Writers of Pro Football Prospectus 2008

Most Recent FO Features

HarvinPer09.jpg

» Impact of the NFL's Kickoff Rule Change

After three NFL seasons of kicking off from the 35-yard line, what has been the impact on touchbacks, returns, field position, scoring and injuries? Also, is this rule responsible for a record number of big comebacks?

16 Sep 2011

Under Pressure: Splitting Seconds

by J.J. Cooper

When it comes to sacks, quarterbacks often get a free pass. Unless you're Ben Roethlisberger, few quarterbacks ever get blamed for their tendency to hold the ball too long -- the offensive line takes the heat.

Just ask Jay Cutler. Admittedly the Bears' quarterback has had plenty of problems with a leaky offensive line, but he also has created many of his own problems by holding the ball too long. In the past two years, more than half of Cutler's 82 sacks came when he held the ball for more than three seconds. His percentage of "long sacks" is sixth-worst in the league among quarterbacks with 200 or more attempts. It's not just offensive coordinator Mike Martz's protection concepts and poor offensive tackles that have given the Bears pass protection problems.

Roethlisberger, Michael Vick, Colt McCoy and Joe Flacco join Cutler in picking up plenty of sacks that really shouldn't be blamed on the offensive line. Every week during the season, Under Pressure will take a look at sacks around the NFL. Since 2009, I've been timing each and every regular-season sack in the NFL from the time of the snap to the time of initial contact on the sack. By doing that, we can find which quarterbacks are getting sacked before they get a chance to make their reads, and which ones are holding on to the ball too long.

So how do we determine a quick sack versus a long sack? Over that two-year-stretch, the average time of all sacks is three seconds, but a more accurate representation may be the median time, which currently stands at 2.8 seconds. After all, a couple of extremely lengthy sacks can adjust the average time up, but there is a pretty fixed limit on the lower end of how quickly a sack can happen -- there are 64 sacks over the two years that took six or more seconds. There are only two sacks that were recorded in less than one second.

The median sack time has hovered between 2.7 and 2.8 seconds throughout the two years of the tracking.

Unless a quarterback falls down or drops the ball, a sack of 1.7 seconds or less generally involves a rusher coming completely free. On the other end of the spectrum, once a quarterback holds the ball for three seconds or longer, he’s had time to make his drop, survey the field, and work through his progressions. If it takes longer than that, either the play has broken down or the quarterback is trying to buy time because the play has not developed as he had expected it to.

There are limitations involved with just timing the sacks. Ideally it would be better to time each and every pass attempt for every quarterback, but the amount of turn-around time makes that impossible to do on a weekly basis. Doing this would give a better sense of the benefits or downsides of quarterbacks holding onto the ball. And it would give a better sense of how much holding the ball longer affects a quarterback’s sack rate.

While it has not been possible to log each and every pass attempt around the league, over the past offseason it was possible to log the time of pass for all of Steelers pass attempts in 2010. Since Roethlisberger is considered to be prime example of a quarterback who creates big plays by holding the ball and escaping sacks, his statistics are useful.

What was found was that in Roethlisberger’s case, his sack percentage (the percentage of pass plays that results in sacks) goes up significantly the longer he holds the ball. Roethlisberger’s sack percentage doubled when he held the ball for longer than three seconds -- seven percent for passes of 2.1 to 3.0 seconds, 14 percent for passes that took between 3.1 and 4.0 seconds. Interestingly, his quarterback rating, yards per attempt and first down percentage all dipped when he held the ball for more than three seconds, so it’s difficult to say that holding the ball longer really pays off for Roethlisberger.

On the other end of the spectrum, when the Colts lost Peyton Manning to a neck injury, they not only lost one of the best quarterbacks in the league, but they also lost a quarterback who could make their offensive line's potential weaknesses seem to disappear. Over the past two years, Manning has been sacked only 26 times. When you look at sacks that are created by a quarterback holding the ball too long, however, Manning is even more exceptional. No other quarterback can top Manning’s four sacks of three-plus seconds in 1,250 pass attempts (0.32 percent of attempts).

Percentage Pass Attempts Resulting in Sacks of Three or More Seconds, 2009-2010
Quarterback Pass Att Pct Quarterback Pass Att Pct Quarterback Pass Att Pct
Josh Johnson 150 6.00% Donovan McNabb 944 3.07% Matthew Stafford 482 1.87%
Michael Vick 393 5.34% Seneca Wallace 228 3.07% Chad Henne 958 1.77%
JaMarcus Russell 259 5.02% Kevin Kolb 294 3.06% Matt Ryan 1040 1.73%
Ben Roethlisberger 941 4.89% Brett Favre 917 3.05% Jake Delhomme 478 1.67%
Daunte Culpepper 165 4.85% Mark Sanchez 898 3.01% Keith Null 121 1.65%
Colt McCoy 233 4.72% Kyle Orton 1070 2.90% Marc Bulger 251 1.59%
Joe Flacco 1032 4.26% Drew Stanton 175 2.86% Carson Palmer 1069 1.59%
Jay Cutler 1030 4.17% Josh Freeman 786 2.80% Matt Hasselback 947 1.58%
Bruce Gradkowski 320 4.06% Derek Anderson 523 2.68% Tom Brady 1074 1.58%
Troy Smith 151 3.97% Ryan Fitzpatrick 686 2.62% Jon Kitna 323 1.55%
Charlie Whitehurst 103 3.88% Alex Smith 732 2.46% Kurt Warner 520 1.35%
Kyle Boller 187 3.74% Matt Moore 288 2.43% Drew Brees 1187 1.26%
Trent Edwards 241 3.73% Philip Rivers 1052 2.38% Eli Manning 1061 1.23%
Jason Campbell 868 3.69% Sam Bradford 604 2.32% Shaun Hill 578 1.21%
Vince Young 430 3.49% Brady Quinn 262 2.29% Byron Leftwich 115 0.87%
Aaron Rodgers 1051 3.33% Jimmy Clausen 306 2.29% Kerry Collins 498 0.80%
Chris Redman 123 3.25% Rex Grossman 136 2.21% John Skelton 127 0.79%
David Garrard 911 3.18% Matt Schaub 1183 2.20% Peyton Manning 1254 0.32%
Matt Cassel 973 3.08% Tony Romo 780 2.18% League Average 34912 2.61%
Minimum 100 Pass Attempts (passes + sacks)

But if there is any good news for Colts’ fans, it’s that Indianapolis has replaced Manning with another quarterback who also knows how to get rid of the ball. Of quarterbacks with 200 or more attempts in the past two years, Kerry Collins is second in the NFL in sacks that took three or more seconds (0.80 percent of attempts). Collins may not have Manning’s passing ability, but he should be able to stay out of needless sacks.

With the background information out of the way, on to the first week of the 2011 season.

WORST SACK OF THE WEEK

No matter what Falcons quarterback Matt Ryan does, there will always be some Falcons fans who blame him for not being Vick. Ryan didn’t help himself on Sunday with a reminder that, as we all know, he doesn’t have Vick’s athleticism.

Vick is one of the best in the league at spinning away from pressure. Sensing backside pressure, he'll dip a shoulder, spin around and slip under a charging defensive end.

Matt Ryan cannot effectively execute the spin move. On a third down late in the third quarter of the Falcons’ loss to the Bears, Ryan tried to do the same thing. But with Julius Peppers bearing down on him from the blind side, Ryan’s attempt to spin away turned into a disaster. He stumbled in the middle of his spin, then used his right hand (the one holding the football) to try to steady himself. He did manage to stay upright, but he left the football on the ground. The Bears scooped it up for an easy touchdown that ended the slim hopes that Atlanta had of a comeback.

QUICKEST SACK OF THE WEEK

Eagles' defensive ends Jason Babin and Trent Cole gave the Rams tackles trouble all day, but when Babin sacked Sam Bradford early in the second quarter, one has to hope that right tackle Jason Smith either forgot or couldn't hear the snap count. That's the only charitable explanation as Babin was across the line of scrimmage and had a step on Smith before Smith even came out of his crouch. Smith never laid a hand on Babin although he did feebly try to chase after him. In his defense, he was in perfect position to help Bradford back up once the play was over.

Babin picked up the sack in only 1.7 seconds. Only one of the 90 other sacks this week was timed at less than two seconds.

LONGEST SACK OF THE WEEK

There were four sacks which took six seconds or longer. Not surprisingly, three of them came on bootlegs. Nothing buys a quarterback more time, but also limits a quarterbacks options like a bootleg. And nothing gives the quarterback a visual reminder of the clock ticking in his head like a bootleg -- when he reaches the sideline, he's either throwing the ball, cutting upfield to run, or stepping out of bounds. Ryan's 6.2-second sack was the longest of the weekend, but Tony Romo's 6.0-second sack is the one that may be most remembered: it came on the play where he attempted to dive for the end zone from the Jets 3-yard line. He lost the ball before he landed, didn't make it to the end zone and gave the Jets a second chance.

HOLDING THE BALL

No one could find much to complain about in Cam Newton's pro debut, but if the Panthers' quarterback had any weakness in the opener against the Cardinals, it was his tendency to hold the ball. All of Newton's four sacks came when he held the ball for three or more seconds. Romo doesn't have youth as an excuse, but three of his sacks also came when he held the ball for more than three seconds.

Posted by: J.J. Cooper on 16 Sep 2011

60 comments, Last at 23 Sep 2011, 7:53pm by JPS

Comments

1
by Eddo :: Fri, 09/16/2011 - 1:57pm

Great stuff, I am going to enjoy this feature!

One thing that seems interesting, in addition to the timing, is the yardage lost on sacks. The most frustrating, as a fan, are when a QB keeps sliding back in the pocket, and winds up losing seven or eight yards when he could have stepped up, and only lost two or three. Any plans on incorporating yardage, as well?

15
by J.J. Cooper :: Fri, 09/16/2011 - 3:15pm

Thanks Eddo, lost yardage could be the basis for a later Under Pressure. I have the info logged, so I'll work on pulling that all together.

2
by Anonymous(not that one) (not verified) :: Fri, 09/16/2011 - 2:08pm

Does the 3 second sack stat count against QBs who move away from a rusher, but are eventually sacked?
Does the sack stat used here include plays where after 3 seconds the QB throws the ball away?

14
by J.J. Cooper :: Fri, 09/16/2011 - 3:13pm

The time is from the snap to the initial contact with the quarterback. So if a quarterback rolls away from pressure at 2.0 seconds, but doesn't get pulled down til 3.5 seconds, it gets counted as a 3.5 second sack. That's the only way I could determine to make it consistent, determining when a quarterback first starts escaping pressure would make it a judgement call. Instead I can based it on a fixed point (the snap) and another fixed point (when the defender makes contact). I should mention that I run each play back at least three times to ensure I get a consistent time.

27
by 0tarin :: Fri, 09/16/2011 - 4:18pm

This touches on something I was curious about as well regarding these statistics. Flacco is obviously gaining a reputation for his Roethlisberger-esqe ability to hold the ball far too long as well, but in my observation (most notably last season), he holds the ball forever simply because by the time he finishes his drop, the line has let pressure come through and he must either bootleg or start dancing. Since this would limit his ability to progress through reads effectively (as well as lead to more sacks), is this factor going to be discussed at all in the future? I realize it was beyond the scope of this article itself, but it seemed like something that could be refined into an interesting note, particularly if Roethlisberger can be observed holding the ball for similar reasons. With the Steelers' notoriously flimsy O-line, that seems likely.

41
by Intropy :: Fri, 09/16/2011 - 7:01pm

If the QB stiff arms a rusher at 2 seconds, rolls to his left and gets hit by a second rusher at 4 seconds, which would that be?

47
by J.J. Cooper :: Fri, 09/16/2011 - 9:52pm

In that case a four second sack. If he stiff armed at 2 seconds, was pulled down at 3 seconds but was never really free of the rusher in the interim, it would count as two seconds. Contact that leads to a sack triggers a stoppage of the stopwatch, contact that doesn't itself lead to a sack does not. Admittedly, that's a little bit of a judgement call, but again, I'm trying to make it as consistent as possible.

51
by Intropy :: Fri, 09/16/2011 - 10:39pm

Thanks. May I propose that you stop the stopwatch on the first contact instead? You make a very good point about wanting to be consistent and it being a difficult judgment call to determine when a QB begins to roll away from pressure. But I think physical contact is pretty concrete, in fact that's what you use already, and based on your statement that determining whether contact leads to a sack is a bit of a judgment, it may actually be more straightforward and consistent to stop the watch on any contact. A quarterback who is contacted by a defender has pretty definitively been pressured, and it makes sense to judge that the line has allowed pressure by that point. I obviously don't have figures, but I think that may make some difference for bigger, stronger quarterbacks like Cutler, Vick, Roethlisberger, and Freeman.

57
by RichC (not verified) :: Mon, 09/19/2011 - 10:39am

I agree with Intropy here.

It also doesn't make a whole lot of sense to only time plays where the guy gets sacked.

3
by Jerry :: Fri, 09/16/2011 - 2:12pm

If alt-Roethlisberger always threw the ball away incomplete at three seconds, how would his stats compare to real Ben's? (His entertainment value would certainly be less.) Also, if you can arrange with Aaron to marry the time data to the DVOA spreadsheet, it would be interesting to see the DVOA splits.

30
by fyo :: Fri, 09/16/2011 - 4:28pm

If alt-Roethlisberger always threw the ball away incomplete at three seconds, how would his stats compare to real Ben's?

This is a critical point!

Quarterbacks don't hold on to the ball for the hell of it.

Unless one wants to argue that quarterbacks intentionally decide NOT to throw the ball to open receivers early, the only relevant issue is if they would be better off throwing the ball away.

Hindsight is 20/20, so you have to compare the 3+ second efforts with the alternative of throwing every single one of those balls incomplete.

55
by MJK :: Sun, 09/18/2011 - 12:08pm

Good point, but I wouldn't say that this is the only relevant issue.

You're arguing that you can't use these stats to say whether a QB should hold the ball or not, and I agree...without knowing how successful a QB is when he holds the ball, the analysis becomes murky.

However, this is useful for one very important thing...knowing when a QB's sack-proned-ness (or lack thereof) is due to bad Oline play or not.

Basically, the O-line's job is to protect the QB for 3 seconds. If a sack comes after the 3 second mark, it's not the O-line's fault.

I would be interested to see a table of what percentage of a QB's sacks came split by time. Not what percentage of total pass attempts ended in long sacks.

56
by Jerry :: Mon, 09/19/2011 - 2:13am

As many of the posts here illustrate, there are several relevant issues. As we accumulate more data, we may be able to address some of them.

I asked about Roethlisberger's numbers because (1) he's the guy who J.J. has all the times for, and (2) he seems more likely to make a play after holding onto the ball than most QBs. My impression is that the Steelers are willing to accept some of those extra sacks in exchange for some of the plays that Ben comes up with, and i wonder how that compares to throwing the ball away.

I'll also mention here that I thought about all this while watching Ben hang onto the ball during the first couple of Steeler drives against Seattle. It looked like he was reasonably successful on those plays; later in the game, he was throwing more quick passes.

58
by RichC (not verified) :: Mon, 09/19/2011 - 10:47am

"However, this is useful for one very important thing...knowing when a QB's sack-proned-ness (or lack thereof) is due to bad Oline play or not."

Its not really even good for that though. How many times does Tom Brady hold onto the ball for 7 seconds only to find that Welker finally got open? One could argue that if he got sacked there its because of his own problems for holding onto the ball too long.

All these numbers are really telling you is how likely a guy is to get hit when he holds onto the ball.

The average sack is 2.7 seconds, but that doesn't mean that holding onto the ball longer than 2.7 means a sack is your fault.

4
by Ben Muth :: Fri, 09/16/2011 - 2:12pm

Great article. Looking forward to this every week.

23
by Dr. Mooch :: Fri, 09/16/2011 - 4:08pm

Now when one of your teams gets singled out here you're almost going to be obligated to comment on it in your column as well.

42
by Intropy :: Fri, 09/16/2011 - 7:02pm

Really good point. I look forward to seeing the same play analyzed the heck out of from both perspectives.

5
by Will Allen :: Fri, 09/16/2011 - 2:18pm

Thanks for the effort. There is a long way to go, but without this sort of work, it impossible to start to get good metrics for pass protection. The fact that the turnstiles in Indianapolis have yielded so few sacks in recent years gives some glimpse as to how badly sacks can mislead.

6
by Jim Z. (not verified) :: Fri, 09/16/2011 - 2:19pm

I always find it humourous that Philadelphia Eagles fans constantly complain about the offensive line's inability to protect Vick, when most of the sacks and hits that accumulate are a direct result of Vick's propensity to try to "extend" the play or his inability to find the hot receiver on a quick throw.

16
by J.J. Cooper :: Fri, 09/16/2011 - 3:16pm

If you notice, many of the quarterbacks with a high number of long sacks are mobile quarterbacks. The ability to extend the play with scrambling seems to bring more sacks with it.

21
by Aaron Brooks Go... :: Fri, 09/16/2011 - 4:00pm

The question is correlation or causation. Would those sacks have been quick sacks on a fence post like a Drew Bledsoe?

25
by J.J. Cooper :: Fri, 09/16/2011 - 4:08pm

Possibly, but just as possible the less mobile quarterback throws to the hot route or simply gets rid of the ball instead of trying to buy time for a bigger play.

46
by eagler7 :: Fri, 09/16/2011 - 8:03pm

Being an Eagles fan, a large number of those 3+ second sacks tend to be the offensive line not protecting Vick and him being flushed out of the pocket as a result, making sacks that would have been less than 3 seconds into 3+ second sacks. I will say that I think the offensive line situation is overblown by the national media and not as much eagles fans. While these 3+ second sacks aren't as accurate for a qb like Vick compared to a pocket passer, a lot of sacks also are still because of Vick holding onto the ball too long, so the blame should be more of a combination of the o-line and Vick rather than one or the other.

7
by Aaron Brooks Go... :: Fri, 09/16/2011 - 2:21pm

"Tony Romo's 6.0-second sack is the one that may be most remembered: it came on the play where he attempted to dive for the end zone from the Jets 3-yard line. He lost the ball before he landed, didn't make it to the end zone and gave the Jets a second chance."

He fumbled while across the line of scrimmage. How was that a sack?

8
by Aaron Schatz :: Fri, 09/16/2011 - 2:24pm

Actually, the line of scrimmage was the 2 and he was tackled at the 3 for a one-yard loss.

19
by Travis :: Fri, 09/16/2011 - 3:40pm

The official scoring has the tackle and the recovery at the 3, but the recovery took place somewhere between the 1 and the 2 (4:15 in the video). Weird.

24
by Aaron Brooks Go... :: Fri, 09/16/2011 - 4:08pm

Even neglecting that the fumble occurred, at worst, at the line of scrimmage (http://healthyinfluence.com/wordpress/wp-content/uploads/2011/09/Romo-Fu...), by that point Romo had committed himself to the run and become a runner. A QB can run for a negative gain without incurring a sack. To say Romo fumbled as a result of a sack is to define to term "sack" so loosely that it loses all meaning and utility as a point of analysis.

Poor played, good sir.

26
by Travis :: Fri, 09/16/2011 - 4:15pm

Even if the quarterback commits himself to run, it's a sack as long as (1) he originally dropped back with the intention to pass and (2) didn't gain yardage on the play. That's just the way the scoring rules work.

28
by J.J. Cooper :: Fri, 09/16/2011 - 4:22pm

Do remember the fumble could have occurred a half yard past the line of scrimmage and still be judged a sack--you have to gain a full yard for it to not be a sack. As far as the fact it was ruled a sack, take that up with Elias. The most consistent rule I can apply is if the NFL calls it a sack, I call it a sack.

9
by Southern Philly :: Fri, 09/16/2011 - 2:32pm

"one has to hope that right tackle Jason Smith either forgot or couldn't hear the snap count."

It was worse than that, IIRC he double teamed the DT at first, which is why Babin went right by him.

10
by alien1rock :: Fri, 09/16/2011 - 2:37pm

Really good stuff, but a question about the numbers. Rex Grossman has 2.21% of sacks taken that were longer than 3 seconds. That is basically just during the 4 games with the Skins last year, and NFL.com has him with 9 sacks during that period. So he held the ball longer than 3 seconds on 0.18 of his 9 sacks? What am I missing?

11
by Anonymous221 (not verified) :: Fri, 09/16/2011 - 2:42pm

looks like the denominator is # of attempts, not # of sacks

12
by alien1rock :: Fri, 09/16/2011 - 2:44pm

That makes a ton more sense. Thanks

13
by drobviousso :: Fri, 09/16/2011 - 3:11pm

Very minor quibble: put the league average in place in the chart. Makes it easy to see who is above and below average.

Big thanks for thinking about the median though. With one side of the distribution bounded and the other side unbounded, the median is the right average to use.

17
by JasonG (not verified) :: Fri, 09/16/2011 - 3:18pm

Are you evaluating receiving corps' roles in long sacks? Welker gets open in an instant and Brady gets him the ball immediately. Rinse repeat. Peyton has multiple guys able to, you know, get open. Cutler, of course, bears some blame for holding the ball, but I doubt he deserves all the blame. He seems to be waiting on his receivers nearly every play (and then the line collapses from all sides and he's SOL). Knox and Hester can outrun guys sure, but they aren't beating their men off the line. Basically, Cutler HAS to wait for them because they aren't getting open. Throw it away, dump it off, yes, again he deserves some blame, but more sacks are going to happen if your receivers are mediocre or worse.

Note: My remarks are based mostly on the last two years. WRs and protection were awful. Based on all of one game, everyone seems better this year (the bar was set really low). Seriously though, Cutler seems more in command of the offense (where to go with the ball), the WRs more comfortable (where to be, if not improved in beating coverage), too. Better pass blocking (including TEs) giving just enough more time (though still not a ton). Better use of dump offs. This is all to say, I would expect Jay to be making decisions just a tad quicker and the line to give him just a tad more time and that combo should significantly reduce his 3+ second sack stats this year. I guess we shall see.

31
by J.J. Cooper :: Fri, 09/16/2011 - 4:33pm

Sorry JasonG, there is no way I can figure out to include them. Even if we had all-22 film on every play (and we don't), how would you quantify whether someone was open enough for the quarterback to throw the ball? I'm not saying that doesn't play a part in long sacks, I'm just saying there is no way to quantify how much of a role it plays.

33
by justanothersteve :: Fri, 09/16/2011 - 5:12pm

It's easy to blame the receivers when there is no way for us to judge. However, I did see a lot of Martz's offense with the Rams. (I live in St Louis.) The "Greatest Show on Turf" used a lot of timing routes. Warner would throw to the point he expected his receivers to be open and they were expected to catch the ball almost as soon as they turned for the ball. The ball was often thrown before the receiver even made his break. I assume the Bears offense is designed much the same way. If so, Cutler doesn't have faith in his wideouts to make the throw and assume they will catch it. So the problem falls on Cutler.

34
by Southern Philly :: Fri, 09/16/2011 - 5:14pm

You're ignoring the enormous talent gap between those Rams receivers and the Bears receivers.

38
by justanothersteve :: Fri, 09/16/2011 - 6:27pm

I am not ignoring it. I do agree that there is a talent gap between the two teams. That doesn't change the point that Cutler is not operating the offense as intended.

39
by Jonadan :: Fri, 09/16/2011 - 6:54pm

"Cutler is not operating the offense as intended," may be an accurate observation, but it still fails to place the blame.

For instance:

- Is it possible to operate "as intended" with the Bears' personnel? If not, the blame falls on Martz, unless the system has enough give in it to work even under non-ideal conditions.
- If it is possible, whose fault is it if it's not happening? If the Chi-town receiving corps is good enough/fast enough to get separation, but isn't actually (underperforming), Cutler may be forced (by his own knowledge) to hold the ball... and worse, that kind of thing can get circular fast (WR runs sloppy route -> QB holds the ball -> WR runs a better route -> QB still holds the ball because he "knows" the WR won't be there -> WR adjusts to the QB instead of the playcall -> etc.), to say nothing of any coaching/playcalling responsibilities there might be.

I'm guessing it's a combination of factors, personally.

---
"When you absolutely don't know what to do any more, then it's time to panic." - Johann van der Wiel

43
by JasonG (not verified) :: Fri, 09/16/2011 - 7:16pm

Warner was definitely special in making those throws and no matter the parts around him, I doubt Cutler would ever or could ever do it the way Warner did. However, as DA says the WRs were vastly superior and oh yeah so was the protection. Forte is Faulk-like, but obviously inferior as well. But Warner vs Cutler is besides the point. How do we know Warner didn't take a bunch of long sacks in that offense? It could be Cutler, could be the line, could be the receivers and it could be the offense/play-calling (constant 7-step drops). This is why I posed my question in the first place. Wins are a team stat unwisely attributed to QBs. Football is the ultimate in team sports, 22 orchestrated men at a time. This metric feels similar, trying to attribute something to a QB that really should be attributed to a team's passing game as a whole.

FO tries to separate line play from sacks and line play from RB metrics as well. All I was asking was if we can separate sacks from receiver play/skill? This analysis has it's value, but whatever percentage ends up next to a QB's name on this list, can we say it defines his holding-on-to-it-too-longedness?

As the writer mentioned, charting all release times would add value to the analysis. Throwaways would be great to add to this as well. Both would let us know how quickly guys are making decisions and how quickly they bail on plays that aren't there.

I like the article, I do. And I know there are limits to what data can be attained. I'm just saying, and I doubt the author would disagree, there is more to know. Quick pass, 7-step drop or rollout? 2 receivers or 5? Open or not? Clean pocket or collapsed? Etc. It would take full game-tape review and charting to do this, but I think it would take that level of analysis before you can attribute certain strengths/weaknesses to particular players, in this case holding on to the ball to QBs.

Take Joyner for example. He analyzes tape for QB bad decisions regardless of the outcome (bad decisions may not be punished with an INT, good decisions could end up in an INT), just to analyze the QB's decision-making in and of itself.

If you went that far, then you could make statements like Ben just holds on to it too long, Flacco was indecisive/tentative, Rodgers extends plays all the time by rolling out, Cutler's WRs were never open and his center was constantly pushed into his face, Campbell doesn't know how to ID open receivers, whatever. We make those statements now based on basically just raw sack totals and watching the game. But if you had a breakdown like this, those statements could be affirmed or refuted with data. Just as FO ignores fumble recovery and just looks at fumbling, in this case, don't look at the result (the sack, quick or not) but why that sack happened or didn't happen. That's another piece of info missing. How about sack avoidance stats? How many sacks has Vick avoided by scrambling, Brees by throwing it away, Ben by shedding tacklers? That's great info, too. Maybe you live with 10 extra long sacks because they avoided 30 would-be quicker sacks.

Anyway ... good article.

45
by Intropy :: Fri, 09/16/2011 - 7:59pm

Your points are all sensible. However I saw the article in a fundamentally different way. You're trying to use it to measure QB-holding-it-too-long-ness. I read these figures and I see short sacks except in pretty unusual cases, such as a fumbled snap, are the lines fault. Long sacks are not likely to be the line's fault. Granted there is some window dressing in the article about QB being at fault. But I think the real main thrust is evaluative of the line play rather than the quarterback play.

18
by DGL :: Fri, 09/16/2011 - 3:28pm

"Interestingly, (Roethlisberger's) quarterback rating, yards per attempt and first down percentage all dipped when he held the ball for more than three seconds, so it’s difficult to say that holding the ball longer really pays off for Roethlisberger."

More than interestingly; I think this is against the perception of pretty much everyone that watches the Steelers. My guess is it's a case of selective memory - the few cases where something good happens outweigh those where it doesn't. It'd be interesting to see all the numbers behind this.

20
by drobviousso :: Fri, 09/16/2011 - 3:45pm

Early in his career, one of the networks had a graphic they would show every time he got hit then made a completion. The graphic showed his QB rating on plays he got hit and plays he didn't. It was like 15 points higher on plays he got hit on.

Haven't seen that graphic in a few years. I also haven't seen him get hit, run around, and launch a long bomb nearly as often. I don't know if this is due to more diligent tackling, different coverage, or if some of his early agility is gone die to age / getting the crap kicked out of him.

59
by RichC (not verified) :: Mon, 09/19/2011 - 10:55am

The fact that his QBRating is lower than normal after holding the ball 3 seconds doesn't mean its lower than if he'd thrown all those balls away.

22
by qed :: Fri, 09/16/2011 - 4:01pm

Excellent research. Could you include each quarterback's raw sack percentage in the table? It would be interesting to see how the "long holders" come out overall in terms of sack rate.

29
by zenbitz :: Fri, 09/16/2011 - 4:26pm

Hey, Hey, something Alex Smith is above average at!

32
by Anon (not verified) :: Fri, 09/16/2011 - 4:58pm

This is a great basis for research, but a lot still needs to be done. After all, it seems hard to believe that holding onto the ball is the quarterback's fault, or at least not wholly.

Quarterbacks don't throw the ball arbitrarily; A QB throws when he sees an open receiver, or when he's intentionally throwing it away to avoid a sack. Receivers that get open faster will help bring sack rates down by allowing QB to throw the ball earlier. Also, scheme greatly effects when those receivers are open: a team that runs the West Coast offense has shorter routes, but teams that run a Mike Martz offense have longer routes, and the QB can't throw the ball until the routes develop properly.

That said, the QB isn't blameless. A QB's perception of "open", release speed, route progression, and mobility in the pocket (both in the Brees-"avoiding pressure" and Vick-"scrambling" senses) all factor in. What we've measured here is all of these factors together, and it's going to take a lot of work to try and parse out who is responsible for what.

36
by Jerry :: Fri, 09/16/2011 - 5:29pm

Everything you said is true, but let's start by thanking J.J. for bringing us some data that we haven't had. After we see it for a while, we may be able to discern (or just guess at) some cause and effect, but the first step is compiling the data.

35
by nibiyabi :: Fri, 09/16/2011 - 5:29pm

Fascinating stuff, but I'm curious about a few potential confounding variables / overlooked assumptions.

I would be pretty shocked if any QB logged a significant number of pass attempts over the course of a season and ended up with a positive correlation between performance and elapsed time before pass. By definition, if a QB takes a long time to get rid of the ball, the play did not develop as planned -- either his primary receiver is covered, someone ran a sloppy route, someone missed a blocking assignment, etc.

So of course Roethlisberger performed worse when he held the ball longer -- what I'd be interested in seeing is how this dip in performance measures up to the rest of the league (which, of course, would require a huge undertaking in terms of volunteer game charting).

Using a baseball analogy, if a hitter has a .270 EQA given a 0-0 count and a .220 EQA given a 0-2 count, you can't simply say "he should swing earlier to avoid 0-2 counts, as he clearly suffers from a huge drop in performance when he reaches 0-2". What if the average hitter earns, respectively, a .270 EQA and a .180 EQA in the same scenarios? This would indicate that while our hero is an average hitter, he excels in two-strike scenarios compared to the rest of the league.

Of course, comparing to other players isn't sufficient either -- even if Ben suffers the smallest performance drop in the league in terms of quick passes vs. slow passes, it still might make sense to replace some of those sacks with a few checkdowns and dumpoffs. And you also have to consider context -- with 0:04 to play in the game and his team down 6 at midfield, he should be willing to risk a much higher chance of taking a sack to even remotely increase his chances of completing a long pass.

All that being said, this is the first time anyone has recorded this type of thing, and I'm intrigued to see how it will evolve and improve as more people get involved and the dataset becomes larger. Keep up the good work -- I'm excited to see more.

40
by Intropy :: Fri, 09/16/2011 - 6:57pm

I think this is really insightful. I agree it would take some doing, but FO already has game charters. Would recording time at throw be practical? That would be a wealth of data for JJ and others to mine and could yield really fascinating results.

37
by NMS (not verified) :: Fri, 09/16/2011 - 6:13pm

It seems like alot of the league's more athletic QBs are in the far left column and the league's slowest QBs are in the right column.

This could mean that the more athletic guys believe in their legs too much and hold the ball and dance around too long, thinking they can always out run a pass rusher, or it could be because (as another poster suggested) the quicker guys simply manage to run away for a second or two before going down instead of going down immediately like a slow QB would

44
by dbt :: Fri, 09/16/2011 - 7:59pm

One thing to be wary of when drawing conclusions about the data is selection bias, especially in the "mobile" vs "immobile" qbs. An immobile QB who can't make a hot read and get the ball out quickly for at least some portion of his career simply doesn't show up on any list, because he doesn't have NFL talent.

This is some great data. Looking forward to further analysis. One thing I would like to see is the full distribution of time/sack numbers to see if there are obvious inflection points other than at 3.0 seconds.

48
by EasyLikeSundayMorning (not verified) :: Fri, 09/16/2011 - 9:55pm

JJ, great work. While many of the suggestions would take more time for you, they really show how much interest there is in the topic.

One thing I'd be interested in is how the 3+ second sacks compare to the <3 second sacks for each QB. And with this data, there should be clusters of QBs, eg, few of either kind of sacks, lots of 3+ sacks but few <3 sacks, lots of <3 but few 3+, lots of both types. Without seeing the data, I'd guess that Peyton, Vick and Ben would fall into 1, 2 and 4.

49
by KK Probs (not verified) :: Fri, 09/16/2011 - 10:20pm

What about flipping this data and looking at the defense? That is, which defensive players got sacks the fastest? Perhaps specific defensive players, or schemes, are a more significant factor than QBs or OLs. It would be interesting to see if there are certain threshholds for defensive players. For example, are there only a few players who can ever get to the QB in less than 2.5 seconds, and the rest of the NFL needs more time than that to record a sack? If you're collecting the data anyway, maybe this would be an easy add-on.

This is innovative research with a lot of work put into it. Thank you for this.

50
by J.J. Cooper :: Fri, 09/16/2011 - 10:24pm

KK, thanks for the comment. I'm planning to do exactly that for an upcoming post. Just wait a couple of weeks (or maybe next week depending on how things come together).

52
by KK Probs (not verified) :: Fri, 09/16/2011 - 11:01pm

Very anecdotally, the top 7 on the list includes a natural outlier in Vick, arguably the greatest NFL running QB of the last few decades, three guys who were not playing at an NFL level during the years in the sample (Culpepper, Russell, and Johnson), and three players who are generally considered solid-to-good NFL QBs (Roethlisberger, Flacco, and McCoy). The three in the last group have in common that they all played in the AFC North, and their teams have faced 75% of the same defenses over the last two years. It looks like that may have something to do with the defenses on their schedule.

53
by NHPats (not verified) :: Sat, 09/17/2011 - 9:10am

I think this is great stuff, but as I picture Tom Brady standing behind the line last week, chatting on his cell phone for all the time he had on some plays, it just whets my appetite for more, especially for context.

As I think about this, I'd want to compare this with time-to-release data on all drop backs as well, and then cross tab that with results. What are the times for successful completions (success in DVOA terms)? What are the times for unsuccessful completions? For on-target incompletions or interceptions, and finally for true throw sways. How long until the qb begins scrambling and what was the outcome there? Obviously there's a bit of definitional gray area in a couple of those categories,

What can we learn about individual qb performance, about o-line performance, about defensive line play and secondary play? How much time does a d-line need to improve the probability of a negative outcome? is there such thing as a "coverage sack" or is it crappy route-running? What quarterbacks are more effective holding the ball, buying time for themselves, which aren't? How does that relate to overall performance measured by DYAR?

54
by Uncle Rico (not verified) :: Sun, 09/18/2011 - 8:54am

It's interesting. But I'm not sure how much value it is. It doesn't take into account the number of drop steps by the QB, from under center or gun, and it doesn't take into account routes and route progressions. It's based entirely on league median. What if that median works out that the QB is being sacked on just his 2nd progression? Seems then that 3 seconds would set the bar too low. Lot of different offenses out there. Many with mostly 3 and 5 step drops, and many with mostly 7 step drops with longer developing routes. The guy getting sacked on a 3 step drop is going down almost instantly. Those sacks would skew it down. I dunno. It doesn't seem like 3 seconds is a well reached benchmark. We know on 5 and 7 step drops there is no branch on the route tree sprouting shorter than 12 yards. We know from the combine, most WRs are running 1.5 to 1.8 10yard splits on their 40. That's of course sans pads, sans a defender jamming/rerouting, from sprinters blocks/stance, straight line, no breaks, no powerdowns, ect. Takes a QB nearly 2 seconds to get to the back of his 7 step drop. Obviously, that's coordinated with the route designs. Without timing receivers in their routes, seems just over 2 seconds is a generous guess for the shortest routes. If so, that means the QB has less than one second to cycle his progressions and get the ball out. I know presnap reads and his drop should help the QB eliminate one maybe two receivers from that cycle. But still, is that enough? Maybe it is. I don't know. But it just doesn't seem accurate/useful to say league median is ~3 seconds, so anything above that is on the QB.

60
by JPS (not verified) :: Fri, 09/23/2011 - 7:53pm

Here's an opportunity for "Charters II: Guys With Stopwatches" to record snap->throw times along with snap->sack contact times.