Stat Analysis
Advanced analytics on player and team performance

FO Mailbag: Does Defense Dip Down in December?

by Bryan Knowles

NFL seasons are long. Even though the sheer number of games an NFL team faces is nowhere near the slate of any of the major sports, most other sports don't involve slamming into people 40 to 60 times a game over a period of four months. That wear and tear is bound to have some sort of effect as the season goes along, right?

That's the basic question asked by reader Rocco C on twitter:

While it's a common cliché to talk about a defense being tired at the end of the year, we've never actually gone back and looked through the data to see how defenses actually change throughout the year. Zach Binney's work has shown that linemen tend to have fewer reportable injuries than "mobile" positions like defensive backs and wide receivers. However, it would at least make a degree of sense if the accumulated stress of playing in the trenches would gradually wear down defensive linemen over the course of a season. There is no box on the injury report form for "just really tired and bruised," but NFL bodies and minds are healthier in September than they are in January. I mean, that would logically check out, right? And that's something we can definitely take a look at.

The following table lists total defensive DVOA, pass defense DVOA, and rush defense DVOA over the past six seasons. Since 2012, we're looking at:

  • the average DVOA over the course of the entire season;
  • the average DVOA over the first five weeks of the season;
  • the average DVOA over the first 12 weeks of the season;
  • and the average DVOA for Weeks 13 to 17.

Those last five weeks roughly correspond to December and January, the critical final stretch of the season where we would expect tiredness and fatigue to set in. The difference column is the difference in DVOA between those last five weeks and the rest of the season.

Remember, lower DVOA means better defense.

Defensive Splits, 2012-2017
Defense Overall Weeks 1-5 Weeks 1-12 Weeks 13-17 Diff
Def. DVOA 0.0% 1.0% 0.6% -2.2% -2.8%
Pass Def. 6.3% 8.9% 7.8% 2.7% -5.1%
Rush Def. -7.9% -8.8% -8.7% -6.4% +2.3%

December Drop-Offs

While the data is definitely noisy -- random scheduling variation and injuries will definitely throw a spanner into the works of an apples-to-apples comparison -- a couple of interesting things pop out.

First of all, rush defense does appear to decline in the last five weeks of a season. This is true in total, and in five of the six individual seasons we studied. Average run defense DVOA actually increased in December of 2014, but all defense had an unusually big boost in December 2014 -- pass defense increased by 7.9% that year.

The first ten weeks of the season see a slight bump in run defense DVOA -- it rises from about -10.0% to -7.0%, bottoming out in late October, before improving back to around -10% by Veterans Day. From there, however, run defense begins a steady nosedive, with DVOA approaching -5.0% over the course of three or four weeks and sticking there for the rest of the season. It happens every year in much the same pattern.

If this were a case of fatigue and wear and tear building up over the course of the season, we would expect rush defense to get worse and worse each week. Instead, it's a pretty stark drop-off, starting somewhere between Weeks 10 and 12 depending on the season. Stick a pin in that; we'll get back to that in a moment because there's an interesting explanation.

Passing defense, on the other hand, actually gets better as the year goes along. I found this counter-intuitive; it seems to be the general accepted wisdom that defenses have an advantage over offenses for the first few weeks of the season, as a lack of practice time stops chemistry from developing between a quarterback and his receivers. That would imply that pass defense should get worse as the season goes along, but the reality appears to be quite the opposite.

It's not a perfectly smooth improvement throughout the year. Pass defense gets notably better throughout September, then sort of hits an even keel around 6.0% or 7.0% until that same mid-November transition point we saw in rush defense. Somewhere between Weeks 10 and 12, depending on the season, pass defense notably improves, with DVOAs hitting about 3.0% and staying there.

The following graph shows the average rush and pass defensive DVOA for each week of the season from 2012 to 2017. Because that data is very, very noisy, we've also included a "smoothed" DVOA -- a five-week average, centered on the week in question, to better show overarching trends. Pay more attention to that, which has many more data points, than the individual week-by-week lines, which can be skewed strangely by a few outliers.

(No, I have no idea what's going on with passing in Week 9.)

So, that seems to answer that, right? Run defense gets worse and passing defense gets better at the end of the season. Since that's the period of time when defenders are the most worn out after a long season, that would explain the drop-off, right?

Well, maybe not.

The Weather Outside is Frightful

If the drop-off in run defense was an issue of rest and tiredness, we would expect to see this same drop-off be present throughout the league. Yes, maybe one or two teams would avoid the trend due to random variation or perhaps age-related issues, but on the whole, we'd expect to see teams react pretty much the same.

That is not the case. When we split out the data for all 32 teams, some trends become apparent.

This table shows the Weeks 1-12 and Weeks 13-17 splits for each team in both run and pass defense. We also have each team listed by weather category -- Cold, Warm, Mid, and Dome -- based on where they play their home games in December. We mostly use this for special teams work, but it's actually fairly crucial to understanding what's going on here. Teams are sorted by change in late-season run defense DVOA, from the team that improved the most (Seattle) to that which suffered the sharpest decline (Chicago):

Defensive DVOA Splits, 2012-2017
Team Run Defense DVOA Pass Defense DVOA Weather
Weeks 1-12 Weeks 13-17 Diff Rk Weeks 1-12 Weeks 13-17 Diff Rk
SEA -14.6% -25.9% -11.4% 1 -7.3% -15.8% -8.5% 10 Mid
LAR -13.2% -20.1% -6.9% 2 2.7% -5.6% -8.3% 11 Dome/Warm
TEN -2.8% -9.1% -6.3% 3 15.8% 10.6% -5.2% 17 Mid
LAC -1.4% -6.5% -5.1% 4 13.3% 3.7% -9.6% 7 Warm
ARI -17.9% -22.4% -4.5% 5 -10.1% -4.0% +6.1% 30 Dome
NE -8.5% -12.5% -4.1% 6 10.8% 2.5% -8.3% 12 Cold
HOU -10.3% -13.8% -3.4% 7 2.8% -1.3% -4.2% 20 Dome
OAK -5.9% -9.0% -3.0% 8 23.0% 12.3% -10.7% 5 Mid
IND -1.7% -4.4% -2.7% 9 17.6% 2.4% -15.1% 1 Dome
SF -7.6% -9.4% -1.8% 10 9.1% 10.1% +1.0% 26 Mid
MIN -9.6% -11.1% -1.5% 11 7.8% 3.4% -4.4% 19 Dome
ATL -0.8% -2.3% -1.5% 12 16.9% 6.7% -10.2% 6 Dome
JAC -4.9% -5.6% -0.7% 13 8.6% 9.7% +1.1% 27 Warm
NYG -9.5% -9.6% -0.1% 14 7.2% 3.8% -3.4% 22 Cold
CIN -5.6% -4.8% +0.8% 15 2.3% -10.4% -12.7% 3 Cold
NYJ -19.9% -17.7% +2.2% 16 11.7% 7.6% -4.1% 21 Cold
Team Weeks 1-12 Weeks 13-17 Diff Rk Weeks 1-12 Weeks 13-17 Diff Rk Weather
NO -0.5% 2.6% +3.2% 17 16.4% 13.1% -3.3% 23 Dome
PIT -11.3% -7.4% +3.8% 18 8.0% 7.6% -0.3% 24 Cold
KC -4.2% 0.4% +4.7% 19 6.5% 0.3% -6.2% 16 Cold
DAL -5.4% -0.2% +5.2% 20 16.6% 7.6% -9.0% 9 Dome
MIA -7.9% -2.7% +5.3% 21 9.9% 15.5% +5.6% 29 Warm
CLE -5.6% -0.1% +5.5% 22 16.7% 9.8% -6.9% 15 Cold
PHI -14.0% -8.2% +5.8% 23 3.4% 11.1% +7.7% 32 Cold
DET -12.5% -6.3% +6.2% 24 13.1% 13.0% -0.2% 25 Dome
CAR -15.5% -9.2% +6.4% 25 -2.9% -17.0% -14.2% 2 Mid
BUF -2.7% 4.8% +7.5% 26 2.6% -8.2% -10.8% 4 Cold
DEN -19.5% -11.9% +7.6% 27 -6.8% -16.4% -9.6% 8 Cold
TB -14.7% -7.0% +7.7% 28 13.8% 6.3% -7.5% 13 Warm
WAS -7.8% 0.7% +8.5% 29 12.2% 5.1% -7.1% 14 Cold
GB -9.8% 2.4% +12.2% 30 9.6% 4.8% -4.8% 18 Cold
BAL -18.4% -5.0% +13.4% 31 0.0% 3.4% +3.4% 28 Cold
CHI -7.8% 6.1% +14.0% 32 2.0% 9.6% +7.6% 31 Cold

Team Weeks 1-12 Weeks 13-17 Diff Rk Weeks 1-12 Weeks 13-17 Diff Rk Weather
NFL -8.7% -6.4% +2.3% -- 7.8% 2.7% -5.1% -- --
Cold -10.3% -4.5% +5.8% -- 6.1% 2.2% -3.9% -- --
Others -8.2% -9.0% -0.8% -- 9.3% 3.9% -5.4% -- --

It's worth noting that the sample sizes here get smaller. About 45 percent of games take place in cold stadiums, 30 percent take place in mid or warm stadiums, and the remaining 25 percent take place in domes with the roofs closed. In addition, don't read too much into defenses being better or worse in one environment or another. This isn't like home and road splits, where every team plays an equal number of games in each box; the quality of defenses in each group is not random. We're more interested in the rates of change rather than the raw numbers here.

As expected, we see that the drop-off in run defense is primarily located in cold-weather stadiums. Other outdoor stadiums actually see a slight improvement in run defense in December, though that's well within the noise margin. It's not just teams who live and train in cold weather getting worse; it's everyone who shows up to play in frigid conditions.

And, as expected, we see all three groups improve their pass defense. Interestingly, there does seem to be a more significant improvement in mid- and warm-weather stadiums. My gut feeling there is it's a case of "splits happen," as it doesn't really hold up on the individual team level, but that's something worth keeping an eye on as we go forward.

A couple thousand words later, and we can finally cycle back to the original question. Yes, it does appear that run defense gets worse and passing defense gets better as the year goes along. Blaming it on players wearing down over the course of the year, however, appears to be a case of correlation, not causation. The weather seems to be a clear factor for poor rush defenses late in the year, above and beyond any issue of injuries or exhaustion.

I'll leave you, then, with a discussion question. If run defense gets worse in the cold, does it behoove a team who plays plenty of December games in the cold to have a higher investment in running backs? Does it make more sense for the Steelers to pay Le'Veon Bell than it does for the Rams to pay Todd Gurley, because the Steelers play in a division with four cold-weather teams? Does a Saquon Barkley make more sense in New York than he would have in Indianapolis and a division filled with warm cities and domes? I don't think these results come close to outweighing all the evidence we have warning teams against over-valuing running backs, but it is, at least, another interesting point in the discussion.


14 comments, Last at 14 Aug 2018, 12:54pm

3 Re: FO Mailbag: Does Defense Dip Down in December?

Also curious about the mid-season spike, but my suspicion is that there are 3 or 4 ridiculous offensive explosions that all just happened to hit in week 9 skewing the results. Bryan could check that by converting the time-series plot to a violin plot, or just creating a table of the top 4 games in each week from 2012-2017 and then manually check to see if there's anything fishy.

5 Re: FO Mailbag: Does Defense Dip Down in December?

Do defenders on losing teams care in the final weeks of the season? Defense requires motivation, offense is fun. As soon as a game isn't close I'm guessing they give up on some level. Not overtly but in a game of inches you might not hang onto a tackle or chase down a ballcarrier with the same effort that you might have done earlier.

And possibly even true for the teams that have locked up playoff spots and seeding.

8 Re: FO Mailbag: Does Defense Dip Down in December?

somwetimes will see that in fg ames. like 2-10 tema gets down 27-3 in ealrly 4th quarter. next thong you knwo safety rtakes a "weird:" angle and cornerbakc makes quesitonable tackle effort on 64-yard touchdown run by runjibng back. final score ends up being 41-3.

do not notice it too often where there will be bad effort all game long like that but do notice it sometimes with basd teams being beat bad in a gievn game

6 Re: FO Mailbag: Does Defense Dip Down in December?

It could also be a matter of run offenses tightening the screws as the season goes by. It's easy to work on the passing game in practice, since 7-on-7s and other passing drills don't require any hitting. On the other hand, run blocking requires intense coordination between 5 to 8 blockers, and there's only so much this can be practiced in limited contact environments like training camp.

9 Re: FO Mailbag: Does Defense Dip Down in December?

My 2 cents:
1. Since traction plays a part in cold weather, the offense has a slight advantage because it knows where it's going.
2. Passing is harder in cold weather because of the wind as much as anything, making pass "defense" better--not because the defense is better, but because the offense is "worse"--i.e., hampered because of the weather conditions. This correlates to...
3. More rushing plays are run overall in colder weather b/c of bad passing conditions. More rushing attempts can be split among different backs while the others rest. While individual defenders can rest, no NFL team has a full second-team defense that can be subbed out like that individual RB. Which leads to...
4. As the season wears on, depth is tested for most teams. On offense you can hide a weaker player somewhat--whereas on defense, the offense will do its best to attack that weak link. IMO, because rushing defense is about playing your gap, there can be confusion on a play here or there that allows for better rushing overall.
The counterpoint is that many offenses will try to target a weak-link DB, or avoid the great DB. However, if passing overall is harder b/c of the weather conditions, this helps that DB more than the great one--the great one isn't getting targeted regardless of the weather.
Anyway, just my thoughts on all of this.

11 Re: FO Mailbag: Does Defense Dip Down in December?

Regarding the splits by "weather", it might be better to compare home DVOAs. That would help separate two different effects:

  • Dome/Warm teams face less cold/windy weather in December, because at least half their games are indoors/warm.
  • Dome/Warm teams are more affected by cold/windy weather, because they have less opportunity to practice for it.

Looking at home games in isolation would let you say things like:

  • Cold stadium weather in December has less/more effect than warm stadium weather in December on running/passing DVOA.

A more advanced analysis would look at Dome teams playing away games in Cold stadiums in December vs. other months, etc. But you'll have small sample sizes for that, so it might not give a clear signal.

12 Re: FO Mailbag: Does Defense Dip Down in December?

Yeah, you get into a problem of tiny sample sizes soon enough which is an issue. I also wish I could have adjusted more for schedule -- perhaps, over the past few years, good rushing teams have been scheduled to play in cold weather stadiums in December more often than not. I mean, I doubt it, but I can't rule it out 100% yet.

This is more the data being highly suggestive than any solid proof at this point, but it was interesting enough to share.

13 Re: FO Mailbag: Does Defense Dip Down in December?

It makes sense that Seattle would be on top (or least affected) due to weather in run D in late season. Late season scheduling is heavy on division games and Arizona, LA and SF are all dome or warm.

Baltimore plays late season in the arctic.

14 Re: FO Mailbag: Does Defense Dip Down in December?

Another reason that pass D DVOA might improve in the last 5-6 weeks of the season is due to the sheer # of QB injuries. Every year it seems that a third of the QBs starting towards the end of the year were initially backups or 3rd stringers when the year began. Also, some teams decide to hand the reigns to the promising rookie towards the end of the year, with predictable results most of the time. This isn't enough to make a huge difference, but it doesn't seem that there is a huge difference in D DVOA, just a modest one.