Rankings Roundup after Week 2: AP, Coaches, SP+

Submitted by Blue@LSU on September 18th, 2023 at 8:42 AM

Again, the usual early-season caveats about SP+ apply. The preseason numbers are completely phased out by week 7 but they have less influence as we get more games under our belt. Here are the numbers after week 3.

COMPARING AP, COACHES, AND SP+ (RANKINGS)

It was a rough weekend with plenty of ugly games for the top-rated teams, but it didn’t really do much to shake up the rankings. 

A little bit of disagreement among the AP and Coaches with respect to Texas and OSU, but otherwise they are pretty much in agreement about the top teams. 

The big differences are between the humans and the SP+ model. SP likes OSU, Notre Dame, Alabama, and Oklahoma a lot more than AP & the Coaches. On the other hand, they are more down on Texas and FSU especially. This could still be a case of the computer models taking more time to catch up. But then again, maybe it knows something that we haven’t quite figured out: that Texas and FSU are not, indeed, back baby!

Colorado remains the big outlier, ranked 19 by both the AP and Coaches polls, but still sitting just a bit outside the top 25 according to SP+.

SP+ OFFENSE & DEFENSE (RANKINGS)

The SP+ continues to hold on to TAMU, Auburn, and Kentucky in the top-25, none of which appear in the AP or Coaches polls. Wisconsin just sneaks in at #25. 

Anyway, Michigan still sitting at #3 is SP+. It looks like next weekend is shaping up to be a big battle between SP #1 Ohio State and #4 Notre Dame.

As always, it’s fun to compare the offensive and defensive spreads. USC is probably most impressive here, but the Pac 12 teams generally have pretty large differences between offense and defense. Utah is the exception, a defensively-minded Pac team, so it's fitting that they're heading to the...Big 12 (?). 

Speaking of offensive and defensive spreads, don’t forget to take a look at the SP #s for the B1G. Woof.

Who else is looking forward to the first time that Iowa and USC match up in the new B1G? 

TRENDS (RANKINGS)

Last week I said it was probably more relevant to look at trends in SP+ right now as opposed to the raw numbers (which will probably take some time to catch up). With that in mind, here are the P5 teams that have risen at least 15 places in the SP+ since the preseason rankings came out.

Colorado checks in here, rising 18 places from their preseason rank. And look at Rutgers! 

Here’s the graph of teams that have dropped at least 15 spots since the preseason.

B1G schools account for 3 of the 8 biggest falls in the SP+ since the preseason.

SP+ OFFENSE AND DEFENSE (RAW SP+ SCORES)

Unlike the previous graphs, these are raw numbers, not rankings. 

  • Top-right: Good at offense and defense
  • Bottom-right: Good at defense, below-average offense
  • Top-left: Good offense, below-average defense
  • Bottom-left: Not so good at football

Still lots of B1G teams in the bottom-right quadrant. 

SP RANKINGS BY CONFERENCE (RAW SP+ SCORES)

Horizontal lines are the average (mean) SP+ scores for each conference.

Every week I keep waiting for the SEC average to drop. It looks like I've only managed to make it angry. 

 

Anything you find interesting?

McSomething

September 18th, 2023 at 9:06 AM ^

So, on average, the B1G is 4th out of the P5, with only the ACC being worse? Makes sense.

I wonder how far those horizontal lines would drop if the top 2-3 teams/conference were taken out.

Blue@LSU

September 18th, 2023 at 9:38 AM ^

I also prop up the ACC by including Notre Dame as a member. 

Just eyeballing it, though, I'm pretty sure the SEC would still be above all the others. That's been one of my gripes so far every week. The 2nd/3rd worse SEC teams are as good as the average teams from every other conference. I'm just not buying that (and I'm also not buying that South Carolina is the 2nd worst SEC team, fwiw). 

Champeen

September 18th, 2023 at 9:20 AM ^

I think its kind of BS that we did not gain on Georgia.  They were favored by 26.5.  They were losing 14-3 at half, and only won by 10.

We did not cover either, but we did a hell of a lot better than they did.  Missed our spread by 5.  They missed by 11.5, and the game was in doubt up until the 4th Q.

I figured we would at least inch closer to the number 1 spot, but it pretty much stayed exactly the same.

Perkis-Size Me

September 18th, 2023 at 10:34 AM ^

Our respective opponents were nowhere near the same quality. South Carolina may be 1-2 but they are much better (in my opinion) than their record suggests. By and large its the same team that beat two top-10 teams in a row to end last season, and Spencer Rattler is playing his way into being a really solid draft prospect. Maybe not a Day 1 guy, but certainly a Day 2 guy. Georgia was not overly convincing in that game and I'm of the mind that they just went on cruise control after they took the lead, but they faced a far better opponent. 

Bowling Green is an awful G5 team from top to bottom, and McCarthy had far and away his worst game in a Michigan uniform. Nothing about what Michigan did in this game makes me think they deserved to gain any ground on Georgia. Which is unfortunate, because if they played against Bowling Green the way they played against ECU and UNLV, they probably would've started getting some first place votes. 

None of the top-4 looked that good this weekend. Texas clearly had a post-Bama hangover, and FSU probably didn't even deserve to win, but because BC played one of the sloppiest, most undisciplined games I've seen in recent memory, they got to win anyway. 

Buffalowing Blue

September 18th, 2023 at 12:13 PM ^

UGA and S. Carolina are rivals.  We know anything can happen in those kind of games.  

I don't care if Michigan stays #2.  I'm not worried about their play so far but I wouldn't say they've been dominate either.  If the past 2 seasons have shown us anything it's comparing game results to other teams game results don't mean a thing.  Just get better and beat the teams on your schedule.

M-GO-Beek

September 18th, 2023 at 9:29 AM ^

Honest question- what is in the SP+ that likes OSU so much?  Does it include recruit rankings, or is there carryover from last year's team in the early rankings?

DenverMaize

September 18th, 2023 at 9:49 AM ^

This is from an old SB Nation article on what goes into S&P+

My S&P+ projections incorporate recent performance, recent recruiting, and returning production; early in the season, they carry significant weight, and they are phased out with increasing speed. After a team has played seven games, its preseason projections are completely phased out of the S&P+ equation.

agp

September 18th, 2023 at 12:02 PM ^

Professional in the world of sports analytics/modeling here. It's not weird at all! It provides a measure of stability and given the small N of football in particular, it's helpful to essentially fold predictions into each other. Given that S&P is prediction focused (this is good) anything that reduces error is generally helpful. 

jmblue

September 18th, 2023 at 12:08 PM ^

But what if the projection is off-base, or is based on production that doesn't happen  (e.g., a team is projected to have a dominant passing game but its star WR suffers a season-ending injury)?   Then the projection is just skewing the dataset.

OSU's projection is evidently based on the assumption that its passing game won't miss Stroud a bit, but that's far from certain at this point.  Only in the most recent game did they look anything like last year's OSU.

maquih

September 18th, 2023 at 1:27 PM ^

At the same time, teams are figuring it out throughout the season.  Maybe in week 1 the coaches haven't figured out how to use their freak athletes.  Having a bad upset loss in September might not be as valuable information as a massive talent advantage to predict a game in November.  

Ultimately youre asking a hypothetical when these professionals have years of experience at what makes their models as predictive as possible.

agp

September 21st, 2023 at 2:47 PM ^

If there are single season events that drive results up/down out of line with pre-season projections, those are incorporated into the model. If Steph Curry misses ten 3PA in a row, is he suddenly bad? The "program" itself feeds into these models as OSU has a massive surplus of talent, wins at a super high rate, and generally is highly productive. A bit of a QB wobble isn't going to systematically change those underlying factors. 

Blue@LSU

September 18th, 2023 at 12:29 PM ^

Yeah, I read something he wrote from a couple years ago saying he has tried phasing out the preseason components earlier. It didn’t really improve the model and the results just jumped around much more week to week. I wasn’t aware that he’ll keep them in all season, though. Thanks for the update. 

MGoOhNo

September 18th, 2023 at 2:27 PM ^

This college football season has been boring (and this is coming from a fan of even midweek MACtion)!

This early in the season, with the models yet to catch up to reality, I’m looking forward to seeing a couple of blue bloods (god that hurts to type) square off.

However, a pillow fight is still a pillow fight, and I’ve seen little from either ND or OSU to think this game is going to be an epic tilt.

ND homerism and strange things happen at ND probably means ND wins a tight one, and then both teams fade over the season…

MGoOhNo

September 18th, 2023 at 2:35 PM ^

Any game is hard to win when losing the turnover battle, but especially with 4 turnovers.

That said, this D is pretty stout.

My gut feels like the O gets it’s shit together because 3 TO and lackluster line play brings focus and preparation.

My, perhaps unfounded, concern is mobile QB who can also complete downfield passes exposes a somewhat undermanned or rusty defensive backfield.

Synful

September 18th, 2023 at 10:24 AM ^

From a pure trivia pov, who'd have thought that after the trends of the past many years that you'd see a time where there are four PAC schools ranked ahead of Alabama and a fifth directly on their heels.

club_med

September 18th, 2023 at 11:06 AM ^

Colorado definitely looks a bit rickety, but I am surprised at how negative SP+ is on them. Sanders has a 79% completion percentage and is 10:1 TD/INT. The defense does seem to play bend-don't-break and gives up yards, but they stop people in the red zone. Obviously some opponent Qs to be answered yet this weekend at Oregon.

PopeLando

September 18th, 2023 at 11:10 AM ^

So far, this year is shaping up to being a Standard Issue Harbaughffense Year (tm). Decent, but a step or two behind the big boys. (My theory is that the Harbaughffense did not adjust to the new clock rules, and is content to just score less. Which is just great.)

I'll be interested in seeing if we shift into overdrive during the B1G season. We're going to need to. Right now, we're very comparable to Georgia, but SP+ likes Ohio State very much so far.

How far can our #2 defense take us vs. their #4 offense, if our #16 offense is matched up against their #5 defense??

I know this is all SO FAR. Lots of football yet to be played.

1VaBlue1

September 18th, 2023 at 1:51 PM ^

If you're counting on Jim Harbaugh to speed up the pace of offense and generally throw the ball more to run up higher scores, then you are bad at math and cannot be helped.  Every September we tell ourselves that his offense is going to open up and use the weapons it has available in quick strike ways, and it'll be fun, high scoring, exciting, and fast!

And every November we repeat that Harbaugh's offense is run based and doesn't run up scores, and all is well.

MGoOhNo

September 18th, 2023 at 2:49 PM ^

I know clock rules and fewer possessions = higher variance, but I’m curious to see whether or to what extent being able to man ball at least one scoring drive per quarter at a 7-8 minute pace/drive and a D that limits explosives and makes opposing offenses slog to get points is helpful or hurtful under new rules.

I know minter has been aggressive with blitz packages and pressures in these initial games, but keeping the ball in front and inside to make opponents burn clock seems like a better strategy (at least in a game playing with a lead).

I’ve heard thoughts about changing O strategy given clock rules, but what about D strategy?