With the non-Big Ten portion of the season officially over, this is the last week we have to stare at a 0-0 team topping the SP+ rankings. Ohio State gets underway in six days, and we'll finally get to find out if Ryan Day's Buckeyes are as good as SP+ seems to think they'll be. If they aren't, they'll cede control of the rankings to two teams in a hell of a battle at the moment.
Clemson and Alabama have traded off the No. 2 and No. 3 spot in the rankings for a few weeks now, and after two impressive performances -- Alabama pulled away from No. 4 Georgia in the second half for an easy headliner win, while Clemson could have scored 100 on poor Georgia Tech if it really wanted to -- they are basically in a tie. Among teams that have played thus far, Clemson is the only one with a top-5 offense and defense, while, ho hum, Alabama only has an offense that grades out even better than Joe Burrow and LSU last year.
It's early, and things can and will shift, but the Crimson Tide hanging 564 yards and 41 points on what is still easily the top defense in the country pushed them into lofty territory. For comparison, they managed 83 more yards, four more points and 0.9 yards per play more than what Joe Burrow and LSU did to Georgia last year. And this Georgia defense is more experienced and, on paper, better.
Alabama's current offensive SP+ rating of 48.0 adjusted points per game grades out in the 99.8 percentile. In the last 15 years, only the 2017 and 2018 Oklahoma offenses have graded out better. (LSU was at 99.6.) We'll see if the Tide can continue to clear this incredibly high bar, but what they've done so far this season is incredible.
What is SP+? In a single sentence, it's a tempo- and opponent-adjusted measure of college football efficiency. I created the system at Football Outsiders in 2008, and as my experience with both college football and its stats has grown, I have made quite a few tweaks to the system.
SP+ is intended to be predictive and forward-facing. That is important to remember. It is not a résumé ranking that gives credit for big wins or particularly brave scheduling -- no good predictive system is. It is simply a measure of the most sustainable and predictable aspects of football. If you're lucky or unimpressive in a win, your rating will probably fall. If you're strong and unlucky in a loss, it will probably rise.
This week's movers
A quick procedural note: Because of all the oddities of 2020 -- the diminished offseason practice time, the fact that we're whistling along with teams frequently missing one-quarter or one-fifth of their scholarship players for a given game, etc. -- I designed SP+ to be a bit more cautious out of the gates. I set it up so that preseason projections were phased out more slowly than normal while we figured out the predictive effects of all these oddities.
After some study and tinkering over the weekend, I'm confident in making things a little less conservative in that regard. That phasing-out process is now closer in line to how I've done things in previous years, which means that for teams that have played a handful of games now, their ratings are more reflective of how they've played to date.
That should have a positive effect on predictions, but it does bring one oddity to the table: it means teams moved up or down this week based not on Saturday's results but on previous weeks'. Florida State, which had played nowhere near its projections to date, fell quite a bit despite beating North Carolina. West Virginia, meanwhile, seemed to get quite a healthy reward for simply beating up on Kansas. If a team made a decent jump during an off week (hello, Oklahoma State) or made a movement that wasn't in line with results to date, that's probably why.
Five teams jumped more than 10 spots this week:
Liberty: up 22 spots from 77th to 55th
West Virginia: up 18 spots from 61st to 43rd
Army: up 14 spots from 58th to 44th
Marshall: up 13 spots from 53rd to 40th
North Texas: up 11 spots from 106th to 95th
Two teams have separated themselves from the pack in Conference USA: No. 40 Marshall and No. 50 UAB. They beat two preseason contenders -- Louisiana Tech and WKU, respectively -- by a combined 72-31 on Saturday and are the only two top-75 teams in the league.
A few teams also moved down quite a bit.
Florida State (down 17 spots from 47th to 64th)
Georgia State (down 15 spots from 71st to 86th)
Syracuse (down 14 spots from 91st to 105th)
Duke (down 14 spots from 74th to 88th)
Tennessee (down 13 spots from 24th to 37th)
Georgia Tech (down 12 spots from 48th to 60th)
Auburn (down 12 spots from 14th to 26th)
Middle Tennessee (down 11 spots from 102nd to 113th)
Since I don't think many will be surprised that both Auburn and Tennessee fell quite a bit after Saturday's performances -- losses to South Carolina and Kentucky, respectively -- let's talk a bit more about Florida State.
There are two reasons for FSU's big fall. First, as mentioned above, preseason projections were propping them up considerably. The Seminoles had gotten absolutely smoked by Miami, and from the perspective of the predictive stats that go into SP+, they were lucky not to lose to both Notre Dame and Georgia Tech by a lot more than they did. They should have been in the 60s before this week, in other words.
Second, those same predictive stats saw the Noles' big win over North Carolina as more of a tossup. My postgame win expectancy measure -- which takes the key stats from a given game, tosses them into the air, and says, "With these stats, you could have expected to win this game X% of the time" -- saw a game UNC wins 62% of the time. FSU overachieved projections but didn't get a lot of credit for the win.
Why does SP+ hate Arkansas?
The first thing I did after running the numbers was check and Arkansas and see how much the Razorbacks rose after Saturday's win over Ole Miss. The answer: they didn't. That was a bit frustrating, as their story is probably the most interesting and enjoyable of the season thus far. So I dove into why Sam Pittman's 2-2 Hogs are still mired in the 70s, just barely ahead of in-state rival Arkansas State.
Extremely unfriendly preseason projections. Even with me phasing out projections a little more quickly now, they're still holding the Hogs in place a bit. Their preseason SP+ rating was minus-3.2 adjusted points per game, and four games in, that's still more than 50% of their overall SP+ rating.Florida has fallen 17 spots, from third to 20th
Auburn fell pretty far this week, making last week's tossup loss look less impressive. They get less of a boost for overachieving projections in that game.
As good as the defense was against an excellent Ole Miss offense, the offense was equally bad against a dreadful Ole Miss defense. Their defensive ratings rose this week, and their offensive ratings fell.
That damned post-game win expectancy again. It was 46% against Ole Miss, which means they only get so much credit for the performance.
All of these answers are unsatisfying, but stats are antisocial sometimes. If the Hogs keep overachieving their projections, SP+ will likely catch on at some point.