Writers of Pro Football Prospectus 2008

Most Recent FO Features


» 2017 Adjusted Games Lost

Two NFC teams were hit hardest by injuries last year. One already set the AGL record in 2016, while the other has a coach with the worst AGL since 2002. Also: the Rams' incredible bill of health in L.A., and Tampa Bay's questionable injury reporting.

24 Nov 2013

Three-Cone Drill: The New Run to Win?

by Rivers McCown

One of the assertions I made in my ill-fated Deadspin article (sorry, cuz!) on Baylor was that at this point of time in football analytics, we are still popping holes in conventional wisdom. One of the things I offered anecdotally -- for the second time this year, I might add -- is that if the 2009 Texans were so good at play-action (second-biggest positive DVOA difference despite starting nobody but Ryan Moats and Steve Slaton until the last few games of the year), why would anybody think that the success of the play-action pass is in any way tied to the back?

Well, since Deadspin asked me to clarify the position, I went ahead and actually did the research on it. I intentionally left out 2009, because I wanted the Texans example that I knew about left out, lest it skew things. I took the 2010-2012 seasons, found the 10 teams with the biggest positive and negative differences in DVOA for each season, and lined their seasonal running DVOAs up next to them to see what a pronounced effect the run game had.

So, turns out, the 30 teams that were the best with play-action ... actually had a lower rushing DVOA than the teams that were the worst at it. I point your eyes to the table below.

Play-Action DVOA Swings: Top and Bottom 10, 2010-2012
10 Biggest Drops 10 Biggest Increases
Team Year +/- DVOA w/PA Run DVOA Team Year +/- DVOA w/PA Run DVOA
CLE 2012 -25.6% -10.3% BUF 2012 21.3% 5.3%
NYJ 2012 -22.6% -8.2% STL 2012 21.9% -4.5%
DET 2012 -20.3% 0.3% ATL 2012 22.7% -17.2%
OAK 2012 -19.4% -17.1% IND 2012 24.4% -4.3%
GB 2012 -15.1% -0.7% CAR 2012 29.6% 6.0%
NYG 2012 -14.8% 9.2% DAL 2012 30.0% -9.2%
SD 2012 -11.0% -16.5% SF 2012 34.9% 12.7%
BAL 2012 -9.4% 7.5% MIA 2012 57.0% -8.4%
NO 2012 -8.7% 1.6% ARI 2012 60.7% -22.1%
CIN 2012 -7.6% -1.4% WAS 2012 61.6% 16.4%
NYG 2011 -62.4% -4.9% WAS 2011 22.9% -5.0%
JAC 2011 -31.3% -5.2% MIA 2011 28.1% -11.9%
PIT 2011 -26.4% 6.6% CIN 2011 29.9% -9.8%
TB 2011 -22.8% -2.4% HOU 2011 38.8% 4.3%
PHI 2011 -18.7% 17.7% KC 2011 40.6% -13.1%
BAL 2011 -17.2% 3.8% SF 2011 43.0% -7.3%
NO 2011 -15.5% 18.5% ATL 2011 45.5% -7.3%
OAK 2011 -14.7% 3.3% DET 2011 47.1% -4.1%
GB 2011 -13.0% 5.5% ARI 2011 57.0% -4.2%
BUF 2011 -4.4% 4.5% DEN 2011 68.2% -0.5%
CAR 2010 -56.1% -20.7% DEN 2010 35.1% -16.2%
SD 2010 -36.9% -5.4% TB 2010 35.2% -0.9%
MIA 2010 -35.9% -7.2% HOU 2010 38.7% 19.6%
BUF 2010 -28.0% -8.3% TEN 2010 42.6% -9.8%
DAL 2010 -27.9% -3.7% IND 2010 43.2% -0.5%
BAL 2010 -24.0% -2.1% STL 2010 43.2% -16.7%
PHI 2010 -14.4% 24.3% OAK 2010 48.7% 5.1%
SEA 2010 -10.7% -11.3% NYG 2010 58.1% 5.2%
WAS 2010 -10.4% -8.6% CIN 2010 61.6% -11.9%
NO 2010 -4.8% -8.1% MIN 2010 63.3% 4.6%
    -21.0% -1.3%     41.8% -3.5%

So, what can we learn from this? That the absence of evidence is not the evidence of absence?

I'm not exactly ready to jump to the idea that a good running back doesn't help the play-action pass in a way that isn't necessarily picked up in the broad view of stats. Maybe a good running back makes five or ten extra defenders jump the run a year? Does that seem like a feasible hypothesis?

But when you look at this sample, it's hard to follow the thinking that went through the minds of say, Ryan Grigson and Pep Hamilton with the Trent Richardson trade. Coaches and analysts routinely talk up the importance of the running back in selling the play-action fake, but at the NFL level, recent history says it just doesn't matter. Ray Rice and LeSean McCoy were two of the best backs in the NFL over the past three years, and their teams were among the 10 worst with play-action twice in those three years. Did the trifecta of Chris Wells, LaRod Stephens-Howling, and Ryan Williams instill fear into the hearts of the 2012 Cardinals opponents? (Or, at least, the same fear that they instilled in the Cardinals themselves?)

One theory I have going forward that I'll try to come back to in the offseason is just that we need to evaluate a team's success on certain play types -- screens, play-action passes, deep balls -- in the context of who is coaching the team. I have only anecdotal evidence, of course, but it feels to me like that matters much more than who the actual players are.

Until then, I think we've got some convincing evidence that the success of the running game does not necessarily mean anything for the play-action passing game.

Lofty Nuggets

by Andrew Potter

Tampa Bay at Detroit -- 1:00 PM ET (FOX)

In theory, a Calvin Johnson versus Darrelle Revis matchup should be delectable: the league's best receiver against arguably the premier man coverage cornerback in the game. In practice, Tampa Bay has largely deployed Revis in zone coverage and ranks a mediocre 18th against number one receivers. Johnson, of course, leads the league in receiving yards, DYAR, and touchdowns -- if Tampa do plan to use Revis in his natural role, this is the obvious game in which to do so. For the Buccaneers, Bobby Rainey follows up his outstanding game against the terrible Falcons with a much tougher matchup against Detroit's fourth-ranked run defense (-19.3%), but Mike Glennon -- quietly having a very good rookie season, ranking 15th by DYAR and 14th by DVOA -- should find more opportunities against the 27th-ranked pass defense. Still leading the division despite last week's disappointing loss in Pittsburgh, avoiding further disappointment here is imperative for the Lions: it could be years before they get a better opportunity to finally win the NFC North.

Minnesota at Green Bay -- 1:00 PM ET (FOX)

Minnesota came into 2013 with the league's best running back coming off a career year in which he almost single-handedly dragged the team into the playoffs despite an unimpressive passing offense, whereas the Packers entered 2013 with a great passing attack but no clear answers in the ground game. While the Vikings have struggled, the Packers seem to have found their answer: Eddie Lacy ranks seventh by DYAR and eighth by DVOA despite missing time with a concussion. However, the dependence of Green Bay's running game on its passing attack has been plain in the past two weeks: the Packers posted negative rushing DVOA in both games, having previously not done so since Week 1. Scott Tolzien (-19.%) has been serviceable at quarterback in place of Aaron Rodgers (32.9%, third), but unlike Rodgers he won't win games on his own and his presence won't open holes for the ground game. A strong running game being hampered by a lack of passing threat? Vikings fans can tell them all about how that works out.

Jacksonville at Houston -- 1:00 PM ET (CBS)

In Week 11, nine of the thirty active NFL teams started a quarterback who wasn't their opening day starter. A tenth, St. Louis, was on a bye but would otherwise have done so. In seven cases, the starter was injured and inactive, leaving three teams who started a different quarterback voluntarily: those teams have a combined record of 5-25. Two of the three meet in this game, with the atrocious Jaguars (-57.1%, 32nd) astonishingly having the chance to move out of last place in the division with a win over the very bad Texans (-23.9%, 30th). Both quarterback changes are justified as the current starters are much better than their immediate predecessors, but both teams are bad enough that it hasn't really mattered: this is a game for draft position and very little else. One of those Sundays when even a win feels like a defeat … or, for the few remaining optimists, even a defeat feels like a win.

San Diego at Kansas City -- 1:00 PM ET (CBS)

Opposites collide in this divisional game at Arrowhead stadium, as the Chargers and their excellent passing offense (50.7%, second) visit Kansas City's very good DVOA pass defense (-15.7%, third) while the Chiefs and their struggling offense (-3.2%, 18th) get to host San Diego's dreadful defense (22.0%, 32nd). The Chargers come into this game on a three-game losing streak, but their pass offense (50.7%, second) remains elite: the worst game of their season is better than any Chiefs passing performance since Week 4. The Chiefs defense has feasted on bad opponents, but recently had an awful game against Cleveland and its only big game since Week 6 came against the Jeff Tuel Bills. Still, they've only allowed more than 17 points once while the Chargers have only allowed fewer than 20 twice (one of those was against the Jaguars), so something will have to give for the Chargers to win this one and keep themselves in the playoff hunt.

Carolina at Miami -- 1:00 PM ET (FOX)

Finally getting some of the mainstream recognition their strong play deserves, DVOA darling Carolina visits a Dolphins team which still has a good chance at the second AFC wild card berth: Miami's 16.5 percent chance of making the playoffs, per our playoff odds, is barely lower than the Jets and Steelers and higher than the other contenders for the AFC sixth seed. The Dolphins remain mediocre, however: an average defense and a below-average offense with a penchant for the type of close games until recently considered Carolina's bête noire. For Miami to keep this game close against a run-heavy Panthers team, they will have to shore up a run defense ranked 30th by DVOA and find a way of attacking Carolina's solid, balanced defense (seventh against the pass, third against the run) with their own mediocre offense (22nd passing, 15th rushing). The more likely outcome by far is Carolina taking another step toward a wild card berth of their own: a far more likely proposition for the Panthers than for the Dolphins.

Pittsburgh at Cleveland -- 1:00 PM ET (CBS)

Last week's home win against the out-of-conference Lions leaves the Steelers as second-favorites for the final AFC wild card place according to our playoff odds projection, which says more about the general standard of the AFC than it does about the Steelers. Pittsburgh is an average team in an average division, with an average offense (2.4%, 15th), average special teams (1.9%, 14th), and a below-average defense (6.3%, 23rd). That makes the Steelers considerably better than the Browns, who are the fifth-worst team in the league by DVOA yet are also somehow (barely) in the AFC playoff picture. Despite their low DVOA, the Browns have already beaten the Bengals and the Ravens at home, and adding the Steelers to that list would not be a shock. Cleveland's defense in particular matches up well against Pittsburgh: nobody in the league is better against number one receivers (-34.8% DVOA) and they're respectable against other receivers (16th against number two, 13th against 'other'). Cleveland's weakness is tight ends (27.1%, 32nd), and Heath Miller (-15.6%, 38th) is not the Heath Miller of old. A sweep of the division at home would place the Browns ahead of the Steelers and possibly Ravens, and set up a potential winner-takes-all showdown at Heinz Field in Week 17.

Chicago at St. Louis -- 1:00 PM ET (FOX)

This game features a pair of backup quarterbacks whose fortunes stand in stark contrast since being thrust into their respective starting roles. Josh McCown has been one of the most efficient quarterbacks in the league while filling in for Jay Cutler, and his 22.6% DVOA ranks sixth among all quarterbacks with at least 83 passes. On the other sideline, Kellen Clemens has been a huge downgrade from Sam Bradford, and his -20.5% DVOA ranks 38th of 43 qualifying passers. Both defenses are middling (13th versus 16th) but the Rams secondary is the weaker of the two, so a Rams upset will require their vaunted defensive line (second by Adjusted Sack Rate) to overcome Chicago's upgraded protection (fourth by Adjusted Sack Rate). Otherwise, Brandon Marshall and Matt Forte are likely to find space against a Rams defense which is bottom of the league against number one receivers and running backs, and on current form McCown is just as likely to find them.

NY Jets at Baltimore -- 1:00 PM ET (CBS)

Probably the most important game of the week in terms of its implications for the postseason hopes of both teams, this game features two of the league's top eight defenses and two of its bottom five offenses. The matchups are even more extreme: Baltimore has the league's second-worst running game by DVOA (-28.7%) while the Jets have its best run defense (-35.9%), and the Jets have the 26th-ranked running game (-15.0%) against Baltimore's top-five run defense (-18.0%). That means both teams are likely to be forced to rely on their passing games, which clearly favors Baltimore: the Jets pass defense (7.9%, 18th) is comfortably worse than that of the Ravens (-1.7%, tenth), and the Ravens pass offense (-2.4%, 23rd) is far better than that of the Jets (-27.7%, 30th). Given that the Ravens also have home field advantage, all signs point toward these two franchises having identical records at the close of play, adding further murk to an already cloudy AFC wild card picture.

Tennessee at Oakland -- 4:05 PM ET (CBS)

A second game in which both teams are likely to start backup quarterbacks; this time DVOA considers both backups to be upgrades over the starter. Ryan Fitzpatrick (0.9%) is 20th by both DYAR and DVOA, and faces an Oakland pass defense which ranks 22nd (13.5%); Matt McGloin (-1.9%) has been barely less efficient than Fitzpatrick, but faces a slightly tougher opponent (3.3%) and has so far been a much larger upgrade for his team. Surprisingly, both teams have the same record, but unsurprisingly only one is considered to have any realistic chance of making the playoffs. A road win here would be a significant boost to Tennessee's chances; a loss would see them all-but-entirely eliminated.

Indianapolis at Arizona -- 4:05 PM ET (CBS)

Arizona boasts the best defense in the league by DVOA: top four against the pass (-14.6%, fourth) and the run (-22.5%, second), the Cardinals play host to an Indianapolis offense which is far more efficient running (12.8%, third) than passing (9.8%, 15th) despite its much-heralded quarterback. The Cardinals offense has struggled and currently sits 25th with a -11.7% DVOA, but the Colts defense has proven unable to sustain its early season form and currently languishes 24th with a DVOA of 6.6%. That makes the Cardinals just about favorites, but by the numbers this is a tight game between two evenly-matched teams. The Cardinals need the win more as they fight to keep pace at the top of the NFC; it would be a surprise to see anybody other than the Colts atop the AFC South in Week 17 irrespective of the outcome here.

Dallas at NY Giants -- 4:25 PM ET (FOX)

Despite starting the season 0-6, the New York Giants have now won four straight games and a win at home against Dallas would leave both teams one game behind the Eagles for the division lead. The Giants have a balanced top-ten defense by DVOA -- 11th against the pass and sixth against the run -- but an offense underperforming in both the passing (27th) and running game (29th) and dire special teams (31st). Against Dallas, that's not a bad matchup: the Cowboys have the 12th-ranked offense and 25th-ranked defense, but Dallas does boast the second-best special teams DVOA and enjoys a particularly large advantage on punt returns (7.1% versus New York's -24.7% on punts). Given how close the reverse fixture was in Week 1, the numbers don't say anything surprising: no outcome would be a shock here.

Denver at New England -- 8:30 PM ET (NBC)

If it wasn't for the names of the quarterbacks, what would this be? The clear best team in the AFC, with its outstanding offense (35.1%, first) and improving defense -- Denver's defense has shown notable improvement since Week 6 -- travels to visit a solid but unspectacular opponent with a declining, injury-decimated defense (-2.1%, 14th) and an inconsistent offense (8.7%, 11th) which has been much better in its past two games. Both teams are very good, but one is playing for home-field advantage whereas the other will be happy to win a division they lead with some margin for error. Denver is the clear favorite by the numbers, but a New England victory wouldn't be that much of an upset -- though for the Patriots to win, they'll probably have to outpace the Broncos in a shootout. Hey, did you hear who the quarterbacks are in this one?

Editor's note: There will be no Three-Cone Drill next week as we take a lap for Thanksgiving. See you in two weeks.

Posted by: Rivers McCown on 24 Nov 2013

11 comments, Last at 25 Nov 2013, 6:20pm by speedegg


by speedegg :: Sun, 11/24/2013 - 1:01pm

Interesting. The point about the run game doesn't correlate to play action is also echoed by Greg Cosell at NFL Films. Cosell has said several times that it's the O-line and QB that make play action effective because Linebackers and Safeties read run/pass keys (lo-hat vs hi-hat). If the O-line and QB can sell "run action", then the Linebackers and Safeties will bite.

by Bright Blue Shorts :: Sun, 11/24/2013 - 4:23pm

Seem to remember than when Norv Turner got to the Raiders in 2004 he stated something to the effect of "It's having the intent to run the ball, not actually being able to run the ball".

Now I'm pretty certain I've misquoted that and I'm not entirely sure he was talking specifically about play actions and obviously his stint as a headcoach in Oakland (and anywhere else) was pretty poor, he is generally regarded as a good offensive co-ordinator. Which makes me think that there might have been something in that statement.

by The Ancient Mariner (not verified) :: Sun, 11/24/2013 - 6:17pm

Makes sense. After all, no defender ever said, "He's not a very good running back, so we can ignore him." Even bad RBs could rack up yards if the other team never bothered to try to stop the run.

by Pat (filler) (not verified) :: Sun, 11/24/2013 - 7:59pm

In fact, the reverse could easily be true: if they know a RB is no threat, they'll be more aggressive to go after him for a big play.

That'd fit with the discovery that the crappy run game teams have good play action.

by Dan :: Sun, 11/24/2013 - 7:10pm

If you were evaluating defenses at stopping the run vs. stopping the play action pass, it would make sense for there to be a negative correlation. Defenses that sell out to stop the run would be more likely to bite on the play action fake - so they'd be better against the run but worse against play action.

I wonder if there could be something similar happening with offenses? Some offenses run their play action passes & running plays in a way that encourages the defense to sell out more against the run, which leads to worse running stats and better play action stats. Other offenses somehow encourage the defense to wait and see if it's a pass, which leads to more effective running but less effective play action.

by JamesdfS (not verified) :: Mon, 11/25/2013 - 2:34am

I was thinking the same thing about DVOA as a rate stat. Defenses are obviously aware that the other team can and will run play-action, and have extensive scouting on tendencies. With teams like 2011 and 2010 Philadelphia under Reid, the 2011 Saints and the 2012 NYG the biggest +DVOA in the bad PA column, might it just be that they ran PA so often that DBs assumed pass even when the play fake took place? Reid was notorious for his Pass all the time philosophy and the Saints are pretty much a foot on the gas all the time team. No idea about the Giants.
The other side doesn't seem quite as clear, but still has examples. The Denver Tebows are noticeable as having the biggest positive DVOA differential for play action, and we all know what they ran. Arizona 2012 played terrible QB roulette and was probaby assumed to never pass if it could be avoided. I don't remember anything about the other teams.

Point is, I suspect the correlation between run to PA ratio and DVOA differential would be more significant. Players are evaluating the likelyhood of a run versus a PA and not the threat that the run represents.

by speedegg :: Mon, 11/25/2013 - 6:20pm

I think you might be touching on a point that most people point out, but in another way. If the running attack is successful, then you move the Linebackers and Safeties several feet off their assignments and not just a step (or two), which open huge holes in the passing game. I would think this effect might be more noticeable late in games, when the defense starts to get worn down, but not sure.

by jbrown (not verified) :: Mon, 11/25/2013 - 1:00am

I'm guessing you may already be onto this with your note (and knowing your Texan fandom/suffering) but I'll put it out there anyway - I wonder how much "best play action" has to do with teams that plan for it, coordinate around it and build it as part of their identity (i.e. Kubiak). The same could be said for "best RB perfomance" coming from teams that really intend and scheme to run well, with some exceptions.

For some teams it's just an extra "gimmick", and for Kubiak or Shanny it's their livelihood.

by wiesengrund :: Mon, 11/25/2013 - 5:32am

The other thing to note is that DVOA as a rate stat tends to punish teams that run too often. When I look at the ATL teams in there, those were the teams that had a horrible run game, DVOA wise, because they just gave Turner the ball about 5 times too much per game, all for a gain of 3. Of course, it's the existence of these surplus-carries (not their quality) that helps the play action game.

by Mussell Shoals (not verified) :: Mon, 11/25/2013 - 5:10pm

DVOA is not the stat to measure a running back's "fearsomeness" or whatever. And, in reality, PA is a small subset of a back's ability to affect an opposing defense. Most obviously, whether the opponent plays with the safety in the box or not, but also: spacing (how close linebackers play to the line and which direction they move at the snap), personnel against personnel (nickel or base versus 3WR-I, etc. etc. etc.), blitzing and on what downs--including type of blitz (run v. pass), and so forth.

What this calls for is something like +/-. I imagine a +/- for WR--especially #1--would also be interesting, given the state of the Falcons, Stafford w/o Calvin, Brady w/o Gronk, etc.

by MJK :: Mon, 11/25/2013 - 5:46pm

Here's a thought--If I recall correctly, running and passing offenses are correlated. I.e. a team with a good pass offense also tends to have a good run offense, and/or vice versa. (I say "and/or" because I do not know of, or wish to imply, that there is causation either way).

But here's the thing. If a team has a good passing offense, because passes are more dangerous to a defense, the defense may be willing to concede the run and (almost) always play pass-first. On the other hand, if the team has a bad pass offense, teams are willing to chance lighter coverage and try to stop the run more, and hence are more at risk of playaction.

So it might be that playaction is correlated to the strength of the passing offense, not the rushing offense.