Writers of Pro Football Prospectus 2008

Most Recent FO Features

SprolesDar05.jpg

» Clutch Encounters: Week 2

The Eagles horse-collar the Colts in Monday night stunner. Also: Chicago's rope-a-dope, the end of Seattle's streaks, and a comeback 22 years in the making in Green Bay.

26 Oct 2011

Scramble for the Ball: Strange Correlations

by Tom Gower and Mike Kurtz

Passing Correlation

Tom: The published final pieces we show to you here at Football Outsiders isn’t all that we do. A number of us have little research projects or work on data collection that doesn’t always result in any sort of finished product. Take, for instance, the playoff odds. Do the FO playoff odds really matter at the end of the year? I have the data for it, but it’s difficult to think of a meaningful way to portray that data. Perhaps you could use it to show the narrative of a season, the rise and fall of teams, but that’s not satisfying to me.

One thing I’ve looked at a little bit, but never really written about: Do teams, by DVOA terms, have greater success in other aspects of the game if they have success with their passing offense? This came up first during the 2010 season as I was looking at writing a column for a Monday Night Football game that I never ended up writing about, and again during my research in preparation for writing the Houston Texans chapter in Football Outsiders Almanac 2011.

It turned out that there was a positive correlation (covariance was what I actually tested) between the 2010 Houston Texans’ passing offense by DVOA in a game and their rushing offense by DVOA for that game. As they passed the ball better, they ran the ball better. (Or vice versa. The causal mechanism could work in either or both directions.) I thought this might be a useful framing device for the chapter, so I expanded my research to the 2009 Houston Texans, who had an almost equally strong negative connection between passing offense DVOA and rushing offense DVOA. Well, there went that idea.

I went back, then, to the two teams I initially considered writing about. In 2009, the Arizona Cardinals and San Francisco 49ers were at opposite ends of the spectrum. In Kurt Warner’s final season, when the Arizona Cardinals threw the ball better in a game, they ran the ball better that game. (Or vice versa, as the causal mechanism isn’t clear. And remember, this is all according to DVOA.) The 49ers in 2009 were not like that at all. In fact, when they ran the ball better, they threw the ball worse, and vice versa.

There were plausible theoretical reasons why each team might have been the way they were. The Cardinals did not seem like a good enough team to have a good running attack without Kurt Warner having success passing, and if Kurt Warner wasn’t having success throwing the ball, teams didn’t have to worry about the run game. For the 49ers, Alex Smith was more successful in a shotgun spread attack, while Mike Singletary preferred to run the power run game. When the 49ers could run the ball successfully, they stayed in a formation where they couldn’t pass the ball well. When they couldn’t run the ball, they switched to what was, essentially, their passing formation.

And then we get to 2010. Kurt Warner was gone, but the plausible theoretical reasons seemed like they might still be true. I ran the numbers again, and found that wasn't the case. The Cardinals now had less success passing when they were better running, and the 49ers now had more success passing when they had more success running. Both teams had reversed their statistical profile.

I then expanded my research to the rest of the teams in the league, and found that there was a small but very slight positive covariance between a team’s passing offense by DVOA and their rushing offense DVOA for that game, but it wasn’t anywhere near consistent.

In its own way, this could almost be sort of profound, and it can also be seen as deeply counter-intuitive. Should San Francisco in 2009 have run the ball ineffectively on purpose, deliberately calling runs into the teeth of the defense so they could pass the ball better? It’d be far, far too easy to make a "Mike Singletary might actually have done this" joke here, but even he would probably recognize that the idea sounds absurd on its face.

I’m also not proposing to overthrow the game-theoretic foundations of football. Teams can run the ball and throw the ball, and intelligently build those off of each other. I saw the Texans do that last Sunday with great success, just as they did in 2010. Teams should strive to do that, and sometimes take a step back and look at what they do well and try to do more of it. When they find what they don’t do well, they should try to do less of it. Sometimes this ends up going to unfortunate extremes, and sometimes you end up with the Bills’ K-Gun offense of the early 1990’s and four consecutive Super Bowl appearances.

By the way, there doesn’t seem to be any link between passing offense DVOA and passing defense DVOA, either.

Fantasy Football Update

Tom: Well, both my fantasy games this week involved the league's highest-scoring team. In the league where I have the highest-scoring team, I was the windshield. In the other one, I was the bug. And yes, I put up the league's highest score even though my running backs were Reggie Bush and Jacquizz Rodgers.

Mike: How on Earth did you manage that?

Tom: Drew Brees and a bunch of balanced scoring. Five of my 11 defensive players had at least 10 points, led by Charles Woodson's 25.2.

Mike: You are the only person I can think of that actually won a fantasy game on the strength of his IDPs.

Tom: If this was a normal IDP league, I probably wouldn't have, but the scoring settings make them as valuable as other players.

Mike: Fair enough.

Tom: In my other league, my opponent had Brees, Adrian Peterson, Jimmy Graham, and Mason Crosby. Squash.

Mike: That's kind of funny, actually. I have Brees in both of my leagues but for some reason I'm not passed out under a giant sombrero with a bottle of tequila. That reason is that one opponent had Marques Colston and the other had Graham.

Tom: Don't tell me you lost this week.

Mike: I lost in both leagues. Guess who else I have in both leagues, although not by choice? Ray Rice.

Tom: The Ravens ran the ball eight times Monday night. How could Ray Rice not put up 15 fantasy points?

Mike: I know! And I got to face Matt Forte and Arian Foster, who combined for 71 of my opponent's 151 points. Oh, and The Ultimate Leader decided to basically not play Ryan Torain at all, so my total from RB2 and RB3 was a massive 4.8.

Tom: Well, Forte and Foster WERE much of the watchability of the early slate on Sunday.

Mike: The other league I lost by 3.4 points. Basically on the Anquan Boldin touchdown. Since, of course, Darren McFadden is injured. Again.

Tom: The same Boldin touchdown that put me over the top for the league's highest score for the week? I'm sorry.

Mike: It was fairly brutal, yes. This means that I've fallen below .500 in the first league, and am now tied for first in the second. This week has brought great shame to my house.

Tom: I'm sure you can recover. But if you need a second for seppuku, just give me a call and supply me a sword, and I'll be ready and willing to stand in support. I do have a bye week dilemma, though.

Mike: Yeah?

Tom: Too many players with no games this week, including five IDPs. I'm carrying 14 IDP, so I could voluntarily start two inactive players and take my chances against what looks like a mediocre opponent.

Mike: Are most of the IDPs fungible? They often are.

Tom: This is the league where I got something like half of KUBIAK's top-20 IDPs for the league. My IDPs on bye are Clay Matthews, Lance Briggs, David Harris, Woodson, and Charles Tillman.

Mike: Yeah, I can understand not dropping any of them.

Tom: I'll probably just take my chances and hope for the best, so tune back in next week for my tale of woe of losing by 0.2 points because I started two IDPs on bye.

FO Staff League Update

Known Chumpsky (Rivers, 4-3) 97 def. That's Great Hustle! (Sean, 4-1) 83

And then there were none. The final undefeated team has fallen, hamstrung by terrible performances by Vincent Jackson and Ray Rice and a mediocre outing by Matthew Stafford. Credit to Rivers for sticking with Ben Roethlisberger against a weak defense, but no credit for starting Jimmy Graham because I'm pretty sure that's the only tight end Rivers could get.

Intentional Rounding (Danny, 3-4) 84 def. Equipo del Jefe (Aaron, 2-3) 49

Want to know how dire Danny's bye week problems were? He started Jay Cutler and Nate Burleson. Michael Vick, Hakeem Nicks, DeSean Jackson, BenJarvus Green-Ellis and A.J. Green all sat this past week out, and while Tiny Darren Sproles and Dez Bryant came through with 21 and 15 points, respectively, it really didn't make much of a difference. Aside from Tim Tebow's mystifying 22-point excursion -- yes, Aaron started Tebow over Matt Schaub, I don't know why either -- Aaron's best player was Michael Bush with 10 points. Aaron also had bad luck with injuries (Darren McFadden and Sebastian Janikowski), but I'm not sure that there is any excuse for starting Doug Baldwin.

Reverse Jinxes (Elias, 6-1) 69 def. Los Pollos Hermanos (Rob, 2-3) 67

Two points! I think that ties a staff league record for margin of victory. What is interesting is that regardless of the close score, there is no possible way Rob could have won this game: Half his bench was on bye, and the other half (Peyton Hillis, Robert Meachem, and Greg Olsen) combined for two points. While Cadillac Williams contributed negative points, the low scorer for el Hermanos was Santana Moss at 1 point, so the best he could possibly manage is 68. That is, how you say, wah-wah.

Dyscalculia Plus Ones (Will, 5-1) 106 def. Parts Unknown Mufflers (Ben, 0-5) 48

Will goes back into dominance with a huge victory over poor Ben's sad-sack squad. To be fair, Earnest Graham left his game on the second snap and Ryan Torain suffered from Shanahanitis. Even Mike Wallace could not save him from that debacle. Plaxico Burress had a better game, anyway.

Time Passes More Quickly as You Approach the Burrito

Tom: Like Jake Delhomme's delivery of biscuit justice, this came from a reader submission. Frankly, I'm not sure of the actual mechanics of how you fall into a giant burrito.

Mike: What is it about food commercials that brings out the strangeness in writers and directors?

Tom: Most actual food looks horrible on camera, I think. Even if it's warm and delicious in real life, on camera it looks sickly and unappetizing.

Mike: Well, yes and no. Generally yes, but there is a whole branch of film and photography dedicated to capturing food. A lot of techniques do render a formerly delicious item inedible, however.

Tom: Exactly.

Mike: This doesn't even try, though.

Tom: So you can show fake food in something close to its regular variant, or fake food in a very unnatural variant. This commercial takes the latter path.

Mike: That is a somewhat silly way of saying "you can use realistic food or ridiculous prop food," but yes, they did. I imagine because they realized in their heart of hearts that their commercial itself was fairly ridiculous.

Tom: I do like the simplicity. "Our marketing strategy is giving people a lot of food. Let's show a guy with a lot of food."

Mike: True, but making a lot of actual food is trivial. Here, the writer had a dream of being suffocated by a giant burrito or something and decided to base his next piece on it.

Tom: I'm guessing they first went with the "let's show him with something close to a regular variant," like a typical mondo burrito. And it didn't really look like that much food. So then they decided to go to 1,211.

Mike: Please stop referring to food as "regular variant." Nobody else would ever say that.

Tom: Have I ever mentioned I own some plastic food?

Mike: Fortunately, you have not. I suppose formerly fortunately.

Tom: Well, I do. Bought it in Japan. It was supposed to be a present for my sister, who had expressed curiosity about the concept.

Mike: And what do you use this food for?

Tom: As I don't own a restaurant, it's now decorative. Sitting above my kitchen cabinets with other useless items of bric-a-brac.

Mike: Exciting.

Tom: Oh, it is.

Mike: Anyway, it would be easy to fall into a burrito provided the burrito is of sufficient size. That said, they seem to be screwing their customers with this burrito in particular. There is some stuff along the interior surface, but it's mostly air. I guess the burrito is designed to be crawled into and eaten out of?

Tom: It does seem exceptionally like a hollow cylinder for a real burrito. That said, when I go to a place like Chipotle, I got my burrito without the tortilla, which is calories I don't really need.

Mike: Oh, you're one of THOSE people.

Tom: Yes. I get the bowl. Do you have a problem with that?

Mike: I do. You are trampling all over those non-Mexican people's fake traditions.

Tom: I can eat 200 calories worth of blah tortilla, or I can eat 200 calories worth of Oreos later. That's an easy decision to me.

Mike: ... Nutrition does not work that way.

Tom: I know that ... now. The real question I want to know the answer to: if you do fall into a giant burrito like that, how long will it provide you with lunch?

Mike: Well, how long does it take someone to eat approximately twice his weight in food?

Tom: It's actually a pretty difficult problem, since we don't know how much food there is. We see shots of the guy in the burrito, and of the guy outside the burrito, but we never see a good outside shot of the burrito.

Mike: True, although I'm estimating the burrito is roughly twice his mass.

Tom: I'm not sure just how long the burrito is, or how thick the tube is. Fine, let's pretend he's, what, 70 kilos? That's 154 pounds. Or shall we make him 80kg/176lb?

Mike: He's eating fast-food Mexican. Go with 80.

Tom: OK. From Taco Del Mar's website nutritional information, we know that a Chicken Mondo Burrito is 523g. That's roughly 153 meals. I wouldn't want to go to Taco Del Mar every day, but even if he goes three times a week, that's lunch for a year.

Mike: That is impressive. NASA needs to get on this. Instead of expensive capsules, launch astronauts into space in giant burritos. Use the atmosphere to cook them.

Tom: The astronauts or the burritos?

Mike: Unfortunately, they don't have the funding to figure that part out. Shoot 'em off and hope for the best!

Loser League Update

KICKER: Billy Cundiff wasn't really responsible for his team's offensive performance Monday night, but his -1 from a missed field goal and a made extra point made him that game's less valuable kicker.

WIDE RECEIVER: Danario Alexander ... I never understood quite what he was supposed to be aside from a marginal guy with good deep speed. His 0 points were this week's low. Behind him with 1 point were Eddie Royal, Santana Moss, Miles Austin, Austin Collie, Percy Harvin, and Jacoby Ford.

RUNNING BACK: It's not unusual to see Thomas Jones be ineffective carrying the ball for the Chiefs, and this week's 9 carries for 19 yards and 1 points came as no real surprise. Behind him at 2 were Ryan Grant and Justin Forsett.

QUARTERBACK: We're sorry, but Curtis Painter and his -1 were not available to you for this session of Loser League. Nor was Blaine Gabbert and his 4 points, though Matt Cassel, who tied Gabbert, was.

Awards!

KEEP CHOPPING WOOD: Aqib Talib, come on down! You're this week's winner of "Defensive Player Who Needlessly Commits a Stupid Penalty That Gives the a First Down to the Opponent That's Trying to Run Out the Clock."

MIKE MARTZ AWARD: For as duly maligned as he was for his poor play as a rookie and then his inability to beat out Rex Grossman as a starting quarterback, Kyle Orton turned into a pretty decent NFL quarterback. He's not among the league's elite, but better he's than Grossman was when the Bears made it to the Super Bowl. It's been difficult to identify the quality premium that made the Bears willing to trade him AND high draft choices to the Broncos for Jay Cutler. No, he didn't always have the best game, and maybe in a down week, you could do better, but you could also do a lot worse. John Fox decided to do just what the billboards suggested, and we know that's always a winning strategy, right?

COLBERT AWARD: John Fox, same thing.

Scramble Mailbag

Joe: Assuming Andre Johnson returns this week (which it's looking pretty likely), I need to decide between benching one of these four match-ups: Marques Colston vs. St. Louis, Andre Johnson vs. Jacksonville, Maurice Jones-Drew vs. Houston, and Fred Jackson vs. Washington. Our league has standard scoring, PPR. Thanks.

Mike: Colston is clearly back, in a big way.

Tom: I'm starting Quizz Rodgers and Reggie Bush out of necessity, and you have to bench one of these guys? What is this, an eight-team league?

Mike: He probably got Colston for cheap because he was injured and Fred Jackson because nobody believed in his team. It's not outside the realm of possibility. Just excellent foresight (good guessing).

Tom: Yeah, I can see that. And I did really neglect the running back position.

Mike: Anyway, Johnson in a PPR league is an auto-start.

Tom: If he's healthy, Andre Johnson is the focus of Houston's offense, and he's a better player than Rashean Mathis or Derek Cox. Start him if healthy.

Mike: Fred Jackson has been having a great year. MJD just came off a hideous game, so it's easy to say he's the odd man out.

Tom: That would be my recommendation, but the Texans and Jaguars are a great matchup of weakness and strength. The Jaguars run up the middle as much or more than any team in the league, and they're above-average at it in terms of Adjusted Line Yards. The Texans' biggest defensive weakness is number two corner, especially when it's Kareem Jackson, but coming a close second is runs up the middle. Teams run the ball up the middle against them even more than the Jaguars do it on offense, which is insane. And they're terrible at it, 29th in ALY.

Mike: That would be great information if Jacksonville isn't going to be down by like three touchdowns in the second quarter.

Tom: The real reason you bench MJD in this game is because he seems to be on some sort of informal snap count where his work is limited. And Blaine Gabbert does not seem like he'll be able to take advantage of even the Texans' weakness in the secondary, such that MJD will be productive enough on his limited touches.

Mike: Also because Houston is going to destroy Jacksonville and the running game will be rather irrelevant.

Tom: Deji Karim will likely get his fair share of carries, and you have three other good matchups. And Jacksonville mostly mocks your idea that they're going to start throwing the ball a lot, even if they're down by three touchdowns in the second quarter. In fact, being down by three touchdowns in the second quarter is great, because then you don't have to abandon the run, you have lots of time to come back by running.

Mike: That is why Jacksonville is such a quality organization.

Tom: Absolutely. Just ask the Ravens.

Kimble: I'm ready to reinsert A.J. Green into my lineup in my PPR league. We start 2 RB/WRs and 2 WR/TEs. Last week, I had Steve Smith Esquire, Pierre Garçon, Darren Sproles, and Chris Johnson in those slots. My heart says to bench CJ0.7K and leave Garçon in. However, my head says to always stick with your studs. What should I do? My other choices that aren't hurt or on bye are Davone Bess, Lance Moore, and the other Steve Smith. Thanks.

Tom: If the Titans are honest, they'll really start splitting the work between Johnson, Javon Ringer, and probably even Jamie Harper in goalline and short yardage situations.

Mike: ... why?

Tom: Read this post.

Mike: Ah.

Tom: Chris Johnson is not running effectively. He's not breaking long runs and is leaving yards on the field on short runs.

Mike: Right, but is this a "the Titans will do this" or a "the Titans should do this?"

Tom: It's a "even if the Titans don't do this, do you really want 21 carries for 67 yards"?

Mike: Good point.

Tom: He actually does have some added value in a PPR league. I wouldn't expect him to have six catches like he did this past week every week, but he could have four if they make an effort to keep him involved in the offense even in a rotation, which is what I expect. Your question really is, will Lance Moore, the other Steve Smith, or Bess do better than CJ?

Mike: I'm not sure any of them will, especially since Colston is back with a vengeance.

Tom: And the answers I see are maybe, doubtful, and maybe. Colston, Graham, and Sproles are getting a lot of work in New Orleans. And I wouldn't be surprised to see the Saints try to run the ball a lot against the Rams. Bess I see as a week-to-week proposition. He wasn't bad in a PPR league this past week, but Miami seems to be trying to implode right now.

Mike: Yeah, to answer your question, between these options I think he really would want 21 carries for 67 yards.

Tom: I'd start him too, since I think he has both consistency and explosiveness going for him over the other options.

Flores: I have the same 3C2 (3choose 2) dilemma again: Steven Jackson vs. NO, Mike Tolbert vs. KC, or DeMarco Murray vs. PHI (because Sproles vs. CAR sounds like an extremely delicious match up). I'm guessing Jackson is probably worth starting. Murray played his way to a starting position, but he doesn't seem to be a great receiver which is a little worrisome if Dallas falls behind and needs to throw a lot, but Tolbert got nicked up again and is likely the second back at this point. And then I have the QB controversy of Eli vs. Miami or Ben vs. NE. Both have bad pass defenses, but I'm concerned NYG will build a big enough lead that they start just handing the ball off. PIT vs. NE is more likely to be a shootout.

Tom: Getting in before you, I feel like there's a small but non-zero chance of a Ben meltdown against the Patriots. But barring that, I like him better because I expect the Giants to want to run the ball against the Dolphins.

Mike: I agree. In fact, I'm starting Bradshaw myself this week. For reals this time. At this point, Jackson is really St. Louis' only offense, and New Orleans does not have the most terrifying run defense.

Tom: I haven't watched last week's Chargers-Jets game yet, so I'm not sure what role Tolbert played. If Ryan Mathews will be limited this week, he's a more attractive option. Philadelphia is another team that does not have an elite run defense. I'd start Jackson and Murray. And right now Mathews seems fine. But if his condition takes a turn for the worse and he'll be out or severely limited, I'd probably play Tolbert over Jackson, just because of St. Louis's general level of offensive dysfunction.

Mike: I agree in part, but I would replace Murray. Right now he's way too much of an unknown quantity.

Tom: Eh, I watched him in college. I didn't think he was a great NFL prospect, but if the Eagles give up the kind of holes they've been giving up, he can run through them just like he did against the Rams.

We're turning the corner on the season, and now more than every you need fantasy football help! Send your questions to scramble-at-footballoutsiders.com or check out our handy forum topic, which we promise will actually exist this week.

Posted by: Mike Kurtz and Tom Gower on 26 Oct 2011

16 comments, Last at 27 Oct 2011, 6:49pm by MJK

Comments

1
by Ben Muth :: Wed, 10/26/2011 - 2:48pm

At this point I must defend the Mufflers. Once Charles got hurt, we decided to go into suck for Luck mode. That's why I made sure Phillip Rivers ran the two minute drill with the urgency of an old woman slipping into a warm bath. Wait, this isn't a keeper league where we decide next year's draft order by this year's record. Oh no.

2
by tunesmith :: Wed, 10/26/2011 - 2:53pm

Funny to give Fox both awards. I think that illustrates the two sides of the debate pretty well.

3
by DGL :: Wed, 10/26/2011 - 4:25pm

RB/WR problem. Well, not so much a problem, especially as it's a bye week, as a choice: Standard ESPN scoring, 2RB, 2WR, 1RB/WR Flex. Sproles vs. STL definitely starts. Other choices are SJax vs NO, Mathews vs KC, AJohnson vs JAX, SJohnson vs WAS, Washington vs IND, needing one more RB, 2 WR, and one Flex. Thoughts?

6
by Tom Gower :: Wed, 10/26/2011 - 5:04pm

Jackson, Mathews, Johnson, and Johnson.

Washington was productive early in the year, but then he had Britt to draw coverage. Teams are now able to concentrate on him as Hasselbeck's only reliable option, and he's struggling to get open, which makes him a risky fantasy play.

4
by Keith(1) (not verified) :: Wed, 10/26/2011 - 4:44pm

I really enjoyed the introduction to this article. It was very interesting. I like that I can realistically argue, from eany side, "run to pass the ball," "pass to run the ball," and "they have nothing to do with each other." I guess what we all thought was wrong is not categorically wrong.

5
by Intropy :: Wed, 10/26/2011 - 4:48pm

In this burrito you will find a new definition of pain and suffering as you are slowly digested over a thousand years.

7
by Danish Denver-Fan :: Wed, 10/26/2011 - 5:18pm

I like the intro. I think you need to publish your results even when they are inconclusive. Research is research regardless of the result. And finally; being aware that you know nothing is better than living in complete ignorance. Did that even make sense?

8
by tuluse :: Wed, 10/26/2011 - 5:26pm

My favorite part of scramble is the awards by far. Have you guys considered adding some more?

9
by Tom Gower :: Wed, 10/26/2011 - 6:35pm

Not really, but if you have any suggestions we're willing to listen.

10
by brett (not verified) :: Wed, 10/26/2011 - 9:36pm

so, continuing the niners and cards example above, both teams were better overall on offense in the years when there appeared to be positive correlations between pass and run DVOA, and relatively weaker when the negative correlation was there. it seems like this could be an interesting avenue to pursue-- and could lead to more intuitive results. offenses that do both (run and pass) effectively when they are "rolling" would seem to have a higher ceiling and be more difficult for the defense to untrack, as opposed to an offense that settled into one rut or the other.

i would also be kind of interested in seeing the game-by-game progression of these numbers vs. each other, particularly in the cases where there was a negative correlation, to see if it was something that developed as the season went on (primarily due to game-planning on one side or the other), if it was present at the outset and stayed generally the same, or if it was something that the relative offenses managed to improve on through the course of the year.

just a couple thoughts i had.

11
by brett (not verified) :: Wed, 10/26/2011 - 9:47pm

put more succinctly (sorry), does a positive correlation between running and passing dvoas have a positive correlation on overall offensive dvoa? eek.

12
by Tom Gower :: Wed, 10/26/2011 - 11:26pm

I did look at the link between total offense DVOA and passing DVOA as opposed to rushing DVOA, and found that both had a positive covariance but it was typically stronger with passing. That's from a pretty small sample size (AFC South, 2009 and part of 2010), but is something I could look at on a leaguewide basis.

With only 16 games, I'm hesitant to subdivide that into smaller chunks unless there are compelling reasons to do so such as personnel changes. Typically, there's a bunch of noise in a team's DVOA, Every Week, One Season, but for my FOA chapters I make sure I look at it and see if there's anything to use. Sometimes, there is, and see the Titans chapter of FOA2010 for one example.

13
by tally :: Thu, 10/27/2011 - 10:28am

Covariance does not imply causation one way or another--both could be dependent on a third, lurking variable, and the fact that this only applied to some teams but not others and changes from year to year suggests that it could just be random.

14
by Joseph :: Thu, 10/27/2011 - 3:54pm

I have an opinion on the positive DVOA correlation between running & passing vs. the negative correlation.
I think if team X at the end of 2011 has basically positive DVOA (say, > -1.0) in both running and passing, then they are going to positively correlate. In other words, a team that is at least okay at running and passing will force the defense to concentrate on both--this allows them to run against dime defenses and pass when the D puts 8 in the box. This team will probably have a QB who knows the system well and is able to audible to plays that will take advantage of the defense.
On the other hand, if team X has a good running game and a poor passing game (or vice versa), they probably fit that Niners' profile mentioned--they have running formations, and passing formations. Down and distance + personnel/formation give away too many play calls. When the QB sees that the play call won't work against that defense, he either doesn't have the personnel to make the audible work, or the down and distance are against him (let's face it--draws on 3rd & 15 aren't very successful).
Another thought--maybe, like this year's Titans (see Word of Muth), the O-line is just good at one thing and bad at the other. (see IND, last decade)

16
by MJK :: Thu, 10/27/2011 - 6:49pm

Your last thought is a good one.

A O-lineman can be good at run blocking, pass blocking, neither, or both. Linemen that are good at one but not the other are "mediocre" linemen; ones that are good at both are "good" linemen, ones that aren't good at either play for Chicago...I mean...are "bad" linemen.

Assuming talent variations across the league, teams with bad lines and teams with good lines will show passing and running DVOA in sync with one another, but teams with mediocre lines will show them opposed, depending on the philosophy of the coaching staff and the talent evaluators.

15
by MJK :: Thu, 10/27/2011 - 6:46pm

Brian Burke also looks at correlation between running and passing well.

http://www.advancednflstats.com/2010/10/how-coaches-think-run-success-ra...

He found that, league wide, there is poor correlation between running and passing efficiency (not DVOA efficiency, but rather yards per touch efficiency). The two are nearly independent. However, when you look at "success rate", the two tend to be much more strongly correlated. His definition of success rate is a little more sophisticated than DVOA...he defines any play as "successful" if it increases you expectation of scoring points before your opponent does. So it's like DVOA, but taking field position and game situation into account as well.

I would guess that, league wide, DVOA probably falls in the middle regarding correlating running and passing success, since it's more sophisticated than efficiency but less so than expect points. Meaning, by DVOA, there are probably more teams where running and passing move together, but a few where they are opposed.

And the thought that teams should intentionally do one thing badly to do the other is silly...(probably confusing correlation with causation). It's "kneel to win" all over.