One weird statistic to summarize what went wrong in 2021

In all the discussion about what went wrong for the 2021 Seahawks to finish 7-10, I think I can summarize many of the problems in a single, striking statistic. I haven't seen any other articles make the point that I'm about to make, so I decided to write my own.

Football teams tend to score more points and allow fewer when their offense is on the field than when their defense is. That's incredibly obvious. But does having your offense on the field for more plays and your defense for fewer tend to mean that you score more points and allow fewer? That's an empirical question, so let's have a look at the data.

First of all, I want to mostly focus on play difference and point difference rather than raw numbers of plays or points. The idea is that there isn't anything intrinsically better or worse about tending to snap the ball with 15 seconds left on the play clock rather than 5, but the former will tend to mean that a team plays more snaps per game, both on offense and defense. I want to exclude play tempo and just focus on whether it's predominantly the offense or defense on the field.

So let's define some terms. Your play difference is the number of plays that your offense is on the field minus the number for your defense. Special teams plays are excluded. Your point difference is your number of points scored minus your number of points allowed. I'm unable to exclude points scored by the defense or kick or punt returns, so I'm just using total points scored and allowed. The league wide average for play difference and point difference is always zero, for fairly obvious reasons. So the real question is, does a high play difference tend to be correlated with a high point difference?

Empirically, the answer to that is yes. The correlation between them last year was 0.445. It was 0.414 the year before, 0.447 three years ago, and 0.465 four years ago. The correlation is weaker if you go back further, but is 0.343 over the course of the last decade.

One might expect that keeping the offense on the field more is correlated with success. The league average is about 6.0 plays per drive. Three and outs are rather fewer than that. While very long, quick strike touchdowns are possible, getting a touchdown in one or two plays is rare. Drives that end in touchdowns or even field goals often take quite a few plays to get there.

This might seem like a strange thing to focus on until you look at the 2021 stats for it. Here's the top 10:

Team                 play  point
Baltimore Ravens +152 -5
Buffalo Bills +129 +194
Arizona Cardinals +98 +83
Tennessee Titans +94 +65
Carolina Panthers +78 -100
Kansas City Chiefs +74 +116
Dallas Cowboys +71 +172
Chicago Bears +66 -96
Tampa Bay Buccaneers +46 +158
Green Bay Packers +35 +79

The top 10 in play difference includes 7 of the 8 teams to win at least 11 games last year, as well as 7 of the top 11 in points difference. It also includes the Carolina Panthers (who went 5-12) and the Chicago Bears (6-11), who both had substantially negative point differences. It's not perfectly correlated with success.

But let's look at the other end. Here's the bottom five:

Team                 play  point
New York Giants -87 -158
Houston Texans -91 -172
Atlanta Falcons -108 -146
New York Jets -109 -194
Seattle Seahawks -247 +29

The bottom 5 in play differential includes 4 of the bottom 5 in point differential, as well as 3 of the 5 teams to lose at least 13 games last year. And it also includes the Seahawks, who were last in the league in offensive plays, first in defensive plays, and both by huge margins. And yet somehow, they scored more points than they allowed on the way to finishing 7-10. But what's bizarre is that the Seahawks' play difference is more than twice as far below zero as anyone else's in the entire league.

But just looking at one year of stats doesn't paint the full picture, so let's broaden the data set to include the last ten years, for 320 team totals (= 32 teams x 10 years). The Seahawks averaged 56.12 plays per game last year, making them one of only four teams to average 57 plays per game or fewer. The 2018 Dolphins, 2019 Redskins, and 2018 Cardinals are the other three, and they had point differentials of -114, -169, and -200, respectively. Two of those three also finished 3-13.

Meanwhile, the Seahawks defense was on the field for 70.65 plays per game, making them one of only 5 teams to have their defense on the field for at least 70 plays per game. The 2013 Eagles, 2015 Eagles, 2013 Vikings, and 2018 Browns are the other four. Chip Kelly's uptempo offense accounts for those two Eagles teams (and the 2015 version was second in the league in offensive plays run), and is a good reason to focus on play differentials rather than absolute totals.

But to simultaneously be unusually low in offensive plays and unusually high in defensive plays is a special trick. So what happens when we look at the play differences? Here's the top six:

Team                      play  point
2012 Detroit Lions +162 -65
2018 Baltimore Ravens +161 +102
2021 Baltimore Ravens +152 -5
2019 New England Patriots +147 +195
2012 New England Patriots +145 +226
2019 Baltimore Ravens +143 +249

That's some fairly successful franchises, and also the 2012 Detroit Lions. But there's not an extreme outlier there. Let's look at the other end. Here's the bottom 15:

Team                      play  point
2013 Minnesota Vikings -119 -89
2020 Houston Texans -128 -80
2012 Tennessee Titans -129 -141
2014 Tampa Bay Buccaneers -131 -133
2020 New York Jets -133 -214
2013 Dallas Cowboys -137 +7
2018 Cincinnati Bengals -146 -87
2018 Miami Dolphins -146 -114
2018 Arizona Cardinals -163 -200
2017 Cincinnati Bengals -164 -59
2015 St. Louis Rams -171 -50
2016 Miami Dolphins -175 -17
2014 Tennessee Titans -180 -184
2019 Washington Redskins -194 -169
2021 Seattle Seahawks -247 +29

That is not a lot of successful teams. Three of those teams went 2-14 and two others went 3-13. The only two that won more than 7 games are the 2013 Cowboys (8-8) and the 2016 Dolphins (10-6). It seems that leaving your defense on the field all day is not conducive to winning. Among other things, the defense wears down and then you get no pass rush in the fourth quarter.

Meanwhile, the Seahawks are a full 53 plays shy of the next worst team. Yes, yes, 17 games versus 16. But they're the only team from 2021 in the bottom 15. They're only 4 plays away from matching the sum of the play differentials of the next two lowest teams to have a positive point differential. That's really an extreme outlier, and in a way that tends to lead to bad teams.

One obvious question is, do the Seahawks always have a negative play differential? Is this just Pete Carroll's style? The answer there is definitely no. Their next worst of the last decade is -90 (2020), and the their third worst is -38 (2017). Excluding last year, they're at +127 in total for the previous 9 years.

The next obvious question is, how did that happen? How do you get such an extreme result that is normally correlated with being awful, while still scoring more points than you allow? And that, really, is the story of the Seahawks' season.

The first question is, was it because of the offense or defense? To get that extreme of an outlier, you'd expect the answer to be "both", and it largely was.


Let's first look at the Seahawks' offense. The league average was 6.04 plays per drive. The Seahawks had only 5.22 plays per drive. The next lowest was the Saints at 5.52. It's not just that the Seahawks' had the fewest plays per drive in a vacuum. They had the fewest by an enormous margin.

But that didn't mean that the Seahawks had really short drives as measured in yardage. They averaged 29.4 yards per drive, good for 22nd in the league, and not that far below the league average of 31.7. They also averaged 2.07 points per drive, which was above the league average. The Seahawks were also 20th in offensive yards, in spite of running the fewest plays by far.

On a per play basis, the Seahawks offense actually looked good. They averaged 5.8 yards per play, good for ninth in the league, and well above the league average of 5.4.

So in spite of the Seahawks offense looking good on a per play basis, something was cutting those drives short. One obvious answer is turnovers, but the Seahawks 13 turnovers was tied for fewest in the league. Even on a per drive basis, only 7% of the Seahawks drives ended in a turnover, which was third lowest in the league. So turnovers weren't the problem, at least for the offense.

Stopping for field goals can also end a drive, whether you make the field goal or not. But the Seahawks only had 23 field goal attempts, the second fewest in the league. The Browns had 22, and everyone else in the league had at least 27. The Seahawks actually had an unusually low fraction of their drives end in field goals. Which makes some sense, as drives the end in field goals tend to take quite a few plays to drive down the field.

How about touchdowns? Touchdowns can end a drive abruptly, of course. The Seahawks had 48 offensive touchdowns, well above the league average of 42. The Seahawks scored a touchdown on 5.03% of their offensive plays, which was second in the league behind the Buccaneers at 5.35%. On a per-play basis, the Seahawks scored a lot of touchdowns. But while there have been some dominant, quick strike offenses at lower levels of play that would frequently get a long touchdown in two or three plays, that doesn't happen in the NFL.

Third down conversions is an obvious thing to check. After all, if you get three third down conversions in a drive, it's going to be a long drive. The Seahawks converted on 37.3% of their third down attempts, good for 23rd in the league, as compared to a league average of 40.4%. That's not good, but neither is it an extreme outlier.

Where the Seahawks are an extreme outlier, however, is fourth down conversions. They only had 11 attempts, the fewest in the league. Every other team in the league had at least 15 fourth down attempts, and the league average was 24.8. Furthermore, when the Seahawks did go for it, they were bad, converting only 36.4% of the time, the lowest rate in the league, as compared to a league average of 53.1%. They only had four fourth down conversions all season, while every single other team in the league had at least 9. Even a failed fourth down conversion attempt adds a single play to the drive, and a successful one usually adds several plays.

One could argue that if the Seahawks were so bad on fourth down, maybe they shouldn't have gone for it more. But that misunderstands the situation. The Seahawks were only a little below average on third down. They were the only team in the league to have a worse fourth down conversion rate than third down. With more fourth down conversion attempts, they'd probably have been more successful on fourth down.

Outside of some end of half situations, you always go for it on third down. On fourth down, you don't. The sensible thing to do on fourth down is selectively go for it in situations where you're more likely to convert. If you have fourth and inches from midfield in the second quarter, you go for it. On fourth and 10 from midfield, you punt.

But that's not what the Seahawks did. The Seahawks' average yards needed on fourth down conversion attempts was 8.9. Of course you're not going to have a great conversion rate when you need that many yards. So why did the Seahawks go for it? Because they'd usually wait until the situation was get a first down or it's game over before they'd go for it on fourth down.

The Seahawks were 4 of 5 (80%) on fourth down conversion attempts with more than four minutes to go in the game. They were even 3-1 in games in which they attempted a fourth down conversion with more than four minutes remaining in the game, winning all three of the games where they got it. (They had two fourth down conversions in one game.) They were 0 for 6 in the last four minutes, with all six of those attempts requiring at least 5 yards for a first down. The median yards to gain for those desperation fourth down attempts was 12. No wonder the result was five incomplete passes and a 16 yard completion that was also short of a first down.

So if the Seahawks weren't going for it on fourth down and they weren't attempting field goals, then what were they doing? You already know the answer to that: have you ever heard of Michael Dickson? The Seahawks were tied for the second most punts in the league. The Seahawks had 7.55 punts per fourth down conversion attempt. The league average was 2.62. The second highest ratio in the league was the Texans at 4.85.

And even those numbers are made much less extreme than they could be when you consider that a majority of the Seahawks' fourth down conversion attempts were desperation late in games, which wasn't true of most of the league. Outside of that, in a situation where another team would have gone for it on fourth down, the Seahawks usually punted.

If you consider only conversions, which excludes more of the late game desperation heaves, the Seahawks had 20.75 punts per successful fourth down conversion. The second highest ratio in the league was the Texans at 9.78. Third highest was the Steelers at 7.70. The league average was 4.93.

Analytics experts have long said that NFL teams really should go for it more often on fourth down. Most of the league has gotten that memo. Last year, the Seahawks were the only team that hadn't, even if some other teams still punted too often. Even so, there's a difference between punting when the chart says that you're expected to get an extra 0.5 points if you go for it, and punting when the chart says the expected difference is 0.05 points. When the chart says it's close, it's reasonable and not even that harmful to make the "wrong" decision. But the Seahawks often punted when it wasn't a close call.

Shortening your drives by being far too quick to punt is one way to get your offense off the field more quickly. It did mean that Seahawks' opponents started their average drive from their own 26 yard line, the furthest back in the league. For comparison, the league average was to start from your own 28.8 yard line.


Fourth down punts aren't the whole story. The other half is that the Seahawks' defense couldn't get off the field. The Seahawks' defense averaged 6.53 plays per drive, the second highest in the league.

Long drives tend to lead to a lot of points per drive, as both tend to be caused by bad defenses. But the Seahawks only allowed 1.91 points per drive, significantly below the league average of 2.05. The other five teams in the top six in plays per drive allowed were all well above the league average in points per drive allowed. So what happened?

One way to end a drive quickly is turnovers, and the Seahawks' defense didn't force a whole lot of them. Only 8.5% of drives against the Seahawks ended in a turnover, as compared to the league average of 11.5%. That's in spite of the Seahawks' defending drives with more plays than average, so the turnovers per play forced by the Seahawks defense was even worse.

One might suspect that the Seahawks were simply bad on third and fourth downs, but that wasn't the case. The Seahawks allowed a third down conversion rate of 39.3%, which is better than the league average of 40.4%. They allowed a fourth down conversion rate of 41.9%, good for third best in the league. When they had a good chance to get off the field, they were actually fairly good at getting off the field.

The Seahawks' red zone defense was actually pretty good. Only 50.8% of opponents' trips to the red zone resulted in a touchdown, the fourth lowest rate in the league. For comparison, the league average was 58.5%.

The problem was that the Seahawks were rather bad at keeping their opponents out of the red zone. 32.3% of opponents' drives made it into the red zone, as compared to a league average of 30.6%. That's 12th highest in the league, which isn't terrible, but remember that that's with opponents starting with the worst field position in the league. The Seahawks allowed 61 opposing drives to reach the red zone, tied for 6th highest in the league.

So what happened? You probably saw it with your own two eyes. The Seahawks defense would play fairly soft and allow their opponents to slowly march down the field. Then upon reaching the red zone, they'd tighten up, often successfully. But all those long drives kept the defense on the field for far too long, and that wore them out. By the fourth quarter, the Seahawks defense was far more tired than their opponents' defense. That's how you lose most of your close games.

The Seahawks simply allowed too many short throws to slowly eat up yards. The defense's average depth of target (on opponents' passes) was only 7.0 yards, the second lowest in the league. They only allowed 3.30 air yards per attempt, the third lowest in the league, and as compared to a league average of 3.91. (Air yards only count on completions, which is why this is much lower than depth of target.)

But then the Seahawks allowed 3.86 yards after catch per target, as compared to a league average of 3.53. This wasn't a problem of an inability to tackle, either. The Seahawks had 0.255 missed tackles per completion allowed, which is well below the league average of 0.289. Rather, they systematically gave their opponents too much room to catch short throws and run with the ball before they encountered any defender. That's how you give up a lot of long drives.

The good news is that these problems are probably fixable, as they aren't just about player talent. If nothing else, going for it more often on fourth down will help, at least so long as you do so in reasonable situations. It's not just about expected points added from going for it versus punting. It's also about keeping your offense on the field and letting your defense rest. Wearing down your opponents' defense while keeping your own defense fresh tends to make good things happen late in games.

The trade-offs of playing more aggressively on defense versus a softer defense are more complicated. But if the Seahawks could clamp down effectively on third and fourth downs and in the red zone even while worn down from being on the field for so long, that's probably not a problem of player talent.