Regression to the mean is an ugly word in statistical circles.
Typically it means that someone or a team has outperformed expectations and some pull back is expected. That is particularly relevant for Florida’s 2019 season, as I’m not sure many people had Florida pegged for a 10-win season and a top-10 finish in the AP Poll.
The Gators 2017 season was marred by many things. The credit card suspensions, the death threat fiasco and the 4-7 record on the field. Much of that was due to a stagnant offense, the calling card of Jim McElwain during his time in Gainesville.
But amid the offensive futility and the off-field stuff lay another issue. The Gators gave up over 8.0 yards per attempt through the air. The vaunted defense that had always been reliable – even when the offense struggled mightily – had finally broken.
Combine that with an offense that never seemed to get better under McElwain and the 4-7 record should have been predictable.
In came Dan Mullen and Todd Grantham. The Gators defense improved from 67th in yards per play allowed to 37th against FBS opponents, a 30-spot improvement in the rankings. The offense improved even more, as the Gators improved from 104th in yards per play in 2017 to 32nd, an improvement of 72 spots in the rankings.
The result was that the Gators gave up 5.1 points per game less vs. FBS opponents in 2018 compared to 2017 and also scored 8.7 points per game more than the previous year. The almost two-touchdown difference in scoring differential is how you go from 4-7 to 10-3.
The natural expectation is that 2019 will bring more of the same. After all, it’s everyone’s second year in Mullen’s system and Grantham came back rather than accepting the defensive coordinator job with the Bengals, right?
But does that actually happen? Do teams that improve a ton on offense sustain that improvement? And do teams that improve significantly on defense keep those gains or improve ever further?
FBS track record of improvement
To measure this, I went back and looked at 120 teams over the past decade in FBS on both offense and defense.
I like measuring success of an offense or defense by looking at yards per play gained or allowed. Over time, I believe this ends up being a much more reliable measure than just looking at points allowed. Because a unit will execute nearly 1000 plays during a season, looking at things on a per-play basis is a much larger sample size.
Plus, sometimes a defense gets put in a bad position because of a turnover or a defense that can’t get off the field forces an offense out of what it wants to do.
The three points Florida gave up to Georgia last year after Franks fumbled at the one is a perfect example. Those three points were a positive reflection of the Gators ability to hold up, as they gave up zero yards on six plays.
For defense, I looked at teams that improved between 20 to 40 spots in yards per play ranking the year prior and how those teams performed the next season. For offense, I looked at teams that improved between 50 and 90 spots in yards per play rankings the year before.
It shouldn’t be surprising that these teams did not improve significantly the following season. After all, if you were ranked 80th and improved 70 slots, it’s hard to improve much from there. However, the surprising thing to me was how many teams regressed, both on offense and defense.
122 FBS defenses showed mild improvement (20 to 40 ranking spots) over this 10-year time-frame and of those, 85 (70%) regressed. The average regression was 13.5 ranking spots.
For Power-5 teams, the story was much the same.
Change in defensive yards per play allowed ranking the year after a 20-40 ranking improvement vs. FBS opponents from 2009-2018. (Will Miles/Read and Reaction)
You can see the most likely outcomes in the histogram above for the defense for Power-5 teams. 64 defenses showed an improvement of 20-40 rankings spots and 64 percent of those regressed an average of 10.8 slots.
For offenses, the story is much the same.
There were far fewer FBS opponents who showed the level of improvement as Florida (50-90 ranking improvement). 72 teams fit that description, and of those, 64 percent regressed. Overall, those 72 teams regressed an average of 16.7 rankings spots.
Change in offensive yards per play gained ranking the year after a 50-90 ranking improvement vs. FBS opponents from 2009-2018. (Will Miles/Read and Reaction)
The histogram for these teams is actually a little bit more encouraging. The most likely occurrence for those teams was actually a slight improvement (17 teams). But there were only 9 teams that improved significantly in the following year.
Again, this is expected. You can’t see an improvement of 70 spots two years in a row. But the fact that 64 percent of those teams see some sort of regression is concerning.
What does this mean for Florida?
But should it be concerning, really?
Well, if Florida does what the average team has done over the past decade, the Gators will end up 50th in defense and 49th in offense, hardly the recipe for a follow-up 10-3 season.
Were that to happen, Florida would average 29.5 points on offense and allow 25.9 points on defense. That’s a team that likely ends up 7-5.
Offense
Mullen’s track record does provide some hope. Not only is he an offensive coach, but while at Mississippi State, the Bulldogs offense immediately improved 44 spots in yards per play in his first year. The offense then slightly improved in year two.
Additionally, there weren’t any wild swings where the offense improved 50 spots and then dropped that far as well. If anything, we know we’re getting consistency from Mullen.
And on the offensive side of the ball last season, the Gators finished 32nd in yards per play and 37th in points per game against FBS opponents. That indicates that the offense, if anything, was slightly unlucky.
The biggest thing I think that points towards at least maintaining the offensive improvement next year is who is returning. The quarterback, running back, and top-5 wide receivers are all coming back.
The offensive line has some major question marks, but I detailed earlier this summer why I think there is reason to believe that Mullen and Hevesy will be able to get that unit to play at a high level.
Defense
Instead I think how much the Gators can success in 2019 will be tied to whether the defense can sustain its improvement from last season.
Grantham’s track record is a little more spotty that Mullen’s. He showed giant jumps in defensive efficiency at his last two stops (Mississippi State and Florida), but this is the first time he’s going to coach a second year after seeing this type of increase.
In the three years prior to Grantham’s arrival at Georgia, the Bulldogs actually had better defenses than during his four years there. But the three years prior to his arrival at Louisville, the Cardinals were 16 spots worse (26th vs. 10th) and Grantham piloted some elite defenses.
I’m just not sure there’s a lot to point to either way that tells us whether Grantham is definitively a difference maker as a defensive coordinator.
And Florida has some pretty big shoes to fill on the defensive side of the ball.
Chauncey Gardner-Johnson was a dynamo at the star position last season after starring as a freshman. The fact that he struggled his sophomore campaign in a cover-2 scheme forcing him to be a primary tackler shouldn’t be held against him. Gardner-Johnson was perhaps the best player on the Florida defense last season, or at least narrowly behind C.J. Henderson.
A source of constant pressure, the only games Jachai Polite didn’t have a tackle for loss were against Charleston Southern and Georgia. It’s not a coincidence that Georgia and Missouri – games where Polite did not record a sack and only had one TFL – were games where Florida’s defense looked limited. Combined with Jabari Zuniga, the tandem meant that opposing offensive lines couldn’t just double-team one guy.
Perhaps one of the more enigmatic Gators, linebacker Vosean Joseph could both thrill and disappoint, sometimes on the same play. The opposition found big plays when he didn’t play the right coverage or jumped out of his gap multiple times during the season. But when he was freed up to attack the line of scrimmage, I’m not sure there was anyone better in the country.
Florida may be able to upgrade at some of these spots in 2019. All accounts are that Amari Burney is poised to be a star at linebacker stepping in for Joseph. Trey Dean certainly showed flashes outside at corner and should see a benefit from moving inside. And transfer Jonathan Greenard seems poised to step in for Polite.
But that’s replacing 22 percent of the team’s total tacklers, with Joseph as the team’s leader in that category. It’s replacing 49 percent of the team’s sacks and 42 percent of the Gators tackles for loss. It’s replacing 21 percent of all passes broken up and 55 percent of all forced fumbles.
That defense last season finished 24th in points per game while ranking 37th in yards per play. That means that the results on the scoreboard outpaced the results on the field. I suspect a lot of that differential came from turnovers, many of them caused by Polite.
Takeaway
Against FBS opponents, Florida had a scoring differential of +8.6. The four playoff teams had an average scoring differential of +21.5 points with the lowest being +13.2 points (Notre Dame). That kind of record improvement won’t be possible by improving one side of the ball while maintaining the other.
Unfortunately for Florida, the Gators will be competing against Alabama (+27.2) and so the minimum point differential isn’t going to be good enough. Instead, it will require the Gators to improve both offensively and defensively, likely a significant amount.
One thing I do think is interesting is that ESPNs FPI gave Florida a 10 percent chance to make the playoff earlier this offseason.
Interestingly, the data says there is a 35 percent chance of offensive improvement and a 30 percent chance of defensive improvement. Multiply those two percentages together and you get the probability that both units will improve in 2019.
That percentage? 10.5 percent.
Featured image used under Creative Commons license courtesy Chris Short
Michael H
I recall the HBC once said you can tell how good a coach is by looking at what happens to a program after his arrival and again after departure. I wish I could find the quote.
While I never analyzed the data, I sure sensed an obvious huge step function shift upward in 1990 with a new, higher normal even for his “down” years. Then in 2002 it felt like a shift back downward even with Rex as QB. It wouldn’t surprise me if there is not continued sharp improvement this year from last, or even slight regression, but as badly as Mac and Muschamp offenses seemed to underperform year after year, I expect Mullen offenses will have a much higher mean that we hover around. Aside from 2017, defense has been pretty good for a long time, so I hope we can have another strong performance in 2019.