Writers of Pro Football Prospectus 2008

Most Recent FO Features

RiversPhi04.jpg

» 2014 KUBIAK vs. ADP: The Underrated

Going too low in your fantasy draft: veteran quarterbacks, running backs who do more with their hands than their feet, and Houston's (only) two good receivers.

11 Jul 2011

Formation Analysis: Number of WRs Part I

by Danny Tuccitto

Today, we're going to examine how frequently offenses used various numbers of wide receivers on a given play in 2010, and how well they performed based on the number of wide receivers in the formation.

Overall, 38 percent of plays had two wide receivers in the formation, 43 percent had three wide receivers, 11 percent had one wide receiver or fewer, and the remaining eight percent had four or five wide receivers. These ratios were almost identical to those in 2009, and it reflects once again that third slot receiver and nickelback are pretty much starting-level positions in the modern NFL.

Just as it’s generally true that passing is more efficient than running, it also seems to be generally true that using pass-oriented formations (i.e., three or more wide receivers) is more efficient than using run-oriented formations (i.e., two wide receivers or fewer). In 2010, offensive plays run out of three-receiver sets were the most successful (13.4% DVOA), followed by plays with four or more wide receivers (9.4%), two wide receivers (7.2%), and finally one or fewer wide receivers (-1.0%). Although the top two formations were flip-flopped, the general trend was the same in 2009.

Because two- and three-receiver formations are so prevalent, team-specific DVOAs on these types of plays are closely related to overall offense DVOA. Therefore, most of the statistical intrigue is in plays where the offense lines up in a formation that either screams run (i.e., one or fewer wide receivers) or screams pass (i.e., four or five wide receivers).

On plays last season using four or more wide receivers, leaguewide offensive efficiency decreased by eight percentage points from 2009 despite essentially no change in the frequency with which teams employed this formation. Almost all of the decline can be blamed on the Chargers, Vikings, Giants, and Cardinals. For instance, Minnesota’s DVOA in four-or-five-receiver sets fell 142.5% in 2010 even though they used the formation just as much last season as they did in 2009. The same can be said for the other three teams, all of whom seemed to have coaches trying to fit square pegs into round holes after personnel moves, injuries, or (tentative) retirements.

The worst offender in this regard was Arizona. In 2009, they had Kurt Warner, Larry Fitzgerald, Anquan Boldin, and Steve Breaston, so employing four or five receivers on 36 percent of their plays made perfect strategic sense. However, after failing to adequately replace Warner and Boldin during the offseason, it didn’t make sense at all in 2010. Nevertheless, the Cardinals still used four or five receivers more frequently than any other team. DVOA results speak to the lack of wisdom: a 50% drop in DVOA with four or more wide receivers, and a 41.5% drop in overall offense.

On the other end of the spectrum, most of the league leaders in DVOA for formations involving one or fewer receivers were teams generally associated with run-oriented offenses (e.g., Oakland, Minnesota, Cleveland, New York Giants and Jets). However, one pass-oriented team in particular stands out: the New England Patriots. (No, they don’t stand out because I had to take an oath of allegiance upon hiring.) After drafting Rob Gronkowski and Aaron Hernandez, New England completed their transition away from the shotgun spread toward more of a two-tight-end alignment. As a result, their DVOA in formations with one-or-fewer wide receivers increased by 14.4% in 2010 after having dropped from 14.5% in 2008 to 6.5% in 2009.

In the table, I’ve ranked a teams' formation-specific DVOAs provided that they ran five percent or more of their plays from that formation. Sorting the table in various ways didn’t make any particular trend stand out, so it’s in alphabetical order.


TEAM 0-1 WR DVOA Rk 2 WR DVOA Rk 3 WR DVOA Rk 4-5 WR DVOA Rk 0-1 WR Freq. 2 WR Freq. 3 WR Freq. 4-5 WR Freq.
ARI -45.3% 28 -10.0% 30 -46.0% 32 -32.1% 18 6% 34% 29% 31%
ATL 0.4% 13 15.8% 7 14.3% 16 45.1% 4 26% 32% 36% 6%
BAL -29.0% 26 16.4% 5 29.8% 7 20.7% 7 10% 45% 41% 5%
BUF 1.1% -- -16.0% 31 -3.7% 25 4.3% 13 3% 26% 46% 25%
CAR -36.5% 27 -34.1% 32 -17.4% 31 12.0% 10 16% 47% 31% 7%
CHI -13.9% 21 -7.4% 26 -13.9% 30 64.7% 2 11% 29% 50% 10%
CIN -12.7% 20 0.0% 23 24.0% 11 -32.3% 19 6% 31% 57% 6%
CLE 16.4% 7 -3.6% 24 0.2% 23 31.6% -- 13% 45% 40% 2%
DAL -4.0% 16 -9.1% 29 17.0% 13 115.9% -- 20% 34% 43% 3%
DEN -14.8% 22 14.0% 9 17.8% 12 17.7% 8 13% 28% 53% 6%
DET -39.6% -- 8.2% 14 11.0% 20 17.9% -- 2% 39% 55% 3%
GB 5.4% 10 4.8% 20 34.9% 4 22.8% 6 13% 27% 41% 19%
HOU 11.9% 8 33.8% 2 31.5% 6 7.1% -- 11% 46% 40% 4%
IND -48.9% -- 13.1% 10 27.4% 8 8.8% 12 2% 19% 72% 7%
JAC 3.8% 11 15.8% 6 4.5% 21 101.9% -- 16% 38% 43% 2%
KC 8.9% 9 8.0% 15 24.2% 10 -3.7% -- 15% 47% 34% 4%
OFF 0-1 WR DVOA Rk 2 WR DVOA Rk 3 WR DVOA Rk 4-5 WR DVOA Rk 0-1 WR Freq. 2 WR Freq. 3 WR Freq. 4-5 WR Freq.
MIA 1.7% 12 7.3% 17 14.2% 17 12.5% 9 16% 36% 42% 7%
MIN 22.2% 3 2.4% 21 -8.2% 28 -94.1% 21 9% 39% 46% 6%
NE 20.9% 4 65.7% 1 42.7% 2 79.0% 1 15% 39% 40% 6%
NO -4.9% 19 9.5% 11 13.8% 18 35.7% 5 13% 34% 38% 16%
NYG 22.3% 2 8.8% 13 24.6% 9 -9.0% -- 12% 44% 40% 4%
NYJ 18.7% 6 7.9% 16 16.1% 15 -50.0% -- 12% 46% 39% 4%
OAK 23.8% 1 2.0% 22 1.9% 22 27.5% -- 7% 50% 42% 1%
PHI 15.8% -- 19.1% 4 34.7% 5 11.0% 11 3% 34% 51% 12%
PIT -4.8% 18 30.5% 3 16.9% 14 63.9% 3 16% 30% 41% 13%
SD 19.6% 5 8.9% 12 66.8% 1 -32.0% 17 13% 53% 29% 5%
SEA -24.4% 24 -8.1% 28 -5.9% 26 2.3% 14 9% 40% 47% 5%
SF -3.6% 15 6.3% 18 -7.8% 27 -2.1% 15 16% 41% 37% 6%
STL -18.2% 23 -4.1% 25 -12.4% 29 -10.8% 16 5% 35% 49% 10%
TB -4.7% 17 6.0% 19 36.1% 3 42.6% -- 11% 46% 41% 2%
TEN -24.6% 25 15.2% 8 -0.9% 24 -11.3% -- 6% 49% 44% 1%
WAS -3.4% 14 -7.7% 27 13.3% 19 -39.3% 20 12% 41% 42% 5%
NFL AVG -1.0% -- 7.2% -- 13.4% -- 9.4% -- 11% 38% 43% 8%

Posted by: Danny Tuccitto on 11 Jul 2011

56 comments, Last at 14 Jul 2011, 2:54pm by The Delivery Guy

Comments

1
by Karl Cuba :: Mon, 07/11/2011 - 1:35pm

Great stuff, any chance of seeing the stats for running and passing out of these formation groups?

2
by Waldo (not verified) :: Mon, 07/11/2011 - 2:03pm

Interesting fact that probably correlates to nothing, though it seems to:

4 teams had greater than 10% frequency of use in all 4 categories (Chi, GB, NO, Pit). All 4 made the playoffs. 3 of them played on Championship weekend. 2 of them played in the Superbowl.

3
by bill (not verified) :: Mon, 07/11/2011 - 2:11pm

Perhaps talent comes first, yielding an ability to do many different things well, and coaching creativity naturally follows?

My mind, regarding the converse, goes to Sam Wyche's teams (lack of talent + creative game planning = circus).

Bill

5
by Dean :: Mon, 07/11/2011 - 2:35pm

Whenever I think of things like this, I'm reminded of how I used to hear the phrase "so many weapons" used to describe the late 80s 49ers. If you tried to take away Rice and Taylor, Roger Craig would beat you, and when you thought you had the 3 of them taken out of the game, Tom Rathman would get one of his occasional touches and burn you with it. It's of those things which has stuck with me. As much as it drove lots of Philly fans nuts, I always appreciated the fact that Donovan McNabb always seemed to finish the day with 25-30 completions and nobody caught more than 3 balls. If you force a defense to account for EVERYBODY in the passing game - even the FB and the GURT - it opens up that much more room for the guys who really can make a difference. Even when guys are limited, if you draw up a play or two each game to take advantage of matchups and get the ball in their hands, then the defense has to account for them every play. It actually kinda surprises me that there hasn't been a growth of 25-catch fullbacks to fill that niche.

8
by Independent George :: Mon, 07/11/2011 - 3:08pm

Well, I think it had more to do with McNabb having to throw to Todd Pinkston, Freddie Mitchell, and L.J. Smith, as opposed to purposely trying to distribute the ball. Before (and between) T.O. and Jackson, Brian Westbrook was his best receiver.

12
by Theo :: Mon, 07/11/2011 - 3:37pm

Defensive gameplanning is sooooooo much easier when you know you have to take away one or maybe 2 guys (or plays) to get an offense into struggle mode.
.
GURT?

15
by Dean :: Mon, 07/11/2011 - 3:40pm

GURT = Glorified Undersized Right Tackle.

It's a Tanierism from a few years ago. I've made it my personal mission to make it catch on. Obviously, it's not quite working out and the rest of the world hasn't realized just how awesome GURT is. Help is appreciated here, folks!

21
by Theo :: Mon, 07/11/2011 - 7:00pm

Ah.
After a little research, I must conclude that GURT has indeed only catched on with you.
I think your struggle is part explained by the BOLLUAER (Backup Offensive Lineman Lining Up As Eligible Receiver), who has replaced the GURT nowadays.
(see the article about extra linemen of a few weeks ago)

30
by Dean :: Tue, 07/12/2011 - 9:18am

If the world isn't yet ready for GURT, how in the world could they ever embrace BOLLUAER?

It is the worlds loss.

17
by Alaska Jack :: Mon, 07/11/2011 - 4:24pm

It was very striking in Rich Gannon's years with the Raiders. In Gruden/Callahan's offense,it seemed like they were making a deliberate attempt to spread passes around to 7 or 8 guys a game. - aj

6
by JimZipCode :: Mon, 07/11/2011 - 2:41pm

Sam Wyche won a conf championship with the Bengals. If Montana & Rice hadn't been on the other sideline, he'd likely have a Super Bowl ring. As it was he came close.

7
by Danny Tuccitto :: Mon, 07/11/2011 - 2:55pm

Interesting fact that probably correlates to nothing

Actual correlations agree.

4
by Dean :: Mon, 07/11/2011 - 2:25pm

It may not have been efficient for the Cards to use 4 and 5 WR formations, but from the looks of their RBs, I have a hard time imagining they'd have been much better using 2 RBs.

9
by Shattenjager :: Mon, 07/11/2011 - 3:12pm

But maybe they would have been better with no QB.

Note: This is a joke. Obviously, they would not. Having no QB all the time would be extraordinarily difficult and is not likely to be successful in the current NFL.

10
by tuluse :: Mon, 07/11/2011 - 3:18pm

It would have been an interesting experiment however. They needn't even have gone to no QB at all, but just less than 50% of the time. Try a bunch of direct snap stuff or maybe some old T-formation or single wing plays.

14
by Harrison Bergeron (not verified) :: Mon, 07/11/2011 - 3:38pm

I remember seeing a tongue-in-cheek article in some 1999 football season preview magazine - maybe Athlon - right after the Saints drafted Ricky Williams. The article predicted the upcoming Saints season week-by-week, theorizing that Ditka would go through Danny Wuerffel and the Billy Joes before finally settling on Ricky running out of the single wing 60 times per game.

It probably wouldn't have hurt; their QBs ended up throwing 30 interceptions and the Saints won 3 games.

11
by The Powers That Be :: Mon, 07/11/2011 - 3:23pm

You have to have a quarterback because if you don't you're likely to have a lot of muffed snaps.

Apologies to Casey Stengel.

13
by Dean :: Mon, 07/11/2011 - 3:38pm

Sadly, with their RB corps (pun intended) that just might be even worse.

Maybe they could run a 6TE offense?

41
by Aaron Brooks' Good Twin (not verified) :: Tue, 07/12/2011 - 6:55pm

Didn't Carolina actually do this in the last game one season after their two best QBs got injured and they were left with Chris Weinke?

Yeah, here it is, from 2006: http://www.pro-football-reference.com/boxscores/200612240atl.htm

Fun stats
Carolina through 7 passes and ran 52 times, but their TD was scored on a pass.
Both starting QBs started in the 2000 national championship game.
The Atlanta QB (Vick) lead his team in rushing yards.

This game was straight out of the 1940s.

Odd bit, since 2000, teams that have thrown 10 or fewer passes in a game are 6-1 in those games. The only team to lose was Houston, to Indy, in 2005. Granted, they're also the only team in that bunch to give up more than 13 points. 3 of the 7 won in a shutout, and 5 of the 7 surrendered 6 points or fewer.

43
by Shattenjager :: Tue, 07/12/2011 - 7:42pm

Teams that have thrown 0 passes in a game are 1-0 in NFL history.

44
by Theo :: Tue, 07/12/2011 - 8:15pm

I bet those 6-1 teams were leading a defensive battle, running out the clock... am I right?
Correlation/causation, welcome to Football Outsiders my friend.

46
by Aaron Brook's Good Twin (not verified) :: Tue, 07/12/2011 - 10:32pm

There are a lot of late season, playing out the string games between good teams and not so good teams, though.

And a random Cincinnati-Cleveland grudge match between 3-10 teams, where one team only ran and the other only passed.

51
by SandyRiver :: Wed, 07/13/2011 - 11:58am

One was the end-of-season Pats-Bills game in 2008, played in winds gusting over 60 mph. I think Cassel was 5-of-8 or something like that, and unsurprisingly the Pats won handily.

16
by John (not verified) :: Mon, 07/11/2011 - 4:08pm

And winner for most consistent team by a country mile, our Indianapolis Colts.

18
by Jimbo :: Mon, 07/11/2011 - 5:05pm

How is a WR or TE being defined here? If a TE is split out is he still a TE or counted as a WR, and visa versa. One guy who comes to mind is Hernandez of NE. He's listed as a TE but is split out more than half the time. To me he's just a slightly overweight WR.

22
by Danny Tuccitto :: Mon, 07/11/2011 - 7:05pm

As per Aaron (Schatz, not Hernandez), TEs aren't counted as WRs for the purposes of this analysis. Didn't want to get bogged down in methods, so just made it kind of implicit in my points about ARI and NE. ARI somehow found 4-5 actual WRs to line up among a WR corps that included Fitz, Breaston, and 4 of the commenters to this piece. Why they'd even want to is beyond me of course. On the flip side, NE used more 0-1WR sets because, as you say, hernandez or gronkowski were split wide pretty often, and they don't count as "WRs" here. It's basically a tale of 2 teams, one of which adapted scheme to existing personnel, and the other tried to adapt personnel to existing scheme.

24
by justanothersteve :: Mon, 07/11/2011 - 8:10pm

I don't have any proof. But I imagine the Packers would have similar numbers had Finley stayed healthy. He was often split wide before the injury. I'm more surprised with Chicago having 10% 4-5 WRs. Nothing against Cutler, who I think is quite good. But it means either Olsen or Forte out of the formation which takes their two best pass catchers out of the play.

26
by tuluse :: Mon, 07/11/2011 - 11:02pm

Actually, per Aaron, it doesn't mean that.

Also, Olsen isn't very good.

29
by Shattenjager :: Mon, 07/11/2011 - 11:46pm

I'm glad someone else said that about Olsen. I don't understand why people seem to think he's so good. He still hasn't even been above replacement level as a receiver per DVOA (he topped out at 0.0% in 2007).

25
by Aaron Schatz :: Mon, 07/11/2011 - 10:02pm

Actually, technically, here's how it works.

If a TE is split WIDE, he counts as a WR. However, if he is in the slot, he counts as a TE who is flexed out.

RB count as WR if they are wide OR in the slot.

31
by TimTheEnchanter (not verified) :: Tue, 07/12/2011 - 9:21am

When your QB goes from Kurt Warner to Derek SkeltonHall, you have no running game, and you lose Boldin to boot, there's not many schemes that adapt well to your personnel. You're probably gonna suck either way, so might as well stick with what you've been doing because at least it's familiar, no?

35
by Theo :: Tue, 07/12/2011 - 10:49am

With that offensive line, it really doesn't matter who you put around it. They've lost Alan Faneca and didn't address the problem in the draft.
They are trying to fix their offensive line with drafting 2 extra running backs and a TE.

39
by tuluse :: Tue, 07/12/2011 - 3:50pm

According to Ben Muth, their line was actually ok.

36
by SandyRiver :: Tue, 07/12/2011 - 11:07am

Why not. Their offense was wretched for all the formations, but the 4-5 WR sets' terrible -32.1 ranked 18th while the others were 28/30/32nd. 2 WR was their least bad but 4-5 was their "best" compared to the other teams.

Edit: Forgot to account for the unranked <5% teams.
0-1: 28th of 28
2 WR: 30th of 32
3 WR: 32 of 32
4-5: 18th of 21. Still their highest rank (three teams worse vs 2 teams for 2 WR), but this is historically bad.

38
by tuluse :: Tue, 07/12/2011 - 3:49pm

In this case I think it's more instructive to look at the DVOA instead of ordinal rankings. Just because other teams are stupid doesn't mean the Cardinals should be too.

40
by SandyRiver :: Tue, 07/12/2011 - 4:21pm

Agreed. I was just amazed at how uniformly awful their offense was, no matter how many WR were on the field.

37
by Intropy :: Tue, 07/12/2011 - 2:21pm

You could try running an offense based on whatever it was they used to run before Pop Warner invented the single wing.

42
by Aaron Brooks' Good Twin (not verified) :: Tue, 07/12/2011 - 6:59pm

Difficulty: the flying wedge has been illegal for 100 years. That was pretty much what pre-dates the single-wing. In those years, a 0-0 match was considered a high scoring affair. Legends were written about teams that managed to rack up a safety.

45
by Intropy :: Tue, 07/12/2011 - 8:46pm

That was interesting to read about. Thanks for providing the info. Seems the wedge was banned in 1894, and the single wing debuted sometime after Pop Warner started coaching Carlisle in 1907. I'm sure adoption wasn't immediate and universal. Does anyone know what people used in the interim?

47
by Aaron Brook's Good Twin (not verified) :: Tue, 07/12/2011 - 10:36pm

The era between the end of the flying wedge (which never lasted very long, nor really ever worked all that well) and legal forward pass was a morass of changing field sizes and layouts, differing zonal rules, changes in rules for men-in-motion and men on the line of scrimmage. Basically, nothing consistent prevailed, largely because the rules kept changing from season to season and differed between the east coast to the midwest to the west coast. It was basically very organized chaos.

48
by Jimbo :: Tue, 07/12/2011 - 11:01pm

Thanks Danny and Aaron.

I'm not even sure of what else to apply the calcs to but I presume that when the effective teams bring out these sets - if i'm rooting for them - I can be more excited.

In practical terms the TE v. WR difference is huge in terms of perception and it's probably difficult to assign values to the difference per scheme. Depending on how a defensive coordinator perceives that a receiver-type should be covered; whether by a LB, S or CB, if coverage skills increase linearly depending on the aforementioned positions (and yes this is guessing horsecrap I've thought of on a posting whim) the point of figuring out such a difference is key in defensive scheming. The term of base defense are worthless because of this. Teams will continue to go corner (or S/CB) heavy because increasingly the TE/WR differences are gray. "Sub Packages" are the new "Base Defense". 3-3-5 all the way.

19
by Timmah! (not verified) :: Mon, 07/11/2011 - 5:18pm

Buffalo had 3+ receivers on 71% of its plays? Wow.

Well, I guess when you don't have a TE worth anything you might as well, but still, that seems like the Texas Tech offense, not an NFL one.

20
by John (not verified) :: Mon, 07/11/2011 - 6:41pm

Hey now, the Colts had 3 wide receivers on 72% of their plays, 79% counting 4+. Don't be dissing my man Peyton.

But seriously, I see several teams with 60% or more on 3+, so Buffalo isn't all that far out of line.

49
by DEW (not verified) :: Wed, 07/13/2011 - 12:52am

And given their DVOA success on those 72% of their plays, there's good reason why they're doing that well. (And to be honest, for a good chunk of the season, the Colts' 4th WR was some guy out of the stands handed a helmet.)

...three tries for the Captcha to come up entirely in an alphabet which is used in the English language. Mathematical symbols the first time and Greek the second. @_@

56
by The Delivery Guy (not verified) :: Thu, 07/14/2011 - 2:54pm

And I think BUF is pretty much 4 WR's, I read somewhere that the Packer's lead the league last season in 5 WR sets, more than doubling the rest of the league combined in plays out of that formation.

23
by sundown (not verified) :: Mon, 07/11/2011 - 7:23pm

In Arizona's defense, perhaps they figured that as inaccurate as Anderson and Skelton were, their only hope was to have lots of receivers out there.

There must be some amazing stories to be told about the behind-the-scenes goings-on with last year's Cardinals. Leinart was heir-apparent, puts up some decent numbers in the preseason but gets released without ever seeing the regular season with some explanation along the lines of "he was checking down too often" which might normally sound reasonable except for the fact that Anderson had looked absolutely terrible. So, they move into the regular season with a TOTALLY different sort of QB than Warner was (not just that Anderson stunk, but that his style was nothing like Warner even when he was effective), less receiving weapons, et al, yet STILL stick with essentially the same look offensively.

They were likely doomed to fail regardless, but it almost looks like their plan was to fail.

28
by Kyle D. (not verified) :: Mon, 07/11/2011 - 11:24pm

I never thought Leinart would last in the league but I still have no idea why they'd release him in a situation like that. No way was that strictly about how he was playing. He must have said or done something to really piss off Whisenhunt. But still weird because they passed up plenty of opportunities to dump him or bring in a Plan B. Does Whisenhunt have a daughter or a hot young wife?

50
by DEW (not verified) :: Wed, 07/13/2011 - 12:59am

Honestly, I think that the most telling comment on Matt Leinart's place in the NFL is that he was unable to win the Texans' backup job over Dan "Where Was The Back Of The End Zone Again?" Orlovsky. Obviously, AZ's QB situation was a complete train-wreck in actual practice, but I can see why they'd consider going with one guy who at least did have one season of being an adequate NFL quarterback and two rookies who looked like they might have some promise instead of a very overpriced bust. I strongly doubt that Leinart would have even added 3-4 wins to AZ's total, and keeping him around would have just obscured the fact that Hall, Skelton, and Bartel were not going to be the answer. I have to give Wisenhunt credit for at least knowing enough to know when to cut bait.

52
by Marcus (not verified) :: Wed, 07/13/2011 - 1:26pm

Way too much credit to Wisenhunt. If his plan all along was to dump Leinart, why'd he have him on the roster taking valuable snaps away from the guys he did plan on playing? That makes no sense. Cleveland got something in exchange for Brady Quinn, so they probably could have traded Leinart for something if they hadn't waited until halfway through preseason. No way was he released as part of any master plan.

27
by dtmeyers (not verified) :: Mon, 07/11/2011 - 11:19pm

How were offenses league wide above average? The only formations that the league was below average are 1 or fewer wr, which was only used 8% of all plays. It seems to me that league wide offense should be at 0% DVOA. Am I missing something?

34
by bag (not verified) :: Tue, 07/12/2011 - 10:48am

over the course of a large number of seasons, it should be 0. In a given season, offense or defense may have been slightly better. So 0% is not always average, it just depends on what time span you look at.

32
by TimTheEnchanter (not verified) :: Tue, 07/12/2011 - 9:32am

I love the numbers for the Raiders.

They have 20+ DVOA going out of Power (0-1 wr) and Spread (4-5) and about 2 DVOA for everything else. So what do they do? Run 2-3 Wrs 92% of the time - 3rd most in the league.

That's some fine game management there.

33
by dbostedo :: Tue, 07/12/2011 - 10:46am

Here's a problem with actually making use of this data... you can't assume that the DVOA numbers would stay the same if the formation utilization changed.

For instance, if the Raiders ran more Power, it's completely possible that their DVOA from that formation would drop significantly. It's possible that the reason it's high is because it's not utilized much (perhaps making other teams not prepare for it as much, or creating more surprise when it's used, etc.).

So I think in order to use this data, you'd have to be careful about re-looking at it frequently, and understand the reason a certain formation is or isn't successful. It could be very difficult in practice, particularly with formations used very heavily or lightly, where changes in usage could cause the most DVOA swing.

53
by trill :: Wed, 07/13/2011 - 3:25pm

I agree in principle, but in this specific case the discrepancy in DVOA between the personnel groupings looks big enough to merit some thought on OAK's part. Maybe their DVOA in heavy/spread formations drops 5% if they use each on 5% more snaps, but it's still much higher than their DVOA on 2/3WR formations. Only one way to know for sure, and that's to run it in a game.

54
by alaano (not verified) :: Thu, 07/14/2011 - 10:31am

Redskins were 15 DVOA points below average in 2 WR sets so of course they used them more than other teams.

Galloway was one of the receivers for most of the year, which may account for their failings [face palm]. Wait, why wasn't he on the worst 25 list?

Curious about their 2 WR numbers when Galloway wasn't one of them.

55
by PTORaven :: Thu, 07/14/2011 - 10:42am

I am not sure if these numbers are correct. If you compute a team's offensive DVOA from this chart the result is *always* greater than the offensive DVOA from the main stats page. On average, the difference is 5.5% DVOA percentage points (too high for rounding errors or any other factor that I can think of).

For example, the main stats page has Baltimore's Offensive DVOA as 9.5% (weighted is 9.5%, non-adjusted is 8.9%). From this table Baltimore ran 10% of it's plays with 0-1WR and had an ODVOA of -29%, 45% with 2WR at 16.4% ODVOA, etc. so from this table Total Offensive DVOA for Baltimore should be -29*0.1 + 16.4*0.45 + 29.8*0.41 + 20.7*0.05, which gives baltimore an offensive DVOA of 17.7%. That's 8.2 DVOA percentage points off (almost twice their actual Offensive DVOA of 9.5%).

Maybe I'm missing something. Interesting read nonetheless.