Roundup 10/2/2010

Are Hall of Fame coaches really just products of great quarterbacks?

You've heard of WOWY stats (With Or Without You), like hockey's adjusted plus/minus. Now here's a WOWRM stat: With Or Without Randy Moss. I did something (primitively) similar with Troy Polamalu last season.

Gregg Easterbrook thinks the 3-4 defense is part of a cyclical fad. He also says that good players are far more important than schemes. He's right. The small advantage I found for the 3-4 is just that--small. The rest of the variance in performance among defenses has to be attributed to other things, most notably player ability. On the other hand, every advantage counts, no matter how slight.

I also think Easterbrook is right about the 3-4 trend being part of a cyclical trend. But it's much more meaningful than a fad. It's part of a concept I call strategic intransitivity, something I learned from Michael Maubossin, author of Think Twice.

A good example of intransitivity is the game rock, paper, scissors. Imagine a paper-scissors world, one where one strategy has a strong upper hand over another. That's where today's defenses find themselves. Defenses have been the paper, shredded every Sunday by scissors-wielding Peyton Mannings and Tom Bradys. Slowly the paper strategy will adapt towards a more rock-like strategy, pounding the scissors into little pieces. Then offenses respond with more paper-like tactics, and around we go. The evolution of these strategies occurs over decade-long periods. The bottom line is that there is no single "best" strategy, only successful strategies given the current strategy choices of your opponents.

When you have your own website, you can see where other sites are linking to you. Maybe it's vain, but it's fun to click around the various other blogs and message boards to see what people think of my site. This has to be the highest praise this site has ever received. Here it is in English. Tack, Karl!

Luck in Australian rules football.

Steven Johnson on innovation.

Andy Steiner looks at whether offenses can improve their production by choosing to pass more often. This is the kind of thing I was hoping would make its way to Advanced NFL Stats Community. Do you have some analysis to publish? All submissions are welcome.

Are the 3-0 Chiefs for real? I think so, specifically thanks to their defense. But I don't think the luck on offense and special teams will continue. Their offensive EPA is near the middle of the pack, but their offensive SR is near the bottom. I think that will catch up with them.

How to write a news article about a research paper. So true. Helmet-knock: Tango.

Carson on sabermetrics as "words, numbers, and baseball." A very thoughtful essay.

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

5 Responses to “Roundup 10/2/2010”

  1. Jim Glass says:

    Are Hall of Fame coaches really just products of great quarterbacks?

    I'd be much quicker to believe that "great" QBs are good QBs who get a rep for being "great" by playing on great teams put together by great coaches and great GMs.

    According to his biography, Vince Lombardi believed that football is the "most team play" of all sports, and a good QB gets everyone else on the offense into the game contributing optimally, so the team shows its quality as a group.

    In contrast -- this was his initial fear with Starr, who been a bottom-pick bench warmer -- a really bad QB can take everyone else on the O out of the game all by himself by screwing up while the ball is in his hands. So a really good QB's numbers reflect the quality of the entire team, while a really bad QB's numbers are all his own as he destroys the team by his sole self.

    The consequence is that the difference the QB makes in the W-L is a lot larger as the QB's ability rises from "really bad" to "average", than it is as it rises from "average" to "top". Or, a bad QB hurts more than a good QB helps. (To anyone who's studied economics this makes perfect sense: diminishing returns.)

    This runs very much counter to the popular idea that the QB is "the man" on the team, the QB "carries the team" -- it is the *top QB* who wins it all. And it is a pretty extreme version even of that to say "the QB makes the HoF coach!"

    If Joe Montana and Archie Manning switched teams, does anyone really believe John North and JD Roberts would be in the Hall of Fame and Bill Walsh would be a forgotten failure?

    Successful QB play is an two-way interaction between the QB and 10 other players on the field. His numbers aren't his, they are *theirs*. When Moss and Welker showed up in NE, Brady's numbers suddenly rocketed up way higher than ever before. Did he suddenly get better that year? figures the difference between the average starting QB and backup replacement-level QB is a little over 2 points per game, 32 points per season. From looking at how NFL GMs spread payroll, and from some other back of the envelope figuring, I get the difference between an All Pro QB and average-quality starter is maybe 12 to 20 pts a season.

    The difference between a good team and expansion-type team of replacement players, or between a championship quality team and average 8-8 team, runs 150-200 points, a heck of a lot more than the QB accounts for. And that number is what the coach and GM account for.

    But the TV camera is always on the QB, so he gets way too much of the credit that belongs to the rest of the team for being good or bad.

    An intersting fact to me is that you can find a whole bunch of HoF running backs who played on bad teams forever. OJ Simpson's teams averaged 4-10 for his career. Barry Sanders, Walter Payton, Gayle Sayers, the list goes on. But you can't find a HoF QB who played on bad losing teams year-afer-year.

    So either for some reason bad teams have never drafted even one top QB the way they draft so many top RBs, or they did draft top QBs who were never recognized as such because they took the blame for the rest of the team. Which implies the QBs on great teams are "over recognized" the same way.

    The team makes the QB much more than the other way around, and the coach and GM make the team. So HoF coaches get their QBs into the HoF.

  2. Benjamin Morris says:

    Jim: Nice comment, I wish you had made it on the original post (for future reference, I have no problem with cross-commenting). The HoF RB point is interesting, though obviously a lot of factors could be at play. Also, I wouldn't put much stock in those PFR numbers -- while their effort is praiseworthy, statistical analysis of true player value is very much in its infancy.

    As for your argument about no great quarterbacks on bad teams, I don't think it proves what you seem to think it does. Clearly, by most metrics, a QB of a given quality will under-perform on a bad team and over-perform on a good team. But also clearly, teams of a given quality will under-perform with a bad QB and over-perform with a good one. Without further information, this state of affairs tells us nothing about the relative value of each.

  3. James says:

    Jim, the first response that came to mind about HoF RB vs QBs on bad teams: when a bad team drafts a HoF QB, they are no longer bad teams and remove themselves from your sample.

  4. Jim Glass says:

    when a bad team drafts a HoF QB, they are no longer bad teams and remove themselves from your sample.

    That makes perfect sense if you believe that football basically is a one-man game, "the QB makes the team".

    But it is pretty hard to square with's empirical finding that the difference between a starting QB and backup is only a little more than 2 points a game, and NFL payroll data that show on average the QB gets about 11% of the pay of the top 22 players and 9% of top 35 -- instead of 50% or 60%. Even Peyton Manning had only 14% of the Colt's payroll last year instead of what, 70%?

    It's also hard to square with common sense. Each team has 22 starters and say a dozen key special teamers and situation players who affect each game. (Let's ignore the rest of the 53 on each team as equivalent replacement level players.) Give each starter a "share" of responsibility for the quality of the team, and each of the other dozen a half share. How much is the QB worth compared to the average other starter? Two players? Three? If the QB is worth three other starting players, he gets 3 of 30 shares and accounts for 10% of the team's success. Which is about what NFL GMs give starting QBs as their share of payroll.

    So if you've got a bottom-quality team and want to have a top-quality one, drafting a top-quality QB will get you about 10% of the way there. Which leaves your team still 90% bottom-quality, which doesn't appear removed from the sample.

  5. Richie says:

    Technical note: your link to "Think Twice" actually links to a different book.

Leave a Reply

Note: Only a member of this blog may post a comment.