The Panthers' Best Run Defense Might Be A Good Offense
Regular readers of mine know that I never pass up a chance to point out that the Panthers had an historically bad 18.6% run defense DVOA last season. At first, it was a coping mechanism for a fan sad to see the team's Cam Newton era crumble one loss at a time. But when players such as Tevin Coleman, Aaron Jones, and Latavius Murray enjoyed career performances one Sunday after the other, the decline of the Panthers' run defense transitioned from merely sad to very relevant for fantasy players like myself who like to play weekly matchups.
That realization came too late for me to take full advantage. At the start of the season, I didn't expect the Panthers to have so porous a run defense. In 2018, they were better against the run (-7.0%, 18th) than the pass (14.8%, 24th), and they maintained a core of run-stopping defensive starters in Luke Kuechly, Kawann Short, Dontari Poe, and Eric Reid from one year to the next. Still, I benefited from enough plus performances from backs such as Chris Carson and Marlon Mack to offset the occasional miss of a Brian Hill.
At the time, I had a theory that the Pro Bowler Short was a lynchpin for the team's run defense and thereby the reason the team became a plus matchup for opposing running backs. He tore his rotator cuff in Week 2 and missed the rest of the season. But a broader testing of my projection system this offseason made me question whether I had missed the mark entirely. Short's injury may have sabotaged the team's run defense, but I now suspect it was Newton's injury that made the Panthers a plus matchup for opposing backs by virtue of damaging their overall quality.
Backs on bad teams had success against the Panthers last season. Washington runners combined for 242 yards and three touchdowns against them in Week 13, and Hill and his recovered teammate Devonta Freeman had 146 yards and two scores in Week 14. But Hill's earlier letdown in Week 11 was a part of a subtle trend of worse teams enjoying smaller benefits from the plus Panthers matchup. Split simply by whether a team had a below-average or above-average DVOA at the time of the matchup -- relying on DAVE in the early season and weighted DVOA in the late season -- bad teams gained just over half as many extra running back carries against the Panthers as good teams.
RB Carries Over Projected Total,
|Week||Team||DVOA||Proj Att||Att||Surplus Att|
|Above-average DVOA teams per game||3.0|
|Below-average DVOA teams per game||1.7|
|Projections adjusted for venue and weather but not defense|
In theory, my venue- and weather-adjusted attempt projections should capture considerations such as offenses' pass and run tendencies. Subsequent opponent adjustments should then offset any remaining attempt surpluses and do so evenly for all teams over big enough sample sizes. The 16 teams that played the Panthers last season are hardly a big sample size, but if the split those teams showed last season exists for all teams, then it illustrates that opponent adjustments of carries based solely on run defenses will remain biased by team quality.
That bias seemed strange to me until I reframed it around the more familiar concept of game script. Game script was a major talking point of my non-fantasy previews of the Titans' playoff run last season. Those Titans upset the Patriots and Ravens because their early leads allowed them to run the ball, a relative strength for the team. Had they trailed instead, the Titans would likely have been forced to rely heavily on the pass to try to score a lot of points with less use of the clock. That play-calling pattern shows up broadly in a steadily decline from a 60% run ratio for teams with three-score leads to a 30% run ratio for teams with three-score deficits.
I tend to think of it as an in-game concern, but game script is at least somewhat predictable before games by the relative quality of the teams involved. It may not have been the case for the Titans, Patriots, and Ravens in January, but better teams tend to build leads and move the game script to favor rushing attempts. Pretty much every team that played the Panthers last year was better in overall quality. By season's end, Carolina was the NFL's second-worst team with a -26.7% DVOA. But their better opponents enjoyed relatively more surplus carries, and that trend existed for all teams last season with DVOA differential serving as an approximation of their relative quality.
Their relationship has a weak coefficient of determination of just 0.0115 -- a common refrain of the statistical research of so random a sport. But DVOA differential still does a slightly better job of predicting carry surpluses than my yards-per-attempt opponent adjustments did.
And it does a dramatically better job than my touchdown-per-attempt opponent adjustments.
And perhaps most importantly, relative team quality is easier to predict. As Aaron writes every year in the Football Outsiders Almanac, "offense is more consistent from year to year than defense, and offensive performance is easier to project than defensive performance." A healthy Short could be transformative for the 2020 Panthers' run defense. So too could top draft pick Derrick Brown. We project the team to have the worst total defense this season, but its 11.0% DVOA projection is just half the outlier from average as the best (Kansas City, 20.4%) and worst (Washington, -17.8%) teams in projected offensive efficiency. All things considered, we're much more confident that the Panthers will be a bad team in general than a specifically bad defense or run defense. And while you may not care why it is that your running backs continue to produce big fantasy numbers when you play the Panthers matchup, that knowledge can help you make the right matchup plays when your backs' opponents are not quite so extreme.