Writers of Pro Football Prospectus 2008

Most Recent FO Features

LockettTyl15.jpg

» Futures: Kansas State WR Tyler Lockett

The Wildcats receiver isn't the best athlete you'll ever see, but Matt Waldman says he could be an effective pro with small improvements in his technique.

08 Jun 2010

Defensive Drops 2009

by Bill Barnwell

In an article for ESPN Insider last year, I took a look at whether players with prominent reputations as big hitters -- e.g. Ed Reed -- actually induced drops at a unexpectedly-high rate, by using the data from our game charting project.

The answer was no. Reed hadn't seen a single dropped pass on plays where he was in coverage, while a highly-regarded tackler like Antoine Winfield only had a league-average drop rate of 5.9 percent. Perhaps more interestingly, I found that:

  • Passes against safeties and particularly linebackers were far more likely to be marked as the result of a drop than those passes targeted at cornerbacks
  • There was very little year-to-year consistency for individual players in their "Drop Rate"
  • There was inconclusive evidence regarding any idea that teams would be better or worse at forcing drops

We'll be covering the third point in a future article, but let's check and see if 2009 matched up with 2007 and 2008 with regards to the first two topics.

In 2009, safeties and linebackers absolutely yielded more drops than cornerbacks. Passes where those players were in coverage were dropped an even 7.0 percent of the time; passes at cornerbacks were dropped 5.2 percent of the time. I'm thinking the reason why that is has to do with several factors; passes charted with safeties in coverage tend to be longer than passes thrown at cornerbacks, which make them more difficult to catch; they also involve more open space, which may limit the number of times our charters mark a pass as "defensed" instead of "dropped". It could also be the case that passes thrown at those positions are to relatively inferior receivers that work in the slot or as the third or fourth option on a pass play.

As you might expect, the leaderboard for drops is dominated by players from those positions. Among players with 30 targeted passes or more, the leader in 2009 was the venomous, terrifying ... Mike Adams. The Cardinals safety was witness to nine dropped passes amongst his 46 targets, for a Drop Rate of 19.6 percent. That would have also been the best figure for safeties a year ago, although it would not have met DeMeco Ryans' league-leading 23.3 percent (seven drops on 30 targets). Below him was Chad Greenway (18.4 percent), Jordan Babineaux (17.5 percent), Nick Barnett (16.7 percent), and James Farrior (15.8 percent). Farrior also ranked among the league leaders last year, as his 18.6 percent total was second behind Ryans.

The other guys weren't so hot in this category a year ago. Adams saw one drop in 15 targets. Greenway was at 2-for-61 (3.3 percent). Babineaux got one drop in 26 targets. Barnett was hurt for a portion of the year, but was at 2-of-20. Furthermore, while Ryans was atop the leaderboard last year, he almost at the bottom of the table this year; in 35 targets, he was only privy to one drop. That's a figure of 2.9 percent.

Among cornerbacks, the leaders were mostly slot guys and, strangely, undersized fellows.

Player 2009 2008
Chris Carr 12.8% 2.7%
DeShea Townsend 12.3% 5.3%
Jerraud Powers 11.7% --
Jacques Reeves 11.1% 6.8%
Antoine Cason 11.1% 2.8%

If you want to argue that the supreme hitting power of Chris Carr or Jacques Reeves yielded those drop figures -- and did so exclusively in 2009 -- be my guest. But you are a strawman of staggering unlikelihood. Again, what might be getting tracked here is the quality of receiver as opposed to the quality of defensive back; most of these guys played in a slot role or against weaker receivers than the players around them.

Last year's leaders at corner also fell back to the pack. Lito Sheppard was at 15.8 percent, albeit in a mere 38 targets; this year, even though he moved to the blitz-happy Jets, he only got drops on a mere 1.7 percent of his targets. Ouch. Drayton Florence went from 13.3 percent in 2008 to 1.3 percent in 2009. Volume leader Charles Tillman, who led the league with 10 drops in 113 targets (8.8 percent), actually stayed reasonably consistent; he had eight drops in 81 targets this year, for a figure of 9.9 percent.

As for the other side of the coin? Well, there were 14 players in 2009 that saw 30 targets or more and couldn't get the break of a single drop. This bad-luck group includes, sorted by target total: Ronde Barber, Tim Jennings, Kelvin Hayden, Randall Gay, David Harris, Keith Ellison, Dawan Landry, Curtis Lofton, Atari Bigby, Drew Coleman, Michael Griffin, Aaron Curry, Chris Hope, and Eric Weddle. Better luck next year, guys. Literally -- you will have better luck next year. We can almost count on it.

Posted by: Bill Barnwell on 08 Jun 2010

35 comments, Last at 14 Jun 2010, 3:43pm by chemical burn

Comments

1
by Brendan Scolari :: Tue, 06/08/2010 - 2:42pm

I"n 2009, safeties and linebackers absolutely yielded more drops than cornerbacks. Passes where those players were in coverage were dropped an even 7.0 percent of the time; passes at cornerbacks were dropped 5.2 percent of the time. I'm thinking the reason why that is has to do with several factors; passes charted with safeties in coverage tend to be longer than passes thrown at cornerbacks, which make them more difficult to catch; they also involve more open space, which may limit the number of times our charters mark a pass as "defensed" instead of "dropped". It could also be the case that passes thrown at those positions are to relatively inferior receivers that work in the slot or as the third or fourth option on a pass play."

And after all of that, you never thought of the reason that I'm guessing was the primary cause of this article? That being that linebackers and safeties nearly always hit harder than cornerbacks? Passes over the middle result in big hits more often because the players that play there hit harder and are usually in a better position to hit a receiver (ie: not sprinting alongside him down the field).

3
by Joseph :: Tue, 06/08/2010 - 2:51pm

Brendan, I'm guessing that "passes over the middle", right in the middle of your last paragraph, is the key. IMO, passes against a zone tend to be where a WR can take a big hit (causing the drop/alligator arms), and that's where the safeties and LB's are covering. I think that the numbers referenced indicate that it doesn't matter how "hard" the player hits.

23
by chemical burn :: Wed, 06/09/2010 - 10:58am

Yeah, but don't you think it's weird that Barnwell puts out all these explanations for why Safeties and LB's cause more drops than CB's and never once mentions the obvious one: because they hit harder. And don't say that's pure speculation because Barnwell's reasons are even more fanciful (borderline illogical) speculation.

The idea that LB's and S's cause more drops perfectly aligns with the idea that harder hits cause more drops. Is there anything controversial about that?

The other half of the article does proves the rest of Barnwell's point: famous "big hitters" don't have seem to make their presence felt outside of the unit stratification... which I can believe because the sample sizes are tiny and the difference between Ronde Barber and Asante Samuel as hitters is obviously not as significant as the difference between Ronde Barber and DeMarco Ryans.

25
by AlanSP :: Fri, 06/11/2010 - 1:39pm

The idea that LB's and S's cause more drops perfectly aligns with the idea that harder hits cause more drops. Is there anything controversial about that?

As used here, "drops" don't include plays where the defender hits the receiver, causing him to drop the ball; those would be "defensed" passes. This is explained in the original article:

"For the purposes of the project, we define a drop as an uncontested pass that the receiver should have caught, so passes that were at the receiver's feet or knocked loose by a big hit don't count. We'll also be looking at passes defensed, which are plays where the defender either breaks up the pass or hits the receiver hard enough to knock the ball out of his hands."

Perhaps this terminology is counterintuitive, but that's beside the point.

The idea I think they were trying to look at wasn't that big hits cause receivers to drop the ball, but that the possibility/fear of a taking a big hit can cause receivers to drop the ball (i.e. what football announcers often refer to as the receiver "hearing footsteps").

The position stratification is at least consistent with this, although it's also consistent with the alternate explanations that Barnwell mentions. You raise some good points about the sample sizes and the differences between positions being bigger than the individual differences among players of a certain position, but those also apply to at least some of the other explanations (e.g. the differences in skill of the receiver being covered).

30
by chemical burn :: Sun, 06/13/2010 - 6:09pm

Huh. Well, I clearly misunderstood. But that makes me ask, what is the defensive element of "defensive drops?" What's the difference between regular drops and defensive drops? It seems like it doesn't matter.

31
by tuluse :: Sun, 06/13/2010 - 7:50pm

I think it is just a normal drop but assigned to which ever defender was in coverage.

I'm not a charter, so I could be wrong.

32
by chemical burn :: Mon, 06/14/2010 - 10:18am

But what sense does that make? Either the defender had something to do with the drop or they didn't. It's either a pass defensed or it's not. If I am understanding it correctly, then I can't imagine how this stat is saying anything about defense. It's just telling us about who drops what...

33
by tuluse :: Mon, 06/14/2010 - 12:57pm

They chart drops, and they chart coverage. They have all the information, they just put it together.

I think they are trying to see if an intimidation factor exists.

34
by chemical burn :: Mon, 06/14/2010 - 3:42pm

But, I guess what I don't understand is, the intimidation factor seems like the obvious and logical conclusion: safeties and LB's cause more drops. Going over the middle causes more drops (whether it be short or deep.) It lines up perfectly - Barnwell coming to the opposite conclusion makes me feel like I don't even understand what this stat is. That's absolutely the only thing this stat tells us and Barnwell dismisses it out of hand...

2
by Jimmy :: Tue, 06/08/2010 - 2:46pm

More proof that Charles Tillman is a very underrated player. Consistent in forcing drops and fumbles year on year.

It was part of the reason for why I was desperate for the Bears to go after Otogwe as having two DBs who rip the ball loose together with a ballhawk (I hope) at the remaining CB spot would have been a nice complement to Peppers, Tommie Harris (healthy?) and two of the best coverage linebackers in the league.

7
by tuluse :: Tue, 06/08/2010 - 5:45pm

Isn't Otogwe a free agent right now?

There is still hope!

10
by Jimmy :: Tue, 06/08/2010 - 7:50pm

Unless he will play for a free pizza after the game not much, sadly.

9
by Nathan :: Tue, 06/08/2010 - 6:16pm

I remember Cris Collinsworth pointing this out during a prime time game last year. I started watching for it and he sure is good at it.

19
by Rich Conley (not verified) :: Wed, 06/09/2010 - 9:46am

"More proof that Charles Tillman is a very underrated player. Consistent in forcing drops and fumbles year on year."

yeah, I'm not so sure about that. Drops are nice, but he gets targeted an awful lot.

22
by Jimmy :: Wed, 06/09/2010 - 10:51am

He was playing LCB in a Tampa2 (or soft zone defense as the amount of T2 was much lower last year) this means he lines up over the flanker who is generally the offense's best receiver. When the QB reads zone why wouldn't he look to throw to his best target (and probably the player with whom he has the best rapport), so he throws to the flanker. It is also easier to work a flanker open against a zone than a split end due to the extra players around to help work receivers open.

When matchups have forced the Bears into man to man coverage Tillman has generally been excellent (yeah elite speed receivers can trouble him but how many corners can cover those guys?). I think the scheme hangs Tillman out to dry a bit as far as some of the charting numbers go.

4
by bigsnack (not verified) :: Tue, 06/08/2010 - 3:25pm

In what alternative universe is ed reed considered a big hitter. A great player..definitely. A big hitter...clearly you haven;t watched him play very much. He avoids contact.

5
by Bill Barnwell :: Tue, 06/08/2010 - 4:10pm
6
by Dean :: Tue, 06/08/2010 - 4:26pm

Admittedly I don't see a lot of Ravens football, so I'm purely going by intuition here, but I always had the impression that he was equal parts ball hawk and hitter. The proverbial mental highlight reel seems to feature just as many hits as spectacular runs after turnovers.

11
by Yoshiki (not verified) :: Tue, 06/08/2010 - 8:17pm

So which alternate universe are you in? Rarely does one see an online poster put so embarrassingly in their place.

12
by BGNoMore (not verified) :: Tue, 06/08/2010 - 9:37pm

Ah, yes, a sample size of one definitely proves the point.

The poster said that the idea of Sanders as a classic, blow-em-up safety would not stand up to an extensive viewing of Sanders' play (and I hope I am fairly representing his point). Then, you watch one play and torch him. I long for a universe in which people can actually grasp the obvious. Sadly, in such a universe, the world-wide web would be completely inaccessible.

15
by Temo :: Tue, 06/08/2010 - 10:22pm

I have to agree; I've never thought of him as a big hitter. I didn't think of him as a light hitter or anything like that-- it's just that when I think of Reed, I think of excellent coverage and ball hawking. Not so much the blow-em-up hits.

18
by Brendan Scolari :: Wed, 06/09/2010 - 2:32am

Me either. Never heard anyone call Reed a big hitter, he's just a ball hawk. Brian Dawkins, Bob Sanders, and (formerly) Roy Williams are guys that come to mind.

27
by AlanSP :: Fri, 06/11/2010 - 2:32pm

Ditto. It's not that Reed can't hit, but that isn't really what he's known for. He's certainly not someone who I'd use as my one example of a safety with a reputation for being a big hitter.

28
by JimZipCode :: Fri, 06/11/2010 - 3:50pm

Ravens fan chiming in. Ed Reed can't be considered a big hitter, no matter what Knowshon Moreno thinks. Reed has had to avoid contact for at least the last 2, maybe 3 seasons, due to his lingering neck/whatever injury. When he has to tackle someone in the secondary, he's more likely to try to drag them down than lay them out.

Reed is an amazing ball hawk and playmaker. But big hitter? Naw. His intimidation is more in the mind of passers, than in the flinching of receivers.

29
by Nathan :: Fri, 06/11/2010 - 7:50pm

I also read "Ed Reed" and thought whaaaaaa?

35
by chemical burn :: Mon, 06/14/2010 - 3:43pm

Yeah... all the people you could choose to represent "big hitter" than would be unanimously agreed upon and you pick Ed Reed?

8
by rk (not verified) :: Tue, 06/08/2010 - 6:06pm

Can we get some consistency with the Mike/Michael Adamses? The Browns list their DB as Mike Adams, and the Cardinals list theirs as Michael Adams. This is a huge problem in my pathetic universe.

13
by Robert O (not verified) :: Tue, 06/08/2010 - 9:40pm

I've never thought big hitters would induce drops. I've always thought big hitters would impact (no pun intended) in one of 3 (if not all 3) ways:

1) Opposing receivers YPC lower than their usual average. Or simply a lower catch rate (short arm or leg syndrome).

2) Opposing receiver YAC lower than their usual average.

3) Fewere "missed" tackles by the big hitter's team than league average.

This last one is the one I've really thought about. If offensive players aren't nervous (or whatever you want to call it) and can play all out without that little nagging feeling in the back of their minds they are going to be able to play faster and make the defense "miss" more tackles.

This is what big hitters bring to a D IMO and I'd be interested if any data backed up this theory.

14
by BGNoMore (not verified) :: Tue, 06/08/2010 - 10:20pm

Of course, there is a much more relevant issue completely unmentioned: the measurement of a drop is completely undefined and subjective. That such a measurement would not show statistical consistency over time is not a cause for comment; it's rather expected. Also, there does not seem to be any acknowledgment of the receiver being defended; surely, in a 16-game season, who defends "sure-handed-steve" vs. "stone-handed sven", and how many times, is not random.

Really, there is absolutely no evidence presented here to either support or undermine the theory of "drop-inducingness". But there is a very strong suggestion presented here that drops are completely random, and that theory is completely unsupported.

16
by ammek :: Wed, 06/09/2010 - 1:52am

Completely unmentioned apart from the bit where it was, um, mentioned:

'they [passes targeting safeties] also involve more open space, which may limit the number of times our charters mark a pass as "defensed" instead of "dropped"'

And no acknowledgment apart from the, erm, acknowledgement:

"It could also be the case that passes thrown at those positions are to relatively inferior receivers that work in the slot or as the third or fourth option on a pass play."

Whatever happened to reading the article before commenting?

20
by Rich Conley (not verified) :: Wed, 06/09/2010 - 9:48am

What happened to actually replying to a post that your reply is relevant to?

He's absolutely correct.

21
by DeltaWhiskey :: Wed, 06/09/2010 - 10:28am

"the measurement of a drop is completely undefined and subjective." As someone who has been critical of the Game Charting project in the past I will make the following comment. FO has presumably defined and operationalized what constitutes a drop. Subjectivity in determining whether a particular pass is/is not a drop remains; however, with proper operationalization and assessment of rater-reliability, the subjectivity can be reduced and drops should be able to be measured fairly reliably. Unfortunately, because we have no FO articles describing any of this in depth, we have to push the "I believe" button.

17
by Bobman :: Wed, 06/09/2010 - 2:01am

Two Colts on the "unlucky" list is intriguing. Their pass D was off a little after allowing a record low number of TDs in 2008--this "bad luck" might have something to do with that. Interestingly one of the rookie CBs had a high drop percentage and the vets are low.

Colts fans won't be surprised by Tim "The Human Cushion" Jennings on that list. He takes a lot of heat but I suspect he's doing what he's told--keep the guy in front of you. I'd be catching them too if the guy covering me was 10 yards back....

Hayden is another matter. Requires some ponderation.

24
by mrh :: Wed, 06/09/2010 - 11:26am

It might be interesting to look at wr/te drops in light of this hypothesis. A WR with a high drop rate due to running a lot of patterns defensed by a LB/S might in fact be more sure-handed than a WR with a lower drop rate who is consistently covered by CB.

26
by AlanSP :: Fri, 06/11/2010 - 2:26pm

Regarding the point about linebackers and safeties generally covering inferior receivers, don't you have the data to control for this? For example, you could stratify the data based on the player being covered (e.g. TE, RB, WR1, WR2, etc.). If it's a matter of the skill of the player being covered, you would expect there to be no difference in, say, tight ends' drop rate based on whether they're being covered by a linebacker, safety, or corner (or at least a much smaller difference than you see when you lump all offensive players together).

The point about open space is an interesting one. Because your definition of drops requires that the pass be uncontested, a player who is usually in position to contest a pass will likely not be credited with very many drops due to lack of opportunity (I'm curious about the drop rate for "hole in zone," which can't contest passes by definition).

Because of this, targets may not be the right denominator to use; ideally, you'd want something like "uncontested passes which should have been caught," since then you're only looking at passes which could have been dropped under your definition. I'd be pretty surprised if the game charting data is detailed enough to allow this, though. Something like targets minus passes defensed, while not ideal, would still probably be a more useful denominator for drop rates than simply using targets.