Writers of Pro Football Prospectus 2008

27 Mar 2008

FO on ESPN: Quarterback Abuse 2007

This week's ESPN.com column looks at which quarterbacks were knocked to the ground most often in 2007, and which offenses protected quarterbacks the best when we count both sacks and hits. I know, I know, people will complain that this is yet another Patriots-centric article, but I was wracking my brain trying to figure out a good anchor for the piece, and it was the only thing I could come up with. We'll look at defenses (and defenders) with the most hits next week.

Posted by: Aaron Schatz on 27 Mar 2008

23 comments, Last at 01 Apr 2008, 9:40pm by Will Allen


by mm (not verified) :: Thu, 03/27/2008 - 11:38pm

It'd be interesting if you could combine some of this with some of the other data you're gathering. For instance, is there an offense particularly good at protecting a QB when the defense rushes 4 but poor when they rush 5?

by Tom (not verified) :: Fri, 03/28/2008 - 12:58am

Great article! Greatest team ever (excluding fumble luck!). Brady, Cassell, it doesn't matter: 19-0 in '08!

by Sam B (not verified) :: Fri, 03/28/2008 - 7:48am

interesting that Cleveland's offence was second lowest in terms of hits.

Amazing turn around in the offensive line, but also suggests that DA wasn't having to do anything like as much throwing into pressure as other top-DVOA QBs this season.

by John (not verified) :: Fri, 03/28/2008 - 8:24am

Re #3: Might not the low hit/sack numbers for Cleveland be equally attributable to DA's quick release? No argument that the line was vastly improved relative to earlier squads, but DA is reputed to have a quick trigger.

Also, I find the ratio of hits/sacks interesting. The three QBs who have the highest hit/sack ratio (Palmer, Brees, Manning) have a reputation for being intelligent with the ball. This bodes well for young Edwards of Buffalo. Perhaps this is unsurprising; he is a Stanford alum. According to Jim "Not a Michigan Man" Harbaugh, they do real schoolin' there.

The hit/sack ratio reminds me of the BB/K ratio in baseball. That is, hitters with more patience tend to have a higher BB/K ratio. Gamblers tend to have a lower BB/K ratio. That doesn't mean that the BB/K ratio is determinative of a good player, though.

Likewise, the hit/sack ratio may separate the more cautious QBs (who dump the ball quickly to avoid the sack) from the gamblers (who try to extend the play or hang tough in the pocket).

Nice work.

by steelberger1 (not verified) :: Fri, 03/28/2008 - 10:50am

I think the hit being only if the QB is knocked to the ground to be a bit misleading.

I think the "simple contact isnt enough" part skews the outcome and paints an inaccurate picture.

QBs can be hit without falling down. So, if he doesnt fall down...does that mean there was no pressure on him?

by Joseph (not verified) :: Fri, 03/28/2008 - 12:19pm

Wow, I knew Sean Payton got pass-happy last year, but 699!!! passes? That's 43 pass plays per game! However Brees only took 9 sacks, 1 per two games. It will be interesting to see if losing Jeff Faine at center will affect these sack numbers (his backup, a UFA, was resigned at a much more reasonable contract).

by Dr. Mooch (not verified) :: Fri, 03/28/2008 - 12:27pm

Someday I need to see Rob Johnson's numbers in these stats.

by Dinger (not verified) :: Fri, 03/28/2008 - 3:24pm

You admit that the stats behind the article possess flaws(Oakland, Denver examples), so why write the article if you believe the stats are flawed? Just doesn't make any sense.

by Bobman (not verified) :: Fri, 03/28/2008 - 3:58pm

I assume QB pressures are yet a third measure. I know they are supposed to be more reflective of the D, but can we add them in here as well? (Or is that only an FO charted item?) While it doesn't measure QB physical abuse, it does measure mental abuse, getting him trigger-happy, flushing from the pocket, etc.

Or is QB hits the same as QB pressures?


by Tom D (not verified) :: Fri, 03/28/2008 - 4:19pm

Re 8:

Did you miss this: "So if we want to see which quarterbacks are taking abuse, it probably is a good idea to adjust hit totals based on the tendencies of the scorers compared to the league average?"

by Bob in Jax (not verified) :: Fri, 03/28/2008 - 5:32pm

This is why I love Football Outsiders. Nice work, Aaron! BTW, where did the other 11 teams rank?

by Sam B (not verified) :: Fri, 03/28/2008 - 6:02pm

#4 while DA's release was undoubtedly a big step up from Frye's, I doubt it ranks particularly highly among QBs - maybe better than average, at a guess.

I certainly doubt that DA's release is quicker than Manning's or Brady's, but he was was pressured less than them.

My point is that this is reasonable evidence that the Cle line was significantly easier to play behind this year than years past.

Either that, or DA is better than Brady/Manning.

by andrew (not verified) :: Fri, 03/28/2008 - 8:18pm

Minnesota's pass blocking the worst in the NFL... yeah, pretty much confirms what I saw charting them all year. For the most part it did not seem a technique problem of anyone person, rather more of a scheme one. Matt Birk is continually lauded as one of the best centers in the NFL, but they always seemed to have their people in the wrong places for blitzes. I don't know if that's Birk's or T-Jack's call (I think Birk's)... More than once I recall six vikings blocking three rushers on one side, and one viking left alone against two rushers on the other side.

Coincidentally Birk is staying away from OTAs this off season to show his displeasure that the Vikings have not offered him a new contract yet...

by John (not verified) :: Sat, 03/29/2008 - 9:42am

Re: #'s 3 and 12. You missed that I agreed with you. No doubt, Cleveland's line was better. No argument there, then or now. My only point was that fewer QB hits are not just due to an improved O-line, but also a QB with a quick release. Frye's documented struggles behind the exact same line demonstrated that fairly clearly.

Still, I think everyone would agree DA's release is better than "above average." Note that New Orleans (and Drew Brees' quick release) was the only team better than Cleveland in terms of fewest QB hits. Of course, both lines were solid last year. Which is perhaps why they were the top two teams.

One more thing: quick release is not the only measure of a good QB. It may well be that Brady/Manning hold the ball longer, on average, than DA. Doesn't mean DA's better (or even close).

by Hummingbird Cyborg (not verified) :: Sat, 03/29/2008 - 10:39am

Wait a gull durn minute.

Didn't Cleveland lead the league in yards per completion? Doesn't that speak well towards their pass protection.

It also interests me that Dallas ranked 2nd in yards per completion.

Anyhow, it seems to me that if a team ranks well in yards per completion and well in fewest QB hits, it follows that their line pass protects very well.

Of course, it is also possible that they have a good QB at recognizing pressure and throwing it away instead of taking the sack and that lowers their QB hits. In that case, they should also have a lower completion percentage.

THIS does match the Browns, but it could also be that their lower completion percentage has more to do with having a game dependent on the deep ball and the fact that the deep ball is more difficult to complete.

I'd like to see this done again with your game charting research.

I find it impressive that Dallas ranks so well in pass protection, yards per attempt and completion percentage (7th).

That speaks very well towards their pass protection.

by Mr Shush (not verified) :: Sat, 03/29/2008 - 1:08pm

How many Dallas games did you watch last season?

I hate Tony Romo (don't ask me why) but you don't have to watch him to long to realise that his outstanding combination of pocket awareness and mobility enable him to hold the ball far longer than most quarterbacks could without being sacked. I think it's more about Romo than the protection.

Writing this post made me think about the website linked on my name for the first time in ages. I'd almost forgotten how awesome it was.

by Hummingbird Cyborg (not verified) :: Sat, 03/29/2008 - 3:17pm

#16: Perhaps I'd be well served to watch a game or two before forming an opinion.

Still, I find myself doubting whether a poor line could allow for a ranking that well in all three categories. Maybe he makes them look better than they appear, but he can't do it all alone.

by Arson55 (not verified) :: Sat, 03/29/2008 - 6:17pm

Some of it is greatness from Romo, but the line is a reasonably good pass blocking line (even if the run blocking at the end of the year was cringe inducingly bad).

by Dinger (not verified) :: Sat, 03/29/2008 - 6:48pm

Re #10
I read that part. The fact remains that the underlying data is flawed even if they are adjusted in some fashion.

by Alex (not verified) :: Sun, 03/30/2008 - 2:01am

You admit that the stats behind the article possess flaws(Oakland, Denver examples), so why write the article if you believe the stats are flawed?

Because even with the flaws, the stats still provide useful information. Even if only one scorer had missed one hit in one game, and all the other hits were accurately recorded, the stats would be "flawed". Does that mean we shouldn't use them, just because they aren't absolutely perfect?

Besides, we're not trying to figure out whether Ben Roethlisberger was hit 64 times or 62, we're trying to figure out if he was hit more than most other QBs, and why. Even if there are minor flaws in the data, we can still use it to make qualitative conclusions, like the fact that Tom Brady got hit a lot, despite taking very few sacks.

The fact remains that the underlying data is flawed even if they are adjusted in some fashion.

So? Flawed is not the same as useless. A slight flaw in the data isn't going to drastically affect the conclusions we base on that data. If we look at the data and say "Wow, Kurt Warner got hit a lot, even though he didn't take many sacks." That's still true even if it turns out that he only got hit 63 times, instead of 68.

We're not trying to create precise rankings of QBs and O-lines, we're trying to break them into categories like "more hits than average", "fewer...", etc, and see what teams in each category have in common, and how they're different. The sample size is small enough that even if the data weren't flawed, we still wouldn't be able to trust it to exactly reflect the true abilities of the players and teams involved.

Still, I find myself doubting whether a poor line could allow for a ranking that well in all three categories. Maybe he makes them look better than they appear, but he can’t do it all alone.

Well, last year, with Bledsoe starting, they gave up tons of sacks, then with Romo playing behind the same line, their sack rate went way down. Sometimes, it really is just the QB.

by Pat (not verified) :: Sun, 03/30/2008 - 3:56pm

I read that part. The fact remains that the underlying data is flawed even if they are adjusted in some fashion.

Have you ever actually worked to analyze a dataset? You're acting as if it's useless.

I dare you to find a data set which isn't flawed. Every single data set I've ever worked with in every field I've ever been involved with has had to be cleaned, bias-adjusted, and tweaked before it was statistically usable.

And we're not talking about just data sets involving humans. Astronomical, high energy particle physics, biological, genetics, all of them. The fact that you can tell a data set is flawed just means you need to clean it.

by justanothersteve (not verified) :: Mon, 03/31/2008 - 7:16am

#2 - To paraphrase Gertrude Stein, a troll is a troll is a troll.

by Will Allen (not verified) :: Tue, 04/01/2008 - 9:40pm

andrew, having the worst receivers in the league, and an inexperienced qb, will greatly affect the percentages of qb hits. Sorry, but you simply cannot measure pass blocking performance well without a stop watch. We are still in the stone age of football performance measurement.