Analyzing NFL team’s injuries and strength of schedule in 2021 to determine potential regression in 2022
by Jason Pauley
A lot of factors that have a heavy influence on a team’s success are factors that they have less control over than their play on the field and talent on the roster. Perhaps they have some control over some of these things, but injuries, record in close games, fumble luck, and schedule strength are factors that aren’t sticky from year to year and they have a significant impact on a team’s record. I believe they are all useful data to project positive or negative regression the next year. They also give fans of teams like the Ravens or the Jets a legitimate reason to believe they are better than their record. In this analysis, I’m only focusing on two of the data points: Injuries and Schedule Strength.
- For injuries, I’m using Football Outsiders Adjusted Games Lost (AGL). You can read more about it here.
- For Strength of Schedule, I’m using Pro Football Reference’s SOS. You can read about it in this blog.
I plotted each team along an X and Y-axis. The X-axis is AGL (Adjusted Games Lost to injury); it moves from the least injured (left) to the most injured (right). The Y-axis is their strength of schedule; it moves from easiest schedule at the bottom, to the most difficult schedule at the top. This creates four quadrants. There is a quadrant with teams that have had the difficult circumstance of a lot of injuries and a hard schedule; this quadrant is on the top right. There is a quadrant with teams who have had a lot of health and an easy schedule; this is on the bottom left. There are two neutral quadrants with teams who don’t fall above or below average for both metrics. The size of the bubble represents the number of wins each team had.
The teams in the top right, theoretically had more difficult circumstances this season and might be good candidates for positive regression in 2022 (personnel changes aside). The teams in the bottom left, theoretically had more favorable circumstances this year, and are good candidates for negative regression in 2022.
Breaking down the data from the quadrants suggests that these two factors do have a heavy influence on a team’s record. Teams in the easy quadrant average 59% more wins than teams in the difficult quadrant. (The two caveats, as always are: correlation does not always mean causation. This is also a small sample size from only one season.)
Easy quadrant (bottom left)
- Average wins 11.0
- % of teams with a winning record — 100%
- % in playoffs — 89%
Neutral quadrants (top left and bottom right)
- Average wins 8.2
- % of teams with a winning record — 56%
- % in playoffs — 22%
Difficult quadrant (top right)
- Average wins 6.9
- % of teams with a winning record — 41%
- % in playoffs — 33%
- Note: The five worst teams (in terms of record) fall into this quadrant, although Houston barely falls in.
Obviously, this data alone doesn’t determine regression; the best and worst shifts in personnel will probably trump these factors, but I think this data is an interesting input and a big part of each team’s story in 2021 and can help to form expectations for the next year.
I plan on analyzing this more over the off-season with past data. I want to back-test if teams truly do regress positively or negatively when their quadrants suggest they should. I also want to see if the average wins per quadrant share the pattern, we see in the 2021 data. More to come on this.
