1. I think we can all agree that if there we had to choose one person in the history of the world with whom we’d like to sit down for an afternoon of NFL action, it would be 17th-century Dutch mathematician and pioneer in the field of probability theory Christiaan Huygens.* And while he would almost surely suffer a series of cardiac episodes if subjected to the relentless, unending stimuli of the modern world, I like to think that he could also weigh in on the statistical probabilities of what’s unfolding on football fields across the nation every Sunday.
With that in mind, I assembled my annual midseason “luck report”—which, this year, came at the exact midpoint of the season (136 games into the 272-game season when you exclude the Ravens-Dolphins game). As I touched on a couple weeks ago when I couldn’t really come up with anything better to write, small-sample-size stats rule the NFL. Plays like fumble recoveries, red-zone stops, fourth-down conversions, missed field goals… they have huge impacts on games, but they also happen so rarely that they can’t really be viewed as anything a team controls in any significant way (and, in the case of fumble-recovery rate or opponent field-goal accuracy, they’re as close to dumb luck as you get in professional sports).
So, in order to get a better grasp on who performed well in the first half of the season, I built out a crude statistical model to try to remove luck from the equation as much as possible. Admittedly, this is less predictive than I’d like considering teams will improve, players will get injured, etc. It also doesn’t take into account injuries or strength of schedule—the latter I’m less bothered by since NFL teams are much more evenly matched than the typical observer realizes, but indeed, if you already got your shot at Jacksonville and Houston, you got a bit of an edge in this model.
A quick look at the factors I tried to mitigate:
• Fumble recoveries: The expected fumble recovery rate for every team in the NFL is 50%. A quick refresher on probabilities: A team that recovered 8 of 10 fumbles to start the year isn’t expected to recover 2 of their next 10, but 5 of their next 10. Which means that team is stealing three fewer possessions over those next 10 fumbles. I moved every team to a league average conversion rate and adjusted by net points-per-possession.
• Fourth-down conversions made/allowed: No one builds a fourth-down defense. I split fourth-down offense and defense into short (fourth-and-3 or less) and long (fourth-and-4-or-more), then moved each team to a league-average conversion rate (58.3% on short, 36.7% on long) for made and allowed. And, again, I adjusted by adding or subtracting net points-per-possession based on possessions gained or lost.
• Red Zone efficiency: It’s a far more volatile stat than most realize, probably because of the importance of that area of the field and the fact that it’s often discussed (in the vaguest of terms) on broadcasts. It seems offenses with a strong run game tend to fare better in the red zone (the Titans and Ravens usually rank toward the top of the league), but this year alone you have outliers in the Cowboys (24th in expected red zone points per possession) and the Colts (28th), both of whom have outstanding rushing attacks. I moved every team to the league-average (4.99 points per red-zone trip) on both sides of the ball.
• Opponent points per kick: Across the league, kickers are making 87.3% from inside 50, 66.7% from 50-plus, and 93.7% on PATs. I moved every team to those league averages on opponents’ attempted kicks.
I then took point differential per game and used the above factors to come up with a luck-adjusted point differential per game. And at the halfway point of the season, here’s what we have for adjusted point differential per game based on the season’s first half (thus, through the Steelers-Bears Monday nighter):
1. Cowboys, +8.9
2. Bills, +7.6
3. Bucs, +6.5
4. Browns, +5.8
5. Chiefs, +5.4
6. Colts, +5.22
7. Raiders, +5.21
8. Cardinals, +4.7
9. Patriots, +3.59
10. Packers, +3.56
11. Rams, +3.1
12. Bengals, +2.0
13. Broncos, +1.4
14. Panthers, +1.3
15. Titans, +0.9
16. Vikings, +0.4
17. Chargers, +0.3
18. Saints, +0.1
19. 49ers, -0.5
20. Eagles, -0.8
21. Steelers, -1.2
22. Ravens, -2.0
23. Seahawks, -3.1
24. Giants, -3.5
25. Football Team, -3.7
26. Jaguars, -4.0
27. Falcons, -5.2
28. Lions, -5.3
29. Bears, -7.5
30. Dolphins, -8.3
31. Jets, -9.3
32. Texans, -11.6
So what did we learn? First, I hate your favorite team and so does math. Second, the Browns, Chiefs, Colts and Patriots are teams that stood out as having better first halves than their win-loss records suggest. Conversely, the Rams, Chargers, Steelers, Saints, Falcons and Ravens were probably not quite as good as their first-half records suggest. The Raiders and Cardinals were for real, and to a lesser extent so were the Bengals and Broncos.
I'm not going to state a defense of this model—it's math, you're arguing with math. But I will point out: The Ravens, 22nd here despite a 6–2 record, went out and lost to the hapless Dolphins in the first game of the season’s second half. So, certainly, based on that one-game sample size, we can conclude that the above model is absolutely foolproof.
C’mon, what about Blaise Pascal? I’d rather watch the game and eat nachos with Pierre de Fermat!
2a. As you likely know, the Rams were the winners of the Odell Beckham Jr. sweepstakes. Being able to recruit like that is another benefit of having a GM whose primary objective is to win games rather than prolong his stay by appealing to a cheapskate owner and the draft-pick fetishists in his fanbase. (Though playing in Los Angeles—where the autumn weather is nice, people are aesthetically pleasing and your chances of spotting the entire cast of on the street is greater than it is in any other city—is also a draw.)
The Beckham addition leaned a little more toward luxury than necessity—though you won’t find a defensive coordinator in the league who’s not aware of Beckham’s location on every snap—until the Robert Woods injury. How much of an impact Beckham will have about 100 hours after signing is up in the air though, and that’s unfortunate considering the struggles L.A. has had against Monday night’s opponent, the 49ers.
It’s a different team with Matthew Stafford under center, but the Rams have lost four in a row to the Niners and in three of those their offense was completely overwhelmed, falling short of 200 net passing yards. Last November, they failed to score on their last two drives to put the game away and ended up losing to a Nick Mullens–led Niners team. This matchup, more than any other, should lay bare the stark difference between Goff and Stafford, though it comes one week after Stafford had a mini-meltdown to cost them against a shorthanded Titans team. They need their MVP-caliber quarterback to play at an MVP level again on Monday.
2b. While Woods isn’t as talented as a receiver as Beckham, his value in formationing (he’s taken a couple handoffs as a traditional running back this season) and as a blocker add up to value well beyond his statistical output. L.A. will likely be fine in the long run either way, but in some weird timeline where they were forced to choose between a healthy Robert Woods or adding Odell Beckham Jr., they’d be better off with the former.
