Keeping the Defense Off the Field

Not statistically significant? Let's see some t-statistics and P-values.

Nice work, Dan. Now get back to finishing your dissertation.

Well written piece that makes the best of an imperfect data set.

The home team is down by a touchdown with eight minutes left in the fourth quarter. For the second drive in a row, its offense sputters and punts after a three-and-out. The defense gets back on the field, beaten, broken, and exhausted. If the offense had been able to sustain drives, the defense might have had the energy to prevent the opposing offense from scoring another touchdown, effectively ending the game.

While that's the narrative that's usually used, it sounds more like the defense is going to be facing some very short fields. By their nature, will result in more scores against, no matter what condition the defense is in.

I'm concerned that we've failed to control for a factor. Teams have limited spending money and talent-attracting-ability. Only elite teams have good defenses and good offenses. There may be a counter-trend, to whatever limited extent that the better your defense, the worse your offense, for the sheer reason that most teams couldn't afford offering awesome contracts in both areas.

Is it possible to compare DVOA of defenses that have run more than a certain number of plays? Say compare the DVOA of teams before and after the 50 play threshold.

The data set would be smaller but you would be looking more directly at the issue. (Maybe, I think. My statistics classes were more than a decade ago.)

Good article though.

I've never understood why defenses get more tired the longer they're on the field, but the offenses purr right along. Is it really more tiring to play defense? Has it been demonstrated?

"One might expect that this has the biggest impact on performance during the fourth quarter. I understand this view, but I doubt that the effect is much bigger in the fourth quarter than it is overall."

This is exactly what I was thinking in the first couple paragraphs. I wish you'd gone into further detail here instead of just dismissing it offhand. It may be that your results would confirm your hypothesis. If so, good, more conventional wisdom debunked. But I think that assuming that 25% of the data would automatically not be overshadowed by 75% sells the story just a little bit short. I'd love to see that data rather than just accepting your assumption.

Otherwise, it's great stuff.

Wait so are you comparing a team's performance against their own performance when the offense is doing better?

Or were you just checking if teams that run longer drives have better or worse defenses than average teams without accounting for anything else?

I have a problem with this article, it doesn't account for differing defensive strengths. I don't think anyone is suggesting that extra rest makes bad defenses good. Just that it allows the defense to play (closer) to it's peak potential. I'd think you would have to look at the same defense at different rates of offensive plays per drive and see how it does.

One of the things I did last season was plot how a defense declined when facing the Oregon offense and noted that when Oregon was able to run 50 plays by around the second half, they started becoming significantly more explosive. My gut suspicion is that the time that a defense gets to 'rest' isn't significant compared to how hard they do end up working when they're on the field, especially in the NFL with TV timeouts, no real hurry up offenses and generous substitution policies.

I'd be curious if there was any statistical correlation being able to be seen given number of plays run by the opposing offense and defensive performance.

First, while defensive performance affects the number of plays that an offense runs per game, it is not likely that defensive performance affects the number of plays an offense runs per drive. Second, performance of the defense affects the average starting line of scrimmage for the offense, and therefore the number of plays that an offense runs per drive.

Huh?

I have both anecdotal and design issues with this piece. Anecdotally, watch the 2011 Ravens-Texans game and watch Jarrett Johnson basically fall over with exhaustion while plodding after the QB and you'll know what I mean.

From a design standpoint, I think you're looking at the wrong thing. You should see if teams with shorter offensive drives (or more defensive snaps, or both) have a greater DVOA differential between the 1st Quarter (or first half, or first 3 quarters) and the 4th Quarter. You might even need to factor in age, and/or number of defensive linemen in the rotation, but you could definitely see it happen in Ravens games this year. Perhaps that was an isolated event and not a greater trend, but it was definitely there.

More fundamentally, all we've really shown is that DVOA doesn't correlate to offensive plays per drive. The underlying premise, though, is rest time. We have no analysis of whether offensive plays per drive correlates with elapsed time.

As an example, Indy runs 5 plays in a far different amount of elapsed time than San Francisco does.

I'm with Aaron Brook's Good Twin and dcaslin. This study is a nice start at something, but it doesn't really measure what is usually meant by the cliche that defenses get tired later in games.

The other two posts cover most of my reservations, but I'll some more:

First, I wonder how well plays per drive can really measure the sort of offensive dominance that results in a "tired defense" -- offenses can run as many as 15 quick passing plays in a 'football minute.' Not only would a sequence of such plays fail to tire the defense, it might even give the defense some emotional strength as they get to slam receivers running short routes over and over again. But a series of 4 or 5 hard-nosed running plays, or passes in which the QB escapes to extend the play (and the time the pass rushers spend chasing) could blow the wind out of most defensive players.

Second, it seems to me that the proverbial situation almost always has some element of the offense beating the cr*p out of the defense in the old-fashioned smash mouth way.

It might be that the change in design these considerations suggest (Time of possession, time per play, actual time vs. football time, run vs. pass frequency, etc.) will force the ivestigation of data that is too picky to make semse of. But from what see presented right now, I'm not willign to draw a conclusion either way.

"First, while defensive performance affects the number of plays that an offense runs per game, it is not likely that defensive performance affects the number of plays an offense runs per drive."

Unless you're just assuming the results of your analysis I don't see why not. It's just the flip side of the "defense gets tired argument." The better your defense the fewer plays the opponent's team offense runs, meaning their defense is more tired when they get on the field.

Anyway, somebody mentioned above the obvious counter to this analysis, which is that teams that have above average offenses are going to tend to have below average defenses given the salary cap and limited resources teams can devote to their team.

I think there may be a bit of a flaw when considering what is to be measured. Instead of looking at simple statistics that measure time of possession and number of plays, actually do the hard work of measuring the amount of time for each individual play. Most plays aren't that long whatsover. In fact, from snap to whistle/tocuhdown/out of bounds etc for most plays is very very short... a long play might last 8 seconds in normative play. Special teams is a little different but it depends on the special team play, a kickoff play is probably longer than a field goal attempt obviously. Regardless, the measure of how many seconds a defense is actually doing work in a play should be utilized instead of simple math with bad measurements.

When you add up the total amount of plays on Defense in seconds of time you'll begin to wonder if anyone really can get tired for only 6-8 min of actual real hard physical exertion/work a game. Imagine, millions of dollars for 8 minutes of work a week (8 min is generous for one side of the ball). Then think, wow, 12-14... maybe 16 minutes of actual play that takes 3-4 hours to watch. Big implications here. Regardless, I think this study should have really considered how much work actually can be compacted into a 6 second play... and why it really just doesn't matter because there aren't enough of them for very long. DO I see an argument for more than 18 games a season?!?!

With that criticism of the measurement used I would say that we'd probably get similar results debunking "conventional wisdom."

As others have mentioned, nobody cares how resting affects performance in the first half since of course there is going to be little correlation. It's the fourth quarter that matters, and the fact is that the offense scores more in the fourth quarter than in the first because the offense doesn't have to work as hard as the defense on each play.

Yea, I never bought into the whole rested defense makes a big difference in improved D play. But I think a good defense that can help give the team a lead makes a huge difference in allowing the offense to play well. Got any stats on that?

Honestly, this is one of those studies where you have to step back from the statistics and see if the results actually make sense. I can clearly remember games where the Bucs Cover-2 defense was exhausted after the other team ran 70+ plays. Defensive players have to react to the offense, therefore they expend more energy on a given down than the offense; so the offense resting at the same time the defense rests isn't a 1:1 ratio.

If you've ever sat in the stands in Lambeau in December or January, you'll know that the defense gets plenty of opportunities to rest during those endless TV timeouts and halftime. With the constant subpackages swapping players, few front 7 defenders play the entire game. Zone defenses limit the amount of continuous sprinting by DBs. I've never bought into the tired defense theme. Perhaps later in the game, the offensive coordinator has set up his offensive scheme, the offensive players have found a rhythm.

to the lesser team's offense against the better team's defense?