Writers of Pro Football Prospectus 2008

09 Nov 2013

3CD: Third-and-Long (Part II)

by Rivers McCown

Last week we looked at how coaches are giving away approximately 25 percent of their third-and-long play calls. That's how many times, in third-and-7 or longer, coaches either ran the ball, or called a screen or pass that traveled less than two yards in the air. Generally speaking, those coaches were trading about 20 percent of their success rate for the down for that conservative approach.

But there's a reason I stopped short of declaring it poor strategy. In fact, the words I used were "what would appear to be sub-optimal decision making." I used those because -- tossing aside game theory for a minute -- there are actually downsides in trying to convert the first down that we hadn't covered yet. This week, we're going to discuss them and the selection bias inherent in them.

Our first stop is sacks. If our theoretical quarterback on third-and-7 or third-and-10 is trying to throw past the sticks, he's likely going to need a little more time than he would otherwise. Our favorite tool for measuring sacks is Adjusted Sack Rate, so let's compare the Adjusted Sack Rate on third-and-long to our normal Adjusted Sack Rate numbers.

As with last time, these numbers are for the 2012 season only, and by "third-and-7" and "third-and-10," I really mean "third-and-7 or longer," and "third-and-10 or longer."

Adjusted Sack Rate on Third-and-Long, 2012
    Pass Attempts Scrambles Sacks* Sack Rate
Third-and-7   2900 132 267 8.80%
Third-and-10   1693 79 154 8.60%
*-includes Intentional Grounding penalties

It's a little counter-intuitive that the Adjusted Sack Rate actually went down on third-and-10s, but when you consider coaching strategy -- "get the ball out," dumpoffs, etc. -- it's probably just due to being more risk-averse in general in that situation.

The average Adjusted Sack Rate in 2012 was 6.5 percent. The highest Adjusted Sack Rate belonged to the Oakland Raiders, at 8.9 percent. So, essentially, throwing the ball on third-and-long turns an average offensive line into a replacement level one as far as sacks allowed.

Surprisingly, changing how mobile the quarterback is didn't really matter in these situations. Here's what happens to the numbers when we remove the three most run-happy quarterbacks of 2012: Russell Wilson, Cam Newton, and Robert Griffin.

Adjusted Sack Rate on Third-and-Long, Minus "Big 3", 2012
    Pass Attempts Scrambles Sacks* Sack Rate
Third-and-7   2692 100 247 8.80%
Third-and-10   1567 57 139 8.50%
*-includes Intentional Grounding penalties

Then we've got the other big reason to be conservative: turnovers. Let's see how that holds up...

Quarterback Fumbles and Interceptions on Third-And-Long, 2012
    Interception Rate Interception Rate on All Balls Fumble Rate Fumble Rate on All Sacks
Third-and-7   2.90% 2.40% 13.40% 14.30%
Third-and-10   3.10% 2.40% 11.60% 14.30%

Interception rates rise slightly, while fumbles on sacks actually taper down a bit. I can think of enough anecdotal football incidents (the fabled armpunt, quarterbacks less likely to try to make something happen on third-and-long) to explain both of those away, but it's a little surprising just how little those rates change over the long haul.

So we said last week that coaches were trading about 20 percent of their success rate on roughly 25 percent of third-and-longs to go conservative. After looking at the upsides of conservative coaching, what they are receiving in exchange is less downs with a replacement-level sack rate (about 2.4 percent difference) and about an 0.5 percent less chance of throwing a pick.

Yes, I'm aware that some of this can be chalked up to game theory. I'm also aware that some of this conservative coaching is coming in field position areas where gaining a few yards can set up a long field-goal attempt rather than a short punt or a fourth-down decision

Still, I can't help but feel like some coaches are sacrificing a few too many third-and-longs to conservative thinking. There are specific instances where it may make more sense -- immobile quarterback, leaky offensive line -- but even if we say the Adjusted Sack Rate should hit 20 percent on those sorts of plays, it's a stretch to say that it's not smarter to throw the ball past the sticks.

Charting Playback: Anthony Castonzo

The narrative surrounding Andrew Luck is that he has a bad offensive line to work with. Luck, of course, owns some of that on his own. By his very nature, his pocket presence means he's more than willing to take some shots to deliver some deep balls. In fact, on his 58-yard touchdown throw to T.Y. Hilton against the Texans last week, he had a pair of defenders honing in on him.

But last season, the line was really bad. Indy finished 17th in Adjusted Sack Rate and Luck got hit more times than any other quarterback. The Colts turned over two spots on their offensive line. Jeff Linkenbach and Joe Reitz were execrable starters by any objective measure, and Winston Justice was a pedestrian-at-best right tackle. The Colts have had some turnover, but are now starting rookie Hugh Thornton at left guard and free-agent signing Gosder Cherilus at right tackle. In both cases, the replacement is probably better (though Thornton still has a long road to be a solid starter). But the real improvement I've seen with the Indianapolis offensive line is the maturation of left tackle Anthony Castonzo.

Charting the second half of this game, with Indianapolis in full comeback mode, he completely shut down Whitney Mercilus. Most of the time, the two were on an island. And while a few more of those plays involved Mercilus playing contain than you'd think (the Texans brought a lot of heat up the middle), Mercilus never even got close to beating Castonzo.

The biggest issue was hand-play. Castonzo's punch has improved a bit from what I remember it being, and he was guessing right on an awful lot of Mercilus' moves. Castonzo practically held his hands on a few plays. The Colts left tackle is still a step slower than you'd prefer for a tackle -- he found himself behind on the play a few times -- but his recoveries were effective if not aesthetically pleasing.

Obviously we're still waiting for the compilation of the charting stats, but last season, Castonzo allowed 8.5 sacks and committed 28.5 blown blocks passing. In this half, in 32 pass snaps for the Colts, he allowed zero sacks and committed zero blown blocks. He did have a blown block leading to a tackle for loss against J.J. Watt while the Colts were trying to run the clock out, but ... well, which AFC South lineman hasn't had that happen to them?

Defensive improvement is a big reason the Colts have become an AFC contender. Luck's maturation and the new offensive game plan that's boosted his completion percentage is a factor. The reliance on jumbo sets and fullbacks have bolstered the running game.

But one of the most overlooked factors has been just how much Castonzo has turned his career around. He led the league in blown blocks in 2012 -- this year he's, at worst, an above-average left tackle.

Lofty Nuggets

by Ben Jones

Seattle at Atlanta -- 1:00PM ET (FOX)

This rematch from the divisional round of the playoffs has lost a little luster. Seattle has struggled as much as a winning team can possibly struggle the last two weeks against St. Louis and Tampa Bay. Atlanta is in a state of free-fall. On the injury front, Roddy White says he is going to play this week while Corey Peters looks unlikely to play for the Falcons. The Seahawks offensive line is all kinds of banged up: Max Unger is out with a concussion, and Russell Okung and Breno Giacomini still haven't returned from their injuries. Is Seattle truly a great team without their offensive line? If so, they should probably be stomping teams like the Falcons. Or the Bucs. Or the Rams.

Cincinnati at Baltimore -- 1:00PM ET (CBS)

Both of these teams are coming off of embarrassing losses last week. The Bengals lost on a walk-off safety to the about-to-implode Dolphins. The Ravens lost to the Jason Campbell-led Browns, snapping an eleven-game winning streak against their old city. During the offseason the Ravens paid Joe Flacco well and watched most of their defense and Anquan Boldin be scattered across the NFL; Joe Flacco is ranked 25th in both DVOA (-10.3%) and DYAR (17) this season. The Ravens defense has still been above average (-5.6% DVOA ranked eighth), and has actually improved from 2012 (2.2% DVOA, ranked 19th), so the exodus has mostly been covered by a healthy Terrell Suggs, Lardarius Webb, and the addition of Elvis Dumervil. Cincinnati has been better than Baltimore this season, but the injuries are mounting. With Geno Atkins now on injured reserve, will the Ravens be able to run on the Bengals when they haven't been able to run well all year? That is the key to a possible Ravens upset.

Detroit at Chicago -- 1:00PM ET (FOX)

With Aaron Rodgers' injury, and Jay Cutler’s impending return the NFC North will be tight. The Packers, Bears, and Lions (all in the top 13 of DVOA) are realistically battling for one or maybe two playoff spots, so these inter-division games are extremely important. The Bears have above-average offense, defense and special teams for the season -- though without Lance Briggs the defense has been shaky at best as of late. The Lions have Calvin Johnson. That's not a good combo.

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

The Eagles get the fortune of playing the Rodgerless Green Bay Packers this week. While this game features two "backup" quarterbacks, Nick Foles torched the Raiders last week. Seneca Wallace last looked good leading the Iowa State Cyclones.

The Eagles' sixth-ranked offense (14.8%) goes against Green Bay's 26th-ranked defense (7.1%). The gap is even larger between the Packers offense and the Eagles defense, but Wallace is no Aaron Rodgers: the Packers posted a -15.9% DVOA on offense with Wallace last week. On the other side of the ball Clay Matthews is expected to come back this week. The Packers struggle against the pass (20.3%) much more than the rush (-10.8%), with their worse ratings coming against No. 1 receivers, other receivers, and tight ends. A big day could be ahead for DeSean Jackson, Jason Avant, Brent Celek, and Zach Ertz.

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

Even with the Colts near-slip up last week, they still are better than the Rams at all three phases of the game. Last week the Titans had a poor game plan to stop the Rams, the Colts struggled to stop Andre Johnson, and the Colts won while the Rams lost. The Colts should learn from the Titans blunders and formulate an aggressive defensive strategy, since they don't have to worry about any Andre Johnson-level talent in the passing game.

Oakland at New York Giants -- 1:00PM ET (CBS)

The Raiders, coming off getting shellacked at home, are traveling cross-country to face a team coming off its bye that is on a two-game winning streak. If the Raiders weren't playing DVOA's 30th-ranked Giants they might be in real trouble. Terrelle Pryor is expected to play after leaving the game last week with a knee injury, but Darren McFadden is expected to miss the game after injuring his hamstring. Last week, Oakland's pass defense was abysmal. They got absolutely no pressure, and receivers were not guarded well. If the Giants can keep pressure off of Eli Manning -- a big if given how New York's offensive line has played -- then similar results should be expected.

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

EJ Manuel returns for the Buffalo Bills as they travel to play the Steelers, who just gave up a franchise-record 55 points to the Patriots. As Scott Kacsmar mentioned in Monday's Audibles, you don't need advanced statistics to tell you that the Steelers defense was bad. They have little pass rush (30th-ranked Adjusted Sack Rate), which in turn exposes their secondary. The Bills on the other hand have an excellent defense (seventh-ranked, -6.4% DVOA) which should be able to keep the Steelers slightly below-average offense (-0.4%) to a simmer.

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

After last week's poor defensive game plan, how can the Titans screw up their plan against the Jaguars offense? Perhaps intertwining the Sean Taylor memorial defense with Buddy Ryan's Polish defense. If they alternate at a 3:1 ratio, they would average 11 men on the field, so that would be legal, right?

Carolina at San Francisco -- 4:05PM ET (FOX)

This is the best matchup of the week. The third- and sixth-ranked teams by DVOA are riding hot streaks into this game. The Panthers have won four in a row by a combined score of 130-48, and the 49ers have won five in a row by a combined score of 172-61. Stomp stompety stomp. The Panthers have lost their only two games that were within one score (in Week 1 and Week 2), while the 49ers have won their only game within one score (in Week 1). If Riverboat Ron has completely turned his head-coaching career around, this'll be a good week to demonstrate that.

Houston at Arizona -- 4:25PM ET (CBS)

The nightmare continues for Houston. Gary Kubiak and Arian Foster are out. Ed Reed has been demoted. On the bright side Gary Kubiak was well enough to stop by practice and Ben Jones (not me -- the Texans guard) wrote him a poem.

On the football field the Texans are facing the best defense in the league (-19.1% DVOA). The marquee matchup should be shutdown corner Patrick Peterson against resident badass Andre Johnson. How that battle goes will go a long way towards deciding the winner of this one.

Denver at San Diego – 4:25PM ET (CBS)

The player/coach used to be relatively common and successful in sports. Some recent examples: Bill Russell and Pete Rose. With John Fox's unfortunate absence, the Broncos might want to make Payton Manning the quarterback/offensive coordinator. Officially. So Adam Gase can be freed up to save the Broncos from anyone who knows what Jack Del Rio coaches like.

San Diego ranks last in defensive DVOA and 31st against the pass (22.2% overall, 31.1% against the pass). Manning leads the league in passing DVOA (46.9%), DYAR (1359), QBR (84.2), EYrds (3734), and more standard statistics such as yards (2855) and touchdowns (29). Denver will score a lot of points.

Can San Diego keep pace in a shootout? Philip Rivers rates second in DYAR (1034) and third in DVOA (36.9%). Denver ranks 21st against the pass (10.8% DVOA) and is poor against No. 2 wide receivers and running backs with DVOA's of 11.1% and 16.9% respectively. Danny Woodhead, Vincent Brown, and Ryan Mathews all have good matchups in theory. If Von Miller really has turned the tide for the Denver defense, though, this could get ugly.

Dallas at New Orleans -- 8:30PM ET (NBC)

New Orleans is ranked seventh and Dallas 12th overall in DVOA (17.1% and 12.8% respectively). Both have better offenses than defenses, and New Orleans has the better offense (14.8% to 6.9%) and defense (-2.7% to 0.9%).

New Orleans defense excels against the pass (-7.7%) and struggles versus the run (4.2%). Someone may want to wake up Jason Garrett and explain that you can run the ball. New Orleans struggles against other wide receivers (12.5%), but is average or better against all other receivers in pas coverage.

Dallas basically has an average defense (0.9% DVOA), though DeMarcus Ware is set to return this week. While Dallas is average overall against the pass (5.4%, ranked 16th), they are horrid against No. 1 (14.3%) and No. 2 (32.2%) wideouts along with tight ends (23.6%). On the other hand, they are superb against running backs (-49.2%). Jimmy Graham will probably be a monster when he's on the field. Marques Colston probably should have a good day, but nobody is sure if his career is still alive at this point. If not, Lance Moore and Kenny Stills are the next men up. Darren Sproles and Pierre Thomas look to be limited in the passing game.

Last but not least Dallas excels on special teams (6.8%, ranked first), where New Orleans is average (-0.4%). While Dallas is at a disadvantage everywhere else, special teams could propel the Cowboys to a victory.

Posted by: Rivers McCown on 09 Nov 2013

2 comments, Last at 11 Nov 2013, 12:37pm by In_Belichick_We_Trust

Comments

1
by williams_482 (not verified) :: Mon, 11/11/2013 - 12:33am

I may have missed it in this or a previous article, but are sacks being included with downfield passes? An 8.8% sack rate will drop the success rate considerably. Also, odds are pretty good that a large number (I would assume a significant majority) of those screen/dump off passes were thrown because superior options further downfield were well covered and the odds of a back juking someone were better than the odds that a forced throw would be caught past the sticks.

Finally, looking at numbers for 3rd down conversion rates with the PFR play index gives some very similar success rates on 3rd and 7 and 3rd and 10 when running or passing in "normal football" situations (within 10 points, 1st or 3rd quarter, outside of the red zone). From that lense, it looks like coaches are actually really good at optimizing their 3rd down play calls.

2
by In_Belichick_We... :: Mon, 11/11/2013 - 12:37pm

The problem with this of course is that the screens and draws may be check down plays and not actually called by the coach.
The success rate of the long pass on third down is only based on how often the play was run, not how often the play was called from the side line.
I think it is very likely that a long pass play is called out of the huddle and then the QB makes a pre-snap decision whether to run it or to check into the draw or screen instead.
For the team to run the "higher rate of success long pass play" more often, the QB may have to run it against a look he doesn't like. Maybe not so successful then.
Also, if a team doesn't run enough draws and screens on third and long, they'd be inviting the defense to blitz all of the time.
A follow up disection would be to try to figure out how often defenses trick the QB into the check down draw or screen and then stop it easier because they know it is coming.