NFC LEast

 

Tonight we’re going to see the latest in a long line of legendary Thursday Night Football matchups. Two bitter division rivals clash as the New York Giants head to Philadelphia to face the Eagles. Both teams still trail Dallas for the division lead, but they are less than a game back, and this early season collision could go a long way to determining which of these teams is a contender going forward.

Obviously, I’m being a bit sarcastic. I’ll be watching this game, but I couldn’t recommend that anyone else do the same, as two of the NFL’s worst teams stumble together for another ugly football contest.

Everything I said in the first paragraph is true though. Both teams are very much still alive in the division race, thanks to the utter ineptitude of the NFC East. It was a bad division a year ago, and it has somehow gotten even worse this year, as young players have failed to develop, established stars have gotten hurt, and just about everything that could go wrong has gone wrong.

We know that the NFC East is bad. But this of course raises the question, just how bad is it? It certainly feels like this particular season is unusual. Through six weeks these four teams have combined to go 5-18-1, for a winning percentage of 23% (assuming you count a tie as half a win, which I am going to do from this point forward). Of course, three of these wins have come in games within the division, where someone has to get a victory. So if we subtract three out from both their wins and their losses, they have a winning percentage of only 14% against the rest of the league.

This seems like a prime opportunity to pour through the data. My plan was to look back at every season to 2002, the first year the league went to its current format with eight divisions of four teams (can you believe the Houston Texans are old enough to vote?) But sometime after I scraped the data going back to 2009, Pro Football Reference added a paywall to their game search data. I’m sure I’ll give in and pay the $8 a month eventually, but for now I’ll stick with the 11 years I have available.

I looked through each division at the six week point in the season and picked the worst from each year, to see how they fared six weeks into the season and how they finished things off.

The short answer is, yes this division is a special sort of terrible. They aren’t even close to the worst divisions of previous years. No previous worst division won less than 30% of their games through the first six weeks, and even when you strip out in-division wins, no other division has been below 20%. The NFC East would have had to find two more wins somewhere just to get up to utterly incompetent.

If there’s good news to be found, it’s that the worst division through six weeks doesn’t always mean that much in the grand scheme of things. In only 4 of the 11 years I looked at was the worst division at the end of the year the same as the worst division through six weeks. Overall a division’s winning percentage through the first six weeks only had a correlation of 19% with their winning percentage the remainder of the season.

(That correlation jumps to 67% if you compare their winning percentage through the first six weeks with their full-season winning percentage, which makes sense since those first six weeks are part of the full season. I’m mostly including this number for later reference.)

Of course, wins and losses aren’t the only way to judge teams. Sometimes good teams catch some bad luck due to strength of schedule or bad breaks in close games. That isn’t the case with the NFC East this year. These teams represent four of the bottom seven in the league in DVOA, a metric that looks at play-by-play success and adjusts for strength of competition. I was able to find weekly DVOA going back to 2004. Here was how things stacked up in each of those seasons.

Once again, we see that this year’s NFC East is truly something special. No division had an average DVOA rank worse than 25 through six weeks in the season. The NFC East is far below that at 28.5. They would need the Cowboys to jump up to the 12^th best team by DVOA just to match the historic bottom. Currently the 12^th ranked team by DVOA is Green Bay.

Things are actually a little closer when you look at the DVOA score, which is measured as percentage above or below average. The 2020 NFC East is 23% worse than average—again the worst among the years in our sample, but there have actually been a few divisions in the past that have come close to this. The 2013 AFC South was dragged down by three teams—Indianapolis, Tennessee, and Jacksonville—with DVOAs below -30%. Jacksonville alone did them in in 2013, with a DVOA of -66%. And San Francisco sank the NFC West in 2005 with an astounding -71%. For comparison, no one in this year’s NFC East is below -28%. The worst team this year is obviously the Jets, and even they are only at -41%.

(It’s possible that some of the DVOA calculations may have changed and not been applied retroactively, dampening the results. Though the worst week 6 DVOA in the entire sample belongs to the 2019 Miami Dolphins at -84%, so they can’t have shifted things too much.)

Another interesting thing to note from this table is that things generally seem a lot more stable. Where the table based on winning percentage never had the same division in two straight years, the DVOA table picks out the same division through six weeks in consecutive seasons on a regular basis. The NFC West “won” four out of five years from 2007-2011, which pretty well matches up with what I remember from that period. The AFC South followed that with a historic streak of five straight seasons, which again checks out.

DVOA also seems a bit better at identifying which division will be the weakest at the end of the year. In 9 of the 16 seasons in my sample the two were the same (and 6 of the 11 where it overlaps with the sample based on records). And while generally the divisions’ DVOAs reverted towards average as the season went along, it was not super dramatic. The correlation between 6 week DVOA and full season DVOA was 78%, even higher than the 67% number I gave above.

There may be reasons to expect this year’s NFC East to get better. First of all, it can’t really get worse. There are a pair of young teams with new head coaches in Washington and New York that should improve as the season goes along. Philadelphia should get healthier on offense, and Dallas…well I don’t really have any reason for optimism for Dallas. But at this point though, the 2020 NFC East has dug themselves a hole unlike any other division in recent history. Only time will tell if they are as uniquely terrible as they appear to be right now.

Football in the Time of COVID

Four weeks into the NFL season, we have already seen our fair share of interesting storylines. There are surprise teams jumping to the top of the league, and disappointments falling to the bottom. There have been thrilling comebacks and last-minute decisions. There have been breakout rookies and aging veterans pushing for one last run.

But beneath it all there’s been another story. The story that is beneath everything happening in the world right now. The story of COVID-19, and how it is shaping every aspect of our lives.

The NFL managed to elude this for the first couple weeks of the season. It was a little strange to watch games happening in empty stadiums, but for the most part you couldn’t tell the difference on TV. The first three weeks went off without a hitch. Then, Tennessee happened.

At this point more than 20 members of the organization have tested positive in the past two weeks. They shut down the facilities a week ago and cancelled their Week 4 game against Pittsburgh, and on Monday and Tuesday it looked like this had worked as they had no more positive tests. But two additional positive tests on Wednesday have put things in further jeopardy (possibly traceable back to unauthorized player-led practices away from team facilities a week ago).

As this was happening, New England was also struck a major blow when starting quarterback Cam Newton tested positive, forcing him to sit out and pushing their game against Kansas City back a day. And on Wednesday reigning Defensive Player of the Year Stephon Gilmore tested positive, suggesting that they may be at risk of a larger outbreak. Long story short, things are not going very well.

Only one game has been cancelled so far, and the league was able to handle this with little damage by shifting around one game later in the season. I doubt they will be as fortunate in the likely event that future cancellations arise. The NFL schedule is a convoluted web, and if you start pulling one strand, the entire thing unravels very swiftly.

The easiest solution—and one the league somehow hasn’t approached yet—would be to add additional buffer weeks at the end of the regular season to accommodate rescheduling. If they eliminate the bye before the Super Bowl, they can keep that on the same date and move each playoff round back a week to create an 18^th regular season weekend. This seems like an easy win—schedule flexibility, no shifts of major events, and an extra week of TV revenue for the league—though if teams start seeing multiple games cancelled, this will be insufficient.

As the NBA, NHL, and WNBA have wrapped up the successful conclusions to their seasons, there has been a great deal of push for the NFL to emulate the bubbles they created. Confine all the players, coaches, and other staff to hotels, eliminate the risks of travel, and keep everyone in a place where they can be carefully monitored.

It’s a good idea on principle, and it worked successfully for the other leagues. But there are a few problems that make it almost impossible for the NFL. The first issue is one of simple numbers. NFL rosters are much bigger than any of these other leagues, and the larger a bubble is, the harder it is to maintain. The NBA invited 22 teams, and each team was allowed to bring around 35 people (roughly 15 players plus 20 staff—coaches, trainers, scouts, equipment managers, etc. That means this bubble was around 800 people. That sounds like a lot, and it took three hotels to accommodate them all.

Each NFL roster has 53 active players. But attrition is much higher in the NFL, and a bubble would make it impossible to sign someone off the street on short notice. I think it’s unlikely they would be able to operate in the bubble without at least the standard 10 player practice squad. The staff required would need to be much larger as well. All together, I think the most optimistic plan possible would be 100 people per team (and even that is likely unfeasible). With 32 teams, that means this bubble would be four times the size of the one the NBA built.

Length is an issue as well. The NBA players entered the bubble on July 7, and the last of them will leave it in the next couple days. That means three months maximum, with most spending much shorter times there. To complete an NFL season, every player would have to remain for the full three months. The teams that make the Super Bowl would be required to remain for an additional month. That’s not including any sort of quarantine period at the beginning before they start playing again. Realistically we would be proposing players spending nearly five months locked inside hotels.

The chances of the players agreeing to this are slim. The NBA made it work by overloading the players with amenities at their Disney campus—pools, fishing, golf, movie screenings. As we enter fall and winter, the number of places that could accommodate this are rapidly thinning, especially at the numbers we are talking.

And it’s not just the hotel facilities that are the problem. Both practice and playing facilities for football are much harder to find than for basketball. A basketball gym is tiny compared to a football field, and practice demands for NBA teams are smaller than for NFL teams. To form a bubble we would need to find adequate facilities for 32 different teams to practice during the week, as well as places for them to play the games. This means not just fields but locker rooms, medical facilities, and of course TV broadcasting equipment.

There is no way a single bubble could work. There is maybe a slight chance they could pull off four NBA-sized bubbles. Put two divisions in each and rebuild the schedule with 10 games—play each division opponent twice and cross over against the other division once each. The best solution would be to find somewhere that is within same-day driving distance of multiple NFL stadiums to play the games at—two at least, but three would be better. A few possibilities that come to mind:

• Los Angeles, assuming the stadium the Chargers played in the past few years is still available, and having multiple college stadiums available if need be
• Orlando: Reasonable driving distance to Jacksonville, Tampa Bay, and Miami as well as having decent local facilities
• Indianapolis: Centrally located with driving access to Chicago and Cincinnati, though I wouldn’t want to see what happens to the turf at Soldier Field if they play more than the normal 8 games a year on it
• Cleveland: Access to Detroit and Pittsburgh, not that the turf at Heinz Field is much better than Chicago, and Heinz Field shares facilities with the University of Pittsburgh so it would be hard to keep isolated
• Baltimore: Access to Washington and Philadelphia, again major field durability concerns in Washington
• Atlanta: Access to Charlotte and Nashville, though some of these driving distances are starting to strain feasibility
• Bay Area: Not sure what condition the Oakland Coliseum is in for hosting football games, but it would be funny to send the Raiders back there
  

There are options. But there are problems with all of them. Maybe when the playoffs come around, they can try to build bubbles around the remaining teams. Until then, I think the only real course is to try to muddle through the season, fighting off these outbreaks as they pop up.

Bye Weeks

One of the first takeaways from the rescheduling of the Titans-Steelers game this past weekend was to feel bad for Pittsburgh. They spent most of the week preparing for a game that ended up not happening, and in doing so they pretty much lost out on the normal week off NFL players receive during their bye. Not only that, they now have to play thirteen consecutive games to finish the season.

That last part addresses some common wisdom in the NFL. An early bye week is generally seen as a piece of scheduling misfortune, to the point that the league departed from past years this season by having no scheduled byes in Week 4. Pittsburgh had what seemed like an ideal bye in Week 8, and because of Tennessee’s outbreak, they were forced to shift their bye to an earlier week.

Of course this raises the obvious question: is there any evidence that an early bye hurts a team? And this isn’t something particularly difficult to find out. I collected season results from the past ten years to see what the average win totals were for teams that had their byes in each given week.

If everything was equal, we would expect to see eight wins on average regardless of the week of the bye. If there is a disadvantage in having an early or a late bye, we would expect to see lower win totals at the ends and higher average totals near the middle. This of course assumes that these byes are distributed independently of team quality, which I think is a fair assumption.

 

The data above seems pretty clear to me. The two best outcomes are for teams with byes in Week 4 or in Week 12, opposite of the common wisdom. But just in case we thought there might some contrary advantage in having long stretches of uninterrupted games, there is a small peak in the center. The whole distribution seems fairly tight around the expected average of eight wins. And a simple T-test on each bucket with a p-value of 5% is not enough to dismiss the hypothesis that each week produces equal outcomes.

So moving the bye was probably a neutral move for the Steelers. But if they still feel like complaining, they can point to the fact that the changed schedule gives Baltimore a bye week right before their crucial showdown with the Steelers that could be a major factor in deciding the division.

But do teams really have an edge the week following a bye? In this case it’s even easier to figure out. Over the past 10 years, teams the week following a bye are 164-154-2. So they are slightly above the .500  But it’s still not significant enough to say that coming off a bye week is a clear advantage for NFL teams. Now it might be more meaningful against an elite coaching staff like Baltimore’s, as they’ve won 80% in weeks coming off a bye over the past 10 years compared to 60% in all other games. But ultimately that feels like a negligible difference as well.

There is going to be more schedule shuffling as the year goes on. But ultimately the schedule doesn’t mean that much. Good teams are going to win games, and bad teams are going to lose games (unless they’re the Bears, who keep winning somehow). These scheduling changes aren’t going to have a major impact on the season. Until the league starts handing out forfeits. Then we can start complaining again.