The danger of garbage inputs
This is when I’m supposed to call him a pencil pushing geek or rail against math nerds who don’t watch the games, but… come on. LSU did everything last season to deserve skepticism. LSU finished last season ranked 45th in the SP+ (and somehow 69th in defense… how did we get so high?).
So LSU getting pushed up to 26th in the preseason poll is actually the formula showing a degree of confidence in LSU. The team was bad last year (though it kept its streak of non-losing seasons alive!), but the numbers support the idea the team will bounce back.
How much they will bounce back is the source of contention. And this is when I get on my soapbox and reiterate one of my core principles: statistical formulas are great and helpful, but it is the principles behind those formulas which are actually important. The numbers are simply a manifestation of those principles, and the concepts are often more valuable than the numerical outputs.
Bill C’s formula is proprietary which I both understand and hate. That’s the way of the world these days and you can’t expect analysts to give away their numbers for free. On the other hand, I grew up on Bill James and Pete Palmer, and those guys not only showed their work, James famously broke the wand and gifted his work for free for everyone’s use.
Still, it’s not the numbers themselves that truly matter, it’s the concepts. And Bill clearly tells us what goes into the formula, essentially what his numbers are thinking. There are three factors going into the projection, in descending order of importance:
1. Returning production
2. Recent recruiting
3. Recent history
Let’s go through them backwards, finishing up at the most important factor.
Year SP+ Rank Off Rank Def Rank
2020 45 (5.6) 38 (33.2) 69 (28.0)
2019 2 (33.1) 1 (48.9) 20 (17.7)
2018 5 (24.2) 30 (34.8) 5 (13.4)
2017 11 (21.1) 23 (35.6) 8 (14.0)
2016 5 (23.3) 41 (34.2) 2 (9.0)
Last season was an outlier, even before we get into COVID. LSU notched a top 5 SP+ ranking in three of the four prior seasons, and #11 in the one other year. But the formula heavily rates the 45th ranked finish last season instead of those others years.
Now, assume there was no COVID and we weren’t talking about LSU. If you saw a team with this pattern of rankings and knew nothing else, would you assume they would bounce back? Probably. Everyone has the right to have a bad year. Then add in that the team had massive personnel losses after 2019 and had to replace both co-ordinators, whiffing on the DC hire who has since been fired.
How are you feeling about that team now? Then add the factor that last year was the Zombie Season and nearly every team who wasn’t Alabama was scraping the bottom of the barrel just to field a roster each week. Heck, even give a bonus for losing the starting quarterback at the end of the first month.
This is a team with a track record of success who had a nightmare year during, well, a nightmare year overall. You don’t get to completely throw out 2020, but I do feel that the track record of success augurs a comeback.
It’s been pretty great. LSU’s recruiting in the 247 composite ranked 4th this season and the last, and 5th in 2019. 247 hasn’t released its 2021 Team Talent composite yet, but LSU ranked 6th in the 2020 version and there’s no reason to think it will drop, as it attained that rank with just 68 recruits.
No one recruits like Alabama except maybe Georgia, but LSU is one of the most talented teams in the country by nay measure. This team had too much talent to be going 5-5, which speaks to the value of depth. LSU lost nearly every starter from its title team, then lost nearly another starting lineup due to opt-out and injuries.
LSU will enjoy the same top tie talent, but now will have the luxury of a second string.
This is the sticky wicket. SP+ rates LSU poorly because it was a mediocre to bad team last season, and there’s only so much improvement you can expect. Which is a fair position to take, and a math formula can’t make exceptions for a particularized situation.
The good news for LSU is that it returns 79% of its production, nearly evenly distributed on both sides of the ball (82% on offense, 76% on defense). This would normally mean great things for LSU’s rating this season, but two factors drag down the impact.
One is that everyone returns a bunch of talent. Normally, only a handful of teams top 80%, and this year LSU ranks 30th in the nation in returning talent at 79%. Due to the eligibility waiver, more teams will return more talent. Last year, 79% would have ranked 10th. Now it ranks 30th. That gives you an idea.
However, only Ole Miss (81%) ranks higher than LSU in the SEC in returning talent, and it is by a marginal amount. Mississippi St return 79%, just like LSU, and Arkansas is at 78%. But the teams LSU are chasing will be replacing a lot: Texas A&M is at just 60% and Bama 55% (not like it will matter in Bama’s case).
Of course, the bigger factor is the second, LSU returns a ton of its production, but it is talent that didn’t produce last year. Returning roughly three quarters of a historically awful defense is sort of a mixed blessing, you know?
This is where I think COVID is wreaking the most havoc on these numbers. The formula doesn’t know the external factors from last year, but we do. The season was weird, and I don’t trust the results. Teams were constantly shuffling lineups due to personnel losses. The schedule itself was a jigsaw, with games getting shifted due to COVID related cancellations, leading to even more bizarre practice schedule.
Nothing was normal about last year for any team. And that’s the 10,000 foot view. When you look at LSU specifically, the number look even less reliable from a predictive standpoint.
And, if you remember, by November, LSU only had 20 upperclassmen in its entire two-deep. That’s almost unfathomable attrition caused by the twin punches of early entrants and the COVID epidemic. LSU was trotting out the equivalent of a JV team last season.
Due to the quarantine restrictions, the team didn’t have a spring or a fall practice to install a new defense from a new coordinator with almost entirely new personnel. Shockingly, it took some time for the defense to find its footing.
That’s even assuming we had a competent defensive coordinator, which is still a bit of an open question. LSU’s defense was awful last season. They allowed over 8 yards per play in five different games, getting lit up by offensively challenged teams such as Mississippi St, Mizzou, and Arkansas.
To put this in perspective, LSU had only allowed 500+ yards 10 times and 600+ just thrice since 1996. They allowed 500+ six times this season, three times going past the 600 yard marker.
LSU allowed over 40 points in half of its games last season. The team allowed 4.8 yards per rushing attempt and 9.7 yards per pass attempt. These are horrific, cover-your-eyes numbers.
The 7.26 yards per play LSU allowed would be the worst in SEC history, if Vanderbilt hadn’t allowed 7.28 last year, nosing past LSU at the finish. The previous record was 6.75. LSU utterly smashed it. LSU allowed 492 yards per game, which also would have been an SEC record, had not another 2020 defense been worse. Take a bow, Ole Miss at 519.
The 2020 LSU defense was historically awful. Not just by LSU standards, but by the entire SEC. So the question is, was that on the players or on everything else? Or was it, most likely, a combination of factors? From a prognistaction standpoint, the only thing about 2021 that will be similar is the players. The coaching staff and (hopefully) the virus will be gone.
A formula can only be as accurate as its input. Garbage in, garbage out. And the problem for predicting the 2021 season is that the 2020 numbers cannot be relied upon at all. The best thing to do is simply write it off. Last year was a mulligan for everyone.
The season was so illegitimate that the NCAA didn’t even count it against your eligibility. It was the Zombie Season. But the break is over. Hopefully this is year is back to normal for everyone, and that means the return of accountability.
LSU isn’t going to be judged based off of 2020, it’s going to be judged by the prior standard. That means a return to the top ten. Bill C might grade LSU on the curve of 2020, but LSU fans won’t.