you leave a comment disagreeing with what I say about the Lions, you might find your comment in the trash (well, you can’t see the trash but that is where I will send it).  Feel free to disagree with the quarterback stuff if you like J

Comments are closed.