Writers of Pro Football Prospectus 2008

27 Nov 2009

Sacks Have No Affect On A QBs Performance

I wrote a post recently (that happily no one read) showing a seemingly strong correlation between how often a QB gets sacks and his numbers. Based off of seven QBs between 1.5 years, it seemed that there was a strong correlation between QBs moving to a new team and how an O-line affects that QB. However, even I realized the huge sample size problem. So I decided to actually extended that sample size to see how sacks affect a QBs performance. To determine this, first, I decided to see how often a QB gets sacked per every 100 passing attempts (S/100). Next, to extended the sample size. To do this I analyzed S/100 on Brett Favre. Favre has led all QBs in passing attempts and has had a long and illustrious career. Plus, I thought if I analyzed one QB and to make the QB performance independent as possible, the dtata would be clearer. So I analyzed how S/100 (again, sacks per 100 passing attempts) affects YPA, completion percentage, passer rating, and TD/INT ratio. Here's what I found:

After plotting out these numbers, I decided to find the R2 or the correlation. For those of you unaware of correlation, +1 means a strong positive correlation, -1 means a strong negative correlation, and 0 means no correlation. The R2 between S/100 and YPA is .002; between S/100 and Comp% is .0008; between S/100 and Passer Rating is .018; and between S/100 and TD/INT ratio is .018. These numbers are so small, that is laments terms, by looking at Favre's numbers, there is no correlation between S/100 and a QB's performance.

Disappointed, I decided to briefly look at Jay Cutler's short career. These numbers were drastically different. Again, there is very little correlation between S/100 and YPA and Passer Rating; however, there is a very strong correlation between S/100 and Comp% and Passer Rating. The R2 for Comp% is very strong for Cutler, being .453 and for TD/INT ratio being .225. There is a strong correlation for Comp% and a slightly strong correlation for TD/INT ratio.

Then I thought to myself, maybe looking at one QB wasn't the best idea. I want to see how sacks affect EVERY quarterback, not just one. So based on the Cutler and Favre numbers, I decided to look at every quarterback last year that had a minimum of 415 passing attempts and looked to see how S/100 affects YPA, Comp%, Passer Rating, and TD/INT ratio. Last year there were 21 QBs that fit this bill ranging from the very good Peyton Manning and Phillips Rivers to the crappy Marc Bulger and Kerry Collins. Here's what I found:

Again, I was disappointed. The R2 between S/100 and YPA is .125; between S/100 and Comp% is .049; between S/100 and Passer Rating is .162; and S/100 and TD/INT ratio is .172. The correlation is slightly stronger when analyzing all QBs as opposed to one, but the correlation is still fairly weak and essentially non-existent.

However, this research is not a total waste of time. Yes, overall, how often a QB gets sacked bears no relevance on how he performs. But that doesn't hold true on every individual QB. Some QBs, like Favre, it does not matter, and even on guys like Aaron Rodgers who gets sacked like crazy. However, on certain guys like Cutler, S/100 does not matter. And sometimes S/100 can show you how crappy certain guys are. Last year, Kerry Collins (min 415 attempts) had the 2nd least S/100, yet put up mediocre to awful numbers. So as a whole, S/100 shows no relevance, but on individual QBs, it may make a difference.

Now, I would love to be proven wrong. If someone finds a major flaw in my analysis or my work, please let me know. If someone had proven data to show how sacks do affect a QBs performance, I'd love to know about it, but until then, I don't think so.

To read the full post and see all graphs, go to:
http://gameofinches.blogspot.com/2009/11/statistical-correlation-between...

Posted by: kaptrap19 on 27 Nov 2009

5 replies , Last at 24 Feb 2010, 3:16pm by Spielman

1
Re: Sacks Have No Affect On A QBs Performance
by Pitt-skee (not verified) :: Thu, 12/31/2009 - 1:53am

Are you serious... have you watched Jason Campbell this season? How about watching games? The pressure causes the int and the fumble. You don't need a math equation to figure out that which is as obvious as it comes.

5
Re: Sacks Have No Affect On A QBs Performance
by Spielman :: Wed, 02/24/2010 - 3:16pm

Pressure can cause an INT or a fumble, but fumbles aren't being taken into consideration as a measure of performance here. And sacks and the passing numbers are discrete... by definition, any play that generates passing stats wasn't a sack.

The suggestion here isn't that pressure doesn't screw things up, or that it isn't a good thing to sack the QB. (There's a strong correlation between QB fumbles and sack rates, believe me.) It's that there maybe isn't the kind of carryover effect that conventional wisdom might have you believe, where getting sacks causes mistakes on other plays.

Although, while I'm guessing that raw sack numbers would correlate well with total QB pressures, it wouldn't be perfect, and by using sacks as a shorthand for pressure, you add another level of uncertainty that likely drops the correlation values.

2
Re: Sacks Have No Affect On A QBs Performance
by Bowl Game Anomaly :: Sun, 01/10/2010 - 1:50pm

The problem with Pitt-skee's comment is that sacks are not the same thing as pressure. Some QBs are often under a lot of pressure but take few sacks (e.g. Joey Harrington), while some QBs take tons of sacks even when they aren't under an abnormal amount of pressure (e.g. Rob Johnson).

I think kaptrap19's research is very interesting- basically what it shows is that taking a lot of sacks does not necessarily mean you can't succeed, and taking a low number of sacks does not necessarily mean you will preform well. But that doesn't mean that pressure is irrelevant. Pressure can hurt performance by causing a lot of incompletions rather than sacks, or by making hits which wear down the QB even though he gets a pass off each time.

If you ran a correlation of sacks+hurries+hits against those other stats, you'd probably find a meaningful relationship. But the QB doesn't control the total amount of pressure (unless he's David Carr), while he does have more control over sacks specifically. (See the Pro Football Reference blog posts about the consistency of QB sacks: http://www.pro-football-reference.com/blog/?p=4152 and http://www.pro-football-reference.com/blog/?p=4395)

So basically kaptrap19's research shows that just because a QB directs pressure into sacks rather than hits and hurries, that doesn't mean he can't be successful, and conversely that if a QB converts pressure into hits and hurries instead of sacks that it won't necessarily make him any more successful.

3
Re: Sacks Have No Affect On A QBs Performance
by tuluse :: Mon, 01/11/2010 - 4:42am

See FO's research on Drew Brees that shows that QBs with low sacks rates who go into a situation with more pressure throw more interceptions.

4
Re: Sacks Have No Affect On A QBs Performance
by kaptrap19 :: Tue, 01/19/2010 - 12:30am

Pitt-skee, my reaction after seeing my results was very similar to yours. I mean how can it be that the correlation between sacks and a QBs performance essentially not be correlated? But I like Anomoly's idea- I hadn't added "pressure" into my equation. When I end up with free time I'll do that and I hope I can find a stronger correlation.

And Tulusa, where can I find this FO article about Drew Bress, it sounds interesting and I'd like to read it

And to the Jason Campbell argument, I counter that and show you Aaron Rodgers. I don't watch too many Redskins games in Chicago but I think Campbell just sucks.

Login or register to post comments