September 2012

I was planning on starting this in week 4, but wanting to know what the Pythagorean expectation was of various teams made me interested in crunching some numbers. I’m not pleased with the solution to the homemade Sagarin. More games would lead to a better result. And I see no real value in adding 2011 data to 2012; see how much good that approach will get you, in figuring out how well the Saints have played.

By many measures (median, pythagorean, SRS) Houston is the best team in football.

To explain the columns above, Median is a median point spread, and can be used to get a feel for how good a team is without overly weighting a blowout win or blowout loss. HS is Brian Burke’s Homemade Sagarin, as implemented in Maggie Xiong’s PDL::Stats. Pred is the predicted Pythagorean expectation. The exponent for this measure is fitted to the data set itself. SOS, SRS, and MOV are the simple ranking components, analyzed via this Perl implementation. MOV is margin of victory, or point spread divided by games played. SOS is strength of schedule. SRS is the simple ranking.

Dallas, interestingly, is not the worst of the 2-1 teams by Pythagorean, Philadelphia is. And Tennessee at 1-2 isn’t quite as lively in many metrics as New Orleans. Even at 0-3, the New Orleans Pythagorean is approaching 40%. Tennessee has an anemic 22% by contrast.

The Fifth Down blog features an article about a new phone app, one that says it will give you the winning chances of every play during the game. In the article, we get this little gem:

According to his analysis, a team that returns a kickoff to its 40-yard line can be expected to score an average of 3 more points on the drive than if it had started at the 20-yard line.

“If you make it to the 40, you essentially just made a field goal, even if you don’t realize that immediately,” Bessire said.

I seriously doubt this. 3 points * 100 yards / 20 yards = 15 points. I don’t know of anyone who scales an expected points curve to be worth 15 points. I don’t know of a single reliable EP solution with slopes routinely greater than 0.08 points per yard in between the 20 yard lines.

You pay your money, and you take your chances. Simply put, I can’t recommend this app.

If more folks followed Chris Brown, rather than getting their play design from such “experts” as Michael Lombardi and Charlie Casserly (1), maybe I wouldn’t need to repost a cut from his twitter feed, but under the circumstances, I think it’s the best thing to do.

A good chunk of Baylor’s offense migrated over to the Redskins. That includes some pet packaged plays from the Baylor playbook.

This lack of understanding of packaged plays (see here and here) badly afflicts football fans, and the worst are the ones “who don’t need to read the article” to figure out what Chris is talking about. They think he’s talking about audibles. Sorry, packaged plays are not audibles.

In Dan Graziano’s article, he interviews Shanahan and mostly confuses the issue..

So I asked Shanahan if this had been by design — if he’d set up that first drive with those quick passes to help his rookie get into the rhythm of the game without facing pressure from the Saints’ defense or pressure to go through progressions while he got his feet under him. Because I figured, if it had been, it was a pretty smart idea.

“No, he has options on those plays,” Shanahan said. “He decided to run it that way.”

So how about that, right? Here I was, ready to give the veteran coach credit for a wise game plan that had helped his rookie quarterback ease into his first NFL game, and it turns out it was the rookie quarterback who’d made that decision on his own.

Duh. RG3 doesn’t get the option unless Shanny puts it into the playbook in the first place.

(1) I was actually told by a fan that in order to understand how plays work in the NFL, I needed to stop paying attention to Chris Brown and pay more attention to Charlie Casserly and Michael Lombardi, as those two had forgotten more about play calling than Chris Brown ever knew. I think this says more about the state of affairs in certain elements of football fandom than it does about the relative expertise of these three.

Keith Goldner is active this season both on Advanced NFL Stats and his own blog, Drive By Football. As he has updated his Markov Chain model (see also here), I’d suggest finding Keith’s new articles on either of these two sites. In my opinion, Keith’s work on his expected points models is a must read for anyone who wants to learn analytics, because he’s perhaps the best at making sure that readers can understand how he sets his models up.

Jene Bramel is a good follow if you like in game analysis on Twitter. After the Cowboys 24-17 victory over the Giants, this tweet caught my eye, where Jene mentions a Bear front.

A Bear you say?

I never found that Bear, but at 5:18 in the second quarter – one of the more interesting drives in the game, from the standpoint of a defensive front junkie – we see this:

Two down linemen, but six players at line depth and two at linebacker depth give this front a Bear like feel.

Diagrammatic representation of the front at 5:18, 2nd qtr. Bruce Carter is the linebacker between T and TE.

Though this is formally a nickel front, and there really isn’t a 3-0-3 diamond here, there are a couple things of note. There are six players across the line. Bruce Carter is in the gap between the RDE and the R (rush linebacker), just inside the tight end. Sean Lee is at the 50 behind Bruce (a few yards in front of the left offensive tackle), and another player is in the other 50, a few yards in front of the right offensive tackle. The “lineman” in the two point stace, to the left of the nose guard in this view, isn’t playing a 5 technique as much as he is playing a 3, and the whole front looks as if Rob Ryan is guessing a run to the left side of the line.

That’s exactly what happened. The Giants ran left. Bruce Carter defeated his block and the run gained almost nothing. And it’s almost pure stubbornness to run a running back into the heart of this kind of formation.

Otherwise, I saw plenty of 2 and 3 man fronts, and at one point, perhaps a 4 man front.

After the game, I found that the day of the game, Chase Stuart had this article online, comparing the relative skills of Eli Manning and Tony Romo. And no, it isn’t the usual media fawning exercise.

Update: for more Rob Ryan fronts, this thread has screen shots of the first 10 Ryan fronts of the season.

I recall a time in my early teens when I participated in a judo club, a small one whose regular competition was one of the largest and most proficient judo clubs of its era. Our competition, in a period when no one had a black belt higher than sixth degree, had a fourth degree black belt, and whose club officers tended to be officers at the national level in the United States Judo Association. Having come to my full height, but not my build, I was tall and awkward, and in matches, tended to be paired with younger members of the better club. Those men were smaller, lighter, faster, more skilled and more nimble. I lost a lot of matches before I ever won one.

When I finally won, it was because my opponent was clearly out of balance. He was afraid of my legs, and was leaning into me as we took hold of each other. It was over in seconds. I executed a clean sacrifice throw. I slipped under him, the arch of my right food cut off his left ankle and he rolled across the mat. It wouldn’t have been possible if my opponent hadn’t been out of position.

An ancient “packaged play”. You run if the cornerback follows the flanker. You pass if the cornerback forces the run.

Oversimplifiying a little, skill in judo is being able to recognize what an opponent is doing and using what he is doing against him. In a real match, seldom is it as easy as beating an opponent off the bat. You have to recognize what the opponent is doing in the match and counter appropriately. Counters in football are old, and perhaps the most accessible explanation of an offensive system and the appropriate counter responses is the set “Vince Lombardi on Football.” In it, they mention the halfback option, a single wing play carried over into the strong side offense of Vince Lombardi’s Giants and Packers teams. We’re going through this long introduction to add a nuance to Chris Brown’s phenomenal discussion of packaged plays, which take the essential idea of judo, and also the halfback pass option, and shows how using an opponent’s own reactions against him are now at the forefront of modern football.

Understand, in an traditional option, as in the modern packaged plays, which way the play goes depends on how the play turns out. There are keys and the offense is reading (keying) on the defense. This isn’t the same thing as an audible. When trying to explain packaged plays to folks who have never read Chris’s article, the most common mistake is to think these plays are audibles. An audible is a pre-snap adjustment of a play. A “packaged play” is a play with 2 or 3 options with built in reads after the play begins.

The thing that makes these plays interesting, is the imaginative combination of run and pass options, with passes to isolate and pick on the reactions of specific defenders. To imagine one combination, think of a strong side run combined with a curl in by X in a pro split T, strong right. The quarterback reads the outside linebacker and left cornerback to see which way to go. It’s potentially a way to exploit an overpursuing linebacker.

I find these notions pretty exciting, another rachet up in the cat and mouse game that dominates offense and defense. I can’t help but feel that better play design is a substantial component of the offensive explosion of the past few years. The idea that “players are getting bigger” can’t entirely account for this; they’re getting bigger on both sides of the ball. Game video should benefit both sides. But if we’re in the middle of a new kind of play design, one that can isolate players at will and make their own natural reactions the enemy, then that kind of play design could in fact be “the thing” that’s powering the offenses of the 2010s.


Get every new post delivered to your Inbox.

Join 243 other followers