After an early season break, I decided to make another go at comparing Football Outsiders’ F/+ Ratings, a statistical analytics approach to rating college football teams, to the highly erratic and sometimes suspect AP Poll. We are now half way through the CFB season, and nearly every team has six or seven games worth of statistical data in the books. All preseason bias has been removed from S&P+ and FEI, so we can get a better feel for how each team is playing.
Here are the F/+ Top 40 teams, with their component S&P+ and FEI ratings along with their AP Poll rankings. There are some curious anomalies that I’ll point out.
And here's some music to listen to as you ponder these numbers. GoT for the win.
Observations from Week 7 data
1) Top 10. Let’s start with the most obvious: the Buckeyes are #1. You may not believe that statement, but according to advanced analytics, tOSU has played the best, most consistent football in the nation through Week Seven in terms of efficiency, explosiveness, and overall bad-assery. But just barely. They are merely 0.30% better than Alabama in F/+ Percentage. The rest of the F/+ Top 10 closely resembles the AP Poll, with most teams landing within three places of their ‘human’ rankings.
A couple notable exceptions:
tOSU is ranked five places lower in the AP Poll than F/+.
Notre Dame is ranked four places lower in the AP Poll.
Wisconsin is ranked eight spots higher by the AP Poll.
My take-away is the pollsters are still over-penalizing teams for an early loss (OSU and ND) to good teams (OU and UGA), and over-rewarding undefeated teams who haven’t played an overly difficult schedule (WI).
2) Underrated. I’m buying a couple of teams that the polls do not give enough credit. UCF is being held back by pollsters because they’re in the American Athletic Conference, and they weren’t preseason ranked. F/+ has them as the #12 team in the country, because they are dominating their competition. If the Knights can escape Navy this weekend, their last test will be USF to finish the season. I would not want to face this team in a NY6 bowl game the way they are playing. UCF is not 2016 Western Michigan.
Despite two losses, Auburn and Stanford are Top 20 teams in F/+. They both have games remaining against quality opponents ranked above them (‘Bama and UW & ND respectively), so they have a chance to move up in the rankings with wins. Auburn faces Arkansas this weekend, while Stanford has a bye.
Iowa State is another underrated, two-loss team. They’ve lost by a combined 13 points: a 10-point loss to Texas and a 3-point OT loss to Iowa. Outside of the Texas game, where the Cyclones only scored seven points, they are averaging over 40 points per game. It remains to be seen if they can keep this up with Texas Tech, TCU, WVU, and oSu coming up.
FSU is the best three-loss team in the nation, statistically speaking. They are rated #22 in F/+, despite losing their starting QB and three games. But look more closely at those losses. A week one, 17-point loss to ‘Bama after losing their QB. The ‘Noles held ‘Bama to 24 points: the only other team to hold the Tide below 40 points this year is Texas A&M. FSU’s two other losses were by six to NC State (#19 in F/+) and by four points to Miami. The biggest tests remaining on the ‘Noles’ schedule are at Clemson and at Florida (only because it’s a rivalry game). It’s entirely possible FSU finishes the season with only three losses.
3) Overrated. In addition to the aforementioned Badgers, the AP Poll has several statistically poor performers ranked this week. The most egregious of these is Memphis, which sits at #25 in the AP Poll, but statistically is #59 in F/+. That’s a 34-place difference. More on that later.
Other overrated teams include: LSU, which is #38 in F/+ but snuck into the Top 25 with a win over Auburn; USF at AP #16 but F/+ #27, SDSU which fell to AP #26 and sits at #36 in F/+, and lastly AP #27 Texas A&M which is just outside the F/+ Top 40 at #43.
4) Noteables. There is very little separation between the Top 11 teams in F/+ raw percentage. From tOSU down to UW the deviation from the mean is about 2 points. So even though tOSU is at #1 and OU is #6, they’re only separated by 2.3 percentage points in F/+. In layman’s terms, “On any given Saturday.” The separation is a little more distinct in S&P+ with tOSU and ‘Bama clearly ahead of the pack.
The B1G is the best performing conference in the Power Five. The conference has six teams in the F/+ Top 25 (five in the Top 20), with tOSU and PSU in the Top 5 and four B1G East teams in the Top 20 (MSU and TTUN). The next best conference is the ACC with five in the Top 20 (six if you add ND), but they don’t start until Clemson at #7. The SEC and Big 12 have two teams each in the Top 5 and Top 10, respectively, but only four total from each conference in the Top 25. I see this as a potential problem for the B1G – conference cannibalization is a very real possibility. I’m hoping the separation between tOSU/PSU and MSU/TTUN is sufficient to prevent this. But the B1G East is the toughest division in CFB right now.
The American Conference is the best conference in the Group of Five, and it isn’t even close. UCF (#12), USF (#27) and Houston (#37) are in the Top 40 of F/+. The Mountain West has SDSU at #36 (one spot above Houston) and Boise State at #39. After these come SMU (American, #42) and Navy (American, #46), before any other Group of Five team makes an appearance in the Top 50: Southern Miss (C-USA, #48) and UTSA (C-USA, #49). In 2016, the American finished with six teams in the Top 50 of F/+; the rest of the Group of five had seven total. In 2015 the American had five Top 50, and the rest had seven. The American is 6-8 in bowl games versus Power Five teams since 2013 and 4-5 in the CFP era including a NY6 Peach Bowl win. For comparison, the Mountain West is 2-6/1-4 over the same periods with a NY6 Fiesta Bowl win, the MAC 0-4/0-3 with a NY6 Cotton Bowl loss, C-USA is 2-2/1-1 (no NY6 bowls) and the Sun Belt Conference has yet to play a Power Five team in a bowl game. If the CFP ever expanded to eight teams (which I hope it doesn’t), the American has a case for an automatic bid just like its former self, the Big East, had to the BCS bowls.
AP Top 25 Match-ups
Penn State (AP 2, F/+ 4) vs. TTUN (AP 19, F/+ 17). This should be what everyone expects, a defensive battle. In S&P+, TTUN is the #2 defense, while PSU is #8. The difference is on offense, where PSU is #13 and TTUN is #76. That’s right, our rival is in the bottom half of the country in offense. Vegas has the line at PSU -9.5; S&P+ is predicting PSU 29 – TTUN 18. Penn State’s strength on offense is rushing; TTUN’s strength on defense is passing. If Franklin establishes a ground game and sticks to it, they should win.
USC (AP 11, F/+ 15) at ND (AP 13, F/+ 9). This is precisely why there should be a law against the AP & Coaches polls prior to mid-October. Vegas set the number at ND -3.5, despite what the pollsters think. And it’s not just because the Irish are at home. S&P+ predicts ND 32 – USC 26. The Irish have the #1 rushing offense according to S&P+, the Trojans are #34 at defending the run. In the passing game, USC is #27 on offense, but ND is #6 defensively. Enough said. I like the Irish by at least a touchdown.
UPSET ALERT: Memphis (AP 25, F/+ 59) vs. Houston (UNR, F/+ 37) tonight on ESPN. Memphis found a cheat code last week (5 turnovers) and shut down Navy’s triple option in the second half. The 3-point win was big for the Tigers knocking off then AP #25, and it catapulted Memphis to #25 with their 5-1 record. Despite all that, F/+ actually dropped Memphis two places, because offensively they just aren't playing that well. Vegas has Memphis as a 2.5-point road underdog, and S&P+ agrees with Houston as a 3-point favorite. Go Cougs!
For anyone interested in seeing a complete breakdown of the Vegas spread, S&P+, & F/+ projections, here’s Bill Connelly’s week 8 projections with a link to the mother of all spreadsheets.
