Writers of Pro Football Prospectus 2008

19 Oct 2012

The Meaning of First Quarter Offense

So, Kevin Pelton of Basketball Prospectus sent me an e-mail this week asking about the relative strengths and weaknesses of his hometown Seattle Seahawks. Kevin seemed to notice that, even after accounting for last week's big comeback against New England, the Seahawks seem to always play better on offense at the beginning of the game. The Seahawks are one of those teams that like to script their first 15 plays; could they just be better on those plays than they are afterwards?

I didn't want to specifically pick out the first 15 plays to look at, since of course the first scripted 15 plays aren't necessarily the first 15 actual plays. Sometimes you have to go off the script for a short-yardage or red-zone opportunity. However, it is clear that the Seahawks offense is better in the first quarter this season. Through Week 6, the Seahawks rank eighth in offensive DVOA in Q1, then 27th, 26th, and 17th in the next three quarters. I don't have numbers broken down yet for last night, but the Seahawks did have two field-goal drives in the first 18 minutes and then went scoreless the rest of the game.

So, will the Seahawks continue to be better on offense in the first quarter for the rest of the season? Actually, that looks more likely than you might think.

At some point last year, we looked and noticed that there's nothing particularly consistent about having a better DVOA in the fourth quarter. (I did this same thing for quarterbacks specifically in ESPN The Magazine a couple years back.) That may not be true about the first quarter. Just to look quickly, I went back and looked at Weeks 1-9 compared to Weeks 10-17 over the past five seasons. In each of those seasons, the 1H-to-2H correlation of Q1 offensive DVOA was higher than any of the other quarters. For the five years in total, the correlation was .43 for Q1 offense, compared to .31 for Q2, .30 for Q3, and .31 for Q4.

Of course, there's also a question about what this might mean from year to year. The Seahawks were scripting their first 15 plays last year too, yet they ranked 24th in Q1 offense but eighth in offense after halftime. Yet another example of how looking for meaningful splits in the NFL can get tricky.

By the way, this whole thing was absolutely not a reason to slip in a mention that the excellent Pro Basketball Prospectus annual by Kevin and Bradford Doolittle is now available for the low price of $9.98. That's just an extra bonus. If you like basketball and you like FO, you'll like PBP.

Posted by: Aaron Schatz on 19 Oct 2012

23 comments, Last at 20 Oct 2012, 5:27pm by nat

Comments

1
by nat :: Fri, 10/19/2012 - 1:13pm

If scripting were a special skill, you'd expect Q1 offensive DVOA to be less predictive of future total offensive DVOA than Q2, Q3, or Q4.

Similarly, if one of the other quarters required a special skill, it might be less predictive, too.

It would be interesting to see the season-to-season correlation of each quarter's DVOA to itself next year and each quarter's DVOA to total DVOA next year, for offense and defense to be complete.

Could each quarter require a specific skill? Certainly.

Q1: scripted offense
Q2: unscripted offense, clock management
Q3: half-time adjustments, and reaction to them
Q4: clock management, risk management

So, Aaron, how do the quarters stack up for being predictive? Is that something you can check for us?

2
by tuluse :: Fri, 10/19/2012 - 1:16pm

If scripting were a special skill, you'd expect Q1 offensive DVOA to be less predictive of future total offensive DVOA than Q2, Q3, or Q4.

I'm not sure that's true because teams don't stop scripting plays in future games, and who knows maybe they start scripting more if they realize it's working better.

7
by nat :: Fri, 10/19/2012 - 3:07pm

If a team were unusually good or bad at scripting, their first quarter offensive DVOA would drift away from their average DVOA, both present and future. That's what being unusually good or bad at scripting means.

If all teams were as skilled at scripting as they were at running their usual offense (relative to the league), it's not statistically speaking a separate skill from running their usual offense.

3
by Jeff M. (not verified) :: Fri, 10/19/2012 - 1:50pm

My first thought was that this was pretty weird for the Seahawks to be systematically better in the 1st. They're a dramatically better rushing than passing team (11th in DVOA vs. 20th, 3rd in ALY vs. 21st in ASR, Lynch is clearly the best offensive player by DVOA/DYAR), and the thing they're good at (power running) is supposed to get better later in games (at least that's the CW--I don't know that I've ever seen good analysis of the numbers).

Then I remembered that the Cowboys game is the only one of seven where the Seahawks held a lead through the middle portion of the game. I'd guess that what explains this is simply difference between DVOA when ahead/trailing and/or between run/pass. I'd like to see the breakdown, but I bet each of those breakdowns is less variable by quarter, but that the Seahawks spend more time ahead/running (high DVOA) in the 1st and behind/passing (low DVOA) in the remaining three quarters.

I don't know how that speaks to the more systematic result across teams, though. Are there clear differences in quarter-by-quarter DVOA for the league as a whole (or just in predictiveness of that DVOA...)? You might expect that all teams can do what they do best on offense in the 1st but may be forced by game circumstances to a weaker are later in the game...

6
by Anonmouse (not verified) :: Fri, 10/19/2012 - 2:03pm

My own two eyes tell me that the Patriot's 4th Q DVOA is much worse than the first 3 Qs. Both offensively and defensively.
I'd love to see the Pat's offensive DVOA ratings split by hurry up and not.
A little love for the home town fans Aaron?

4
by DavidL :: Fri, 10/19/2012 - 1:53pm

Of course, there's also a question about what this might mean from year to year. The Seahawks were scripting their first 15 plays last year too, yet they ranked 24th in Q1 offense but eighth in offense after halftime. Yet

You could argue that they're just better at it this year.

5
by tuluse :: Fri, 10/19/2012 - 1:59pm

I think the more obvious answer would be, rookie QB is better at running the scripted plays he's been drilled in all week than the rest of the playbook.

15
by zenbitz :: Fri, 10/19/2012 - 10:15pm

Or even the defense isn't exactly sure how to play the rookie QB until they see a dozen plays.

8
by DEW (not verified) :: Fri, 10/19/2012 - 3:17pm

I've always wondered how this "play scripting" actually worked; given how sensitive to down and distance situations an NFL offense is ("I don't care if it's 3rd-and-17; the script says we run a dive play out of the I and that's what we're gonna do!"), not to mention the variety of adjustments at the line in terms of protection changes and outright audible calls, what does a script really consist of?

9
by Pat (filler) (not verified) :: Fri, 10/19/2012 - 3:28pm

I'd have to imagine scripting mostly consists of first and second down situations.

You do have to realize though, that early on in the game, you can gain a lot more value out of figuring out what a defense is going to do and capable of than success or failure on an individual down is worth. I mean, on your 3rd and 17 example, who cares if you run a dive, or attempt a pass? Both are pretty damn unlikely to succeed, whereas if you're trying to figure out how the opposing line handles a certain kind of blocking, figuring that out may be way more important than the extremely marginal increase in the likelihood of the drive continuing.

12
by Alex51 :: Fri, 10/19/2012 - 6:47pm

I mean, on your 3rd and 17 example, who cares if you run a dive, or attempt a pass? Both are pretty damn unlikely to succeed, whereas if you're trying to figure out how the opposing line handles a certain kind of blocking, figuring that out may be way more important than the extremely marginal increase in the likelihood of the drive continuing.

Yeah, but on 3rd and 17, the defense is probably going to be in dime, and they definitely won't be playing the run, so the information you get from running that dive probably won't transfer all that well to the rest of the game.

10
by tuluse :: Fri, 10/19/2012 - 3:32pm

There was an extra point that had scans of a Bill Walsh script at one point.

I don't remember it perfectly, but they didn't follow it to a tee, and deviated for certain down and distance situations. It designed to probe the defense in certain ways. I remember one play specifically was a deep pass intended to test one of the safeties.

13
by Jerry :: Fri, 10/19/2012 - 6:51pm

The idea of scripting (as both Pat and tuluse touched on) is to see how the defense reacts to certain offensive plays. There may be a couple of plays that fit 3rd-and-long, or they may go off-script entirely in those cases, but the script isn't really designed to march the team down the field on the first drive. It's supposed to inform the playcalling for the rest of the game.

16
by the cat in the box is dead (not verified) :: Sat, 10/20/2012 - 5:00am

That's my understanding of it too, so you should in theory see a variety of strategies tried out in those scripted 15, and then a team moving to what worked afterwards. I'd almost expect a scripted team to be more hit-and-miss during that first 15 than after.

17
by the cat in the box is dead (not verified) :: Sat, 10/20/2012 - 5:00am

That's my understanding of it too, so you should in theory see a variety of strategies tried out in those scripted 15, and then a team moving to what worked afterwards. I'd almost expect a scripted team to be more hit-and-miss during that first 15 than after.

18
by Jimmy :: Sat, 10/20/2012 - 9:44am

I remember a Walsh quote where he said that you only have two drives to run the scripted plays as game requirements begin to be more important than trying to use the rest of the scripted plays. This made choosing plays that the team executed well vital to trying to both run the script and by inference learn more about the opposition.

11
by The Ninjalectual :: Fri, 10/19/2012 - 6:22pm

Does "8th in Q1 DVOA" mean "if the offense only played in the 1st Q, it would rank 8th, league-wide?" Or does it mean "this Q1 offense is 8th as compared to every other team's Q1 offense?"

14
by MJK :: Fri, 10/19/2012 - 8:13pm

Easier explanation here, I think. Defenses likely get better at adapting to offenses as the game goes on.

If the opposing defense is good and prepared for your offense, you'll get stopped a lot early on and there won't be much data to sway DVOA. If the opposing defense isn't prepared for what you run, you'll have a lot of early success...but then they will adapt and youre success will drop in subsequent quarters. And if this happens, it will tend to skew your first quarter numbers up.

19
by andrew :: Sat, 10/20/2012 - 10:30am

So I guess the next question is how does this compare across the league? is there data on which teams script and don't?

what about the defenses they faced, are they generally better early or late?

20
by Marko :: Sat, 10/20/2012 - 2:07pm

I don't know if the Bears script their plays, but if they do, they need to burn their script. Their first quarter offense has been somewhere between abysmal and atrocious.

However, their offense has been really good in the second half. Several Chicago media outlets have pointed out how much better Jay Cutler and the offense as a whole performs when the team has opened up a lead. In the last three games, the Bears have had small leads until a defensive TD helps them break open the game, followed by the offense starting to click (and in the last two games, another defensive TD). Part of this is due to big plays off play action, which is successful once the Bears get the running game going after opening up the lead.

21
by nat :: Sat, 10/20/2012 - 2:33pm

Funny thing. If the purpose of scripting the first drives is to gather useful information about the opposing defense, then a team that was unusually good at scripting should have a lower Q1 DVOA than the rest of its quarters. Q2-Q4 should be improved by the information gathered by scripting. Q1 doesn't get that boost.

So "has its worst DVOA in first quarter" may translate into "good at learning from scripted plays".

And "has its best DVOA in Q1" means "learns less than other teams from scripted plays".

22
by Marko :: Sat, 10/20/2012 - 4:01pm

That all may be true, but it sounds like you haven't seen much of the Bears games. It's one thing to gather useful information via a play that may not be successful, but is not harmful. It's another thing to have plays that completely blow up in your face and be potentially disastrous.

I'm mostly talking about the Bears' first two drives. In the first game, the Bears had the ball at around their 20 yard line. The first play was a dropback by Cutler which resulted in a sack for something like a loss of 13 yards. They ended up going three and out (there was at least one penalty mixed in there, I think for a false start) for something like -8 yards, setting up the Colts in good field position after the punt. The defense held, but the Colts downed their punt inside the Bears' 5 yard line. The first play was a swing pass to Forte that was easily picked off by a Colts LB, who waltzed in for a 4 yard TD return. Later in the quarter (after another punt), the Bears' finally got their offense going and had a long TD drive that was scored early in the second quarter.

In the next game at Green Bay, again the first play from scrimmage by the Bears was a sack of Cutler for a big loss of yardage. So again,, they put themselves in a big early hole. They never really got the offense going in that game at all.

In the third game, the fans cheered when the first play did not result in a sack of Cutler, as the Bears finally learned to start off with a running play. The first drive ended in a field goal (that's the only points the Bears have socred on their opening drive), but was greatly aided by a roughing the punter penalty, so the offense had been stopped before the Rams gave the Bears a huge gift. I think there was another big defensive penalty on that drive.

The fourth game wasn't disastrous, but the offense didn't do much.

In the fifth game, the first two plays were runs that gained 9 yards. Then on third and 1, Cutler had his first pass attempt picked off at around the Bears' 46 yard line. The defense bailed out the offense and forced a three and out.

23
by nat :: Sat, 10/20/2012 - 5:27pm

The other way your Q1 DVOA can be low is to suck in the first quarter. But that has nothing to do with scripting. So I think I'm agreeing with you.