...as I hinted at before, I think the thing that wins a football game depends very...very heavily on situational coaching and the strengths and weaknesses of your opponent. You're not going to see that in statistical correlations. The dichotomy you present between "rush defense doesn't win many regular season games" and "rush defense is PARAMOUNT in the post-season" can be explained by the strengths and weaknesses of football coaches and the tendency for post-season-bound teams to be more well-rounded than the average NFL team.
Against a generic .500 club during the regular season, you're far more likely to be able to find and exploit obvious weaknesses en route to victory (or get exploited en route to defeat). Example: the Seahawks easily crunched the Cardinals by exploiting the terrified looking rookie QB. Carroll loaded the secondary with extra help so the rookie had no clear passing lanes and the kid panicked and did horribly. That's regular season football. Did the Hawks play much rush defense last week? No they did not...they didn't need it.
In the post-season, though, you have to be able to stop the run or the other team is going to control the flow of the game and eat time off the clock when it matters. Playoff games are battles between very-nearly-equal teams with strong rosters and not a lot of gaping holes to attack...at that point it becomes more about who can manage the game better...(head coaches matter here...run defense matters here...effective running game matters here)...who can take away options from the opponent more often, etc.
See...this is what I mean when I say football is a situational game and baseball...really isn't. This is why there's no signal of clutch in baseball and there's a HUUUUUGE clutch signal in football. Evenly matched opponents win games by having more options than their opponents...just like in chess (a club level player pursues his plan A and maybe has a plan B...a grand master will CLOBBER this kind of player because he's got 30 ways to kill you in his mind at all times). Doc...all your stretched chess metaphors make a heck of a lot more sense in football than they do in baseball. :)
My 0.02
Baseball sabermetrics and football statistical analysis (we need a different name...SABR is the society for american baseball research...you cannot call the analysis of football sabermetrics) need to be approached COMPLETELY differently. Football statistics are in their absolute blastocyst stage...they aren't even infants yet...and the reason is...in football the situation matters a whale of a lot more than it does in baseball. In baseball, every play is one hitter, one pitcher, and three defensive units.
In football, every play involves AT LEAST 18 of the 22 guys on the field and you can't really statistically separate the different offensive and defensive units that have to work together to produce a result...and it gets worse than that...because in football there's a CLOCK...that clock dictates policy...efficiency ratings can be totally FUBARed by a team having a 2-possession lead with 10 minutes to play.
Football is a much harder problem...I'm not smart enough to figure it out without years of study...study much more dedicated than even what I've done with baseball. We need a totally different approach to football...every play must be viewed through a more specific lens than what is commonly stored in the play by play records. We need to make note of who were the key blockers, the key corners, etc. that made events unfold the way they did. Not just Hasselback to TE for a 9 yard TD pass...we need specific accounts of what each guy did on any given play.
.
cpoints x100.
Point well taken on SABR. Maybe it's SAFR to stay away from football with performance analysis? :- )
And yet, it could be fun to bounce SAFR vs SABR off you Matt ... as to your "not smart enough" comment, it reminds of one James comment. "I don't know whether you need a great bullpen to win, and I doubt that Jerome Holtzmann does, either," or somesuch.
.....................
Have you seen this site? Lot of valiant attempts to apply correlations and regression to stuff like "what is the optimal run-pass ratio on 2nd and 3rd down," and the main author seems to have a pretty good idea that he doesn't have 950/1000 light bulbs on.
Yet, there are plenty of times that they don't fully grasp the idea of interdependent variables -- especially as it relates to the factor of play-calling, and defensive positioning......
.
=== Now THAT's a Bonds Shift, Dept. ===
We're used to seeing two infielders move radically for the LH pull sluggers like Thome, Griffey etc. But what do you call the 6- and 7-man defensive backfields that Pete Carroll used OVER HALF THE TIME last Sunday?
If the Giants shifted Ryan Howard with 4 infielders and 2 outfielders to the right of 2B, and just the LF over on the other side ... would Howard's stats be worth anything for those AB's? Yet this is the situation we're trying to capture, down after down, in football...
Meanwhile, Howard keeps stepping back and forth between RH and LH batter's boxes, guessing against the shifts...
........
How do you compare run-D efficiency when the defense has 4 DB's vs 7 DB's on the field? Wow, what happened to the Seahawks' run defense? :- )
We can choose to put the linebackers back on the field next Sunday, and then what happens to your run success ratio...
...........
So if a team can't pass, the defense "Thome-shifts" 8 guys into the box, and these radical "shifts" occur on every play, constantly morphing into other shifts...
This is not taken properly into account, and cannot be...
SAFR analysts don't seem to quite get it here -- hey, half the time SABR analysts don't get the complexity of their problems. A dialogue between SAFR and SABR guys would be helpful. Mostly for the SAFR guys, if we can venture a bit of ego'ed-out-sounding opinion there...
.
.
Comments
That, in general, the SAFR principles apply in a certain way between two mismatched (or flawed) teams and in a different way between two well-matched and unflawed teams...
Concept of critical mass applies: CF defense doesn't decide a baseball game, unless you've got Prince Fielder out there, in which case it decides most the games...
But many other factors apply too, to create the sudden emphasis on run in the NFL playoffs, we're sure...
...I'm not smart enough to do more than throw my best guesses out there as to the nature of football, statistically. The records aren't detailed enough, the game isn't discrete enough, and I don't have as much feel for the physics of the game as I do for baseball either. I watch maybe 8 football games a year...I watch maybe 80-120 baseball games a year...big difference. :)