Couple More Thoughts On McNabb

Continuing this week's theme of absolutely ignoring Sunday night's must-win game against the Bears, I want to continue the McNabb discussion from Thursday.

As usual, you all came through in the comments with good questions and ideas, and Brad even went off and did his own research.  I want to get to all the issues raised, but first, a clarification about yesterday's post.

I wasn't arguing that the Eagles offense is in the ideal position right now.  The running game is a mess and the run-pass balance is even worse.  I'm truly flabbergasted that -- far from adding new "wrinkles" to the Wildcat each week to keep opponents guessing -- we've actually just kept stripping stuff out of it, to the point that it's really just the Mike Vick charity project.  (This post -- not mine -- has proved annoyingly prophetic.)

I actually agree with a lot of the things Lawlor has been arguing.  I just haven't seen statistical support for his grand unifying "deep ball infatuation" theory as to why things aren't as good as they used to be.

Now, back to McNabb.  First question, from ATG:

As to the 1-10 yards changing to behind the [line] passes, is that still an indication of a weak line? Is the running back releasing later because he has to help more? Are teams playing this section tighter, where half the passes go, and forcing us to throw it shorter or longer? Is it just because Westbrook hasn't been around to soak up those passes? Maybe a little of each?

To help answer that question, I pulled the situational stats for all pass catchers for the last two seasons.  But I immediately found something weird:

Passes Caught Behind The Line

2008 -- 24.9%
2009 -- 22.6%

Whoops, that doesn't fit the narrative.  There are two problems here:  1) The receiving numbers are not available by quarterback, which means the total numbers include Kolb's figures, and 2) The numbers I pulled today for receivers finally have the San Diego stats included.  One game may not seem like it could change the numbers for half a season, but remember, McNabb threw 55 friggin' passes in this one. 

Here's the adjusted table:

Lesson #1:  Sample sizes, sample sizes, sample sizes.  This is why you have to be careful throwing around some of those red zone statistics and other numbers I saw today.  Even with San Diego, you're still only talking about 6.5 games for Donovan.  Heck, McNabb's only thrown 23 passes in the red zone all year.  I feel pretty comfortable when we're talking about seven and a half years of situational stats, but week to week, there's going to be variation.

Lesson #2:  In the end, because the Kolb and McNabb numbers are mixed, there's really no receiver effect we can measure, at least at this point.  Westbrook caught 56 percent of his passes BTL last year and 53 percent this year.  Buck and Booker were at 66 percent last year, McCoy's at 70 percent this one.  Running backs as a group caught 27 percent of passes this year and are at 25 percent this year.  These aren't big variations.

- - - - - -

So what's with Kolb, anyway?  And how about that other guy who always comes up in these discussions ... from the professor:

For comparison, what do these numbers look like for Garcia's stint? His game management left an impression on me that an Eagles offense could dink and dunk to move the chains and chew the clock, if it ever chooses to.

Fair warning, the next chart is going to blow your mind.  It did mine, at any rate (definitely click for the full size):

 
That's completely nuts.  Looking just at the far right column, there's almost no significant difference between the distribution of passes called for Garcia in 2006 and McNabb from 2002 through 2008.

That was unexpected.

Now, obviously, we know that chart covers up some year-to-year McNabb variation.  Certainly.  But you're lying if you said you saw that one coming.

As for the completion percentages at the various depths, I'd warn that we're again dealing with small sample sizes.  But I don't find it remotely unplausible that Garcia was a couple percentage points more accurate than McNabb from short to intermediate routes.  McNabb's a 59 percent guy for his career, Garcia is 61.6.  Sounds right.

Understand, though, how that difference scales into the real world.  McNabb's completed about 63 percent of his passes from BTL to 20 yards this year.  Say Garcia would be at 66 percent from the same range.  That's about one short(-ish) pass a game McNabb misses that Garcia would have completed.  That's going to kill a few drives over the course of the season, but combined with the way the distribution of passes really wasn't all that different in 2006, it's not providing much of an explanation for what's different.

And keep in mind that you're trading that one pass a game for significantly better output on all the other passes he does complete.

- - - - - -

As for what explains the difference between 2006 and 2009, it's probably this:

 

Anyway, that chart's too depressing to even try to extrapolate to a 16-game season, so let's talk about Kevin Kolb.  Did you see his numbers up there?  Now that's a different offense.

- - - - - -

Next issue, from Brad:

The decline in the 1-10 yard pass (which is the WCO hot spot) has coincided with an uptick in behind the line passes. Completion percentages and ratings have remained high, but the total amount of the designed screens/check-downs has increased steadily.

On the face of it, this looks to be the most promising angle of attack for those looking to keep the "big play infatuation" theory alive.  Maybe it's not that McNabb's throwing so many more balls downfield, but because he's so focused down there, he's missing guys at the intermediate level and being forced to dump the ball off right before he gets sacked. 

The first problem is that this hypothesis is almost completely unprovable for those of us not privy to what Marty's actually calling (i.e., all of us).  Maybe we are more of a deep ball / check down offense than we used to be.  The second -- and far bigger -- problem with this argument is that even if it's true, it shouldn't be having the effect being claimed. 

Take another look at McNabb's 2009 numbers, and this time check out the YPA and YPC columns:

(If you go back to the big chart above, you'll find similar YPC and YPA numbers across his career.) 

Yes, McNabb's throwing more BTL passes this year, but because he hits on a much higher rate on those, the yards-per-attempt figure is almost identical to passes thrown between one and 10 yards.  So it doesn't really matter, in the end, if he's throwing slants or hanging around for awhile and throwing checkdowns.  The end result is the same.  You're getting five and a half yards (with presumably less variance on the BTL throws, as well, since the completion percentage is so much higher).

- - - - - -

Except -- and here's the big old caveat -- situationally this isn't true.  When you're in the red zone, you'd much rather be good at throwing a five-yard slant than a 45 yard bomb / check down. 

- - - - - -

As a further complication -- and though this question wasn't directly asked -- not all BTL throws are checkdowns.  The Eagles throw a lot of screens.  Might there be a significant difference in yardage gained between those two categories?

To answer that question I went to the 2008 FO game charting and pulled all the BTL passes from last year.  There does seem to be a difference:

I'm guessing the unclassified plays mostly belong with the dumpoffs, because their spreadsheet tries to force you to claim anything behind the line is a screen pass.  I'm not sure how they would define swing passes.  I know when I was charting I marked that for plays where McNabb didn't check everyone else on the field before going to his back out of the backfield.  But one charter's swing pass could be another's dumpoff. 

Bottom line, though, is that there is a difference, and it's probably fair to say that dumpoffs are about a yard worse than 1-10 throws.  So that's maybe two yards a game.

Nothing to sneeze at when you can't convert third-and-one, of course.

- - - - - -

And because Sam will point this out if I don't mention it, it's impossible to square those FO numbers with the ESPN stats above.  That's the downside of mixing and matching data sources.  I think both are telling us something interesting, however.

- - - - - -

Next question:

The only thing I am not sure your stats are accounting for, is when Donovan gets sacked or flat out misses an open guy underneath.  Is that because he is locked on downfield.

Trust me, I am hyper obsessed with Donovan McNabb's sack statistics.  I'm the guy who spent all last year arguing that McNabb's dramatically decreased sack rate had more to do with a change in how he played the game than with the guys blocking for him.  I still think the case is pretty airtight, but after the crap we've seen this year, I'll be a little less sanguine going forward on the importance of the offensive line.

Anyway, yes, McNabb's getting sacked a lot more this year than he did last year.  But he's only at a sack rate of about seven to eight percent (depending upon how you calculate it), which is both in-line with his career average and significantly higher than we saw in 2004 and 2006 (the other years he was bombing away).

As for the second part, that's already been discussed above.

- - - - - -

Last one:

My only concern would be that you're missing that this year's Deep Ball increase (+40), which is statistically significant, appears to be coming directly from the percentage of pass attempts in the 1-10 yard area.

McNabb's historical rate of 41+ throws is 2.2 percent.  This year it's 3.2 percent.  That 1 percent difference is not accounting for the 8 percent difference from 1-10.