Tuesday, August 25, 2009

A not-quite-crisis of confidence

I still see plenty of opinions about the conventional wisdom that states that a good running game improves the passing game and vice versa, contradictory to what I observed in a couple months ago:


Specifically (and against my better judgment), I've gotten into it a bit on a recent post on the Pro Football Reference blog. A commenter trotted out the following bits of wisdom:

1) You can't trust Tarvaris Jackson's numbers last year because of the small sample size; and
2) The Vikings' strong running game was responsible for Tarvaris Jackson's good numbers late in the season.

Note that that running game has been in effect for the last 32 regular season games and, by and large, Vikings quarterbacking has been mediocre to poor over that stretch, excepting the final four games of the 2008 season. So, T-Jack's numbers should be discounted due to a small sample size of games, but clearly the running game helps the team's quarterbacks -- which was only true for a small sample size of games. Eh?

But I'm not here to rag on a fellow commenter/blogger. Rather, I've been thinking about my theories the past few months and, while I still think they're mostly true, I can't help but shake some flaws with the analysis. They aren't necessarily deal-breakers, but they could show there's more to the interaction between running and passing than simple correlation.

As I've mentioned before, I'm no trained statistician or anything, just a football fan with a love of numbers and a willingness to challenge conventional wisdom. Maybe that'll get me in trouble someday, but until someone can come up with real, valid evidence to the contrary -- something more than "Well, it's obvious that a strong running game opens up the passing game, I mean, look at how much better quarterback X was when his team got running back Y!" which ignores all the times a quarterback doesn't improve when his running game improves -- I'll stick to my guns.

That said, here are some of the possibilities for why I might be wrong:

1) I'm looking at the wrong stats. Primarily, I use yards per carry and adjusted net yards per attempt, or an averaging stat for one (rushing or passing) versus a total yardage stat for the other (passing or rushing). Might be I'm comparing the wrong stats or, just as likely, I'm using the wrong tools to compare them.

2) Limitations on salary or coaching proficiency make it impossible to be great at both running and passing. Put simply, if you sink all your money into having a great running game -- a star back, big offensive linemen, a blocking tight end -- you're less likely to have anything left over to invest in quarterbacks and receivers. And some coaches might also just be better at engineering a great running game than passing game, or vice versa. This is the same concept that typically makes teams that are good on offense less likely to be equally good on defense, and it holds true for most major sports (hitting vs. pitching in baseball, offense vs. defense in basketball, etc.). My studies show that teams that run the ball well typically aren't exceptional at passing but maybe that's got less to do with there actually being a correlation from an in-game standpoint and more to do with the lack of ability (or funds) to construct a solid passing game.

3) Maybe there really is an effect, but it's undetectable. If you have the best running game in the league and the 20th best passing game, is it because there's no correlation between the two? Or is there a correlation and your passing game would be the 25th (or worse) best in the league without the strong running game? Using my Drew Brees-to-the-Vikings example, if Brees did take over at QB, maybe Adrian Peterson and Chester Taylor would have better years, but how would we separate Brees' effect on the running game from Peterson's, Taylor's, and the offensive line's normal development and improvement. I don't know.

I'd welcome anyone with strong analysis and the ability to use facts (instead of "conventional wisdom") to support or refute my arguments, because even I admit I feel a little tentative to expound my theories too loudly. But if you're going to disprove what I've written, at least try to be somewhat nice about it :)


Anonymous said...

There is no correlation between running and passing game, contrary to conventional wisdom and analysts. You gave many of the reasons but personally I think you missed two main ones.

1) Blocking. It requires a different skill set to run block well than pass block well. Most lineman do one better than the other. Most coaches build teams with a bunch of guys that generally do one better than the other. You touched on this general concept but focused more on the skill position players. Only elite blockers do both well but there are many guys out there that are great at one and terrible at another.

2) Most importantly, formations. WR's are good at catching passes. TE's, RB's, and FB's are good at blocking and running. Most good passing teams use 3 and 4 WR and don't run well out of those base formations. Most good running teams use TE's and FB's a lot with 1 or 2 WR formations and don't pass as well. When a running team has to pass, they bring on the extra WR. When a passing team has to run, they bring on extre TE's and such. Yet we sit scratching our heads when Green Bay uses 3 to 5 WR every play and can't run worth a damn. And we still scratch the ol head when the Vikings almost always have a total of 1 or 2 WR on the field at a time.

Now I don't think the Vikings could go 3 to 5 wide and be good at passing. I think then they'd suck at both running and passing. But that doesn't mean they can load the field with slow strong guys and expect to pass like mad men. Well that and their QB sucked but that is fixed this year for the time being.

Luis said...

Hey Jason. I just took the yards per attempt analysis one step further. Take a look.

By the way, how do I get people to visit my site?


Luis said...

Hey. Mr. Favre stepped up. But he's still not cutting it.

I intend to make millions by exposing the current qb rating as a hackjob.

Go to www.nflstatanalysis.net - I explain why it's a hackjob.