130-team S&P+ rankings, where Washington’s the next top-tier challenger

Below are this week’s S&P+ rankings.

A reminder: S&P+ is intended to be predictive and forward looking.

Good predictive ratings are not résumé ratings, and they don’t give you bonus points for wins and losses. They simply compare expected output to actual output and adjust accordingly. That’s how a given team can win but plummet or lose and move up.

Through five weeks, the S&P+ rankings are performing quite well, hitting 55 percent against the spread and 55 percent on the over/under point totals for the year.

As you would hope, the absolute error — the average size of miss between projection and reality — is shrinking each week as well.

If you’re interested in a decent résumé ranking of sorts, I encourage you to visit this post on strength of schedule. I created a Resume S&P+ ranking and will be updating it throughout the rest of the season.

Below, however, are the predictive ratings, the actual S&P+.

(You can find full unit rankings, plus a yearly archive, at Football Outsiders. The offense and defense pages will start getting updated in the coming weeks.)

A quick procedural note:

As non-conference play ends and conference play begins, the scoring margins tend to get closer on average. As a result, the overall spread of S&P+ ratings — which is distributed along the bell curve for scoring margins — tends to get smaller, too.

You’ll notice that Alabama’s S&P+ rating fell from plus-33.8 adjusted points per game to plus-30.3 despite obliterating UL Lafayette on Saturday. That “fall” is a product of the scoring adjustment, not anything the Tide did on the field. Their percentile rating remained 99.2 percent in each week.

Because of this, you’ll also notice that all the top conferences’ average ratings fell, while all the bottom conferences rose. Same concept there.

Okay, now that that’s out of the way ...

Your No. 5 Team of the Week: Washington

It’s painfully clear who the top four teams in the country — or at least, the four teams with the most proven upside — have been, and it’s basically who we expected: in some order, Alabama, Ohio State, Georgia, and Clemson. Alabama has been No. 1 since Week 2, and the other three continue to shuffle amongst themselves, but the most drama in the rankings has come just below the top four.

The top team not named Bama/OSU/UGA/Clemson has shifted.

• Washington began No. 4 but fell after losing to Auburn.
• Oklahoma jumped to second before settling back in near its No. 8 preseason ranking.
• Boise State’s early dominance produced a surge to No. 5, and the Broncos immediately crashed and burned at Oklahoma State.
• Mississippi State took the No. 5 spot from BSU and immediately got smoked at Kentucky.
• Michigan took over, damn near lost to Northwestern, and fell two spots. A blessing, really.

The Curse of No. 5 nearly befell the Wolverines, but now they’re out of the five-hole, and this week’s new “top teams not named Bama/OSU/UGA/Clemson” is an old name: Washington.

Chris Petersen’s Huskies have been absolutely dominant defensively — they’re currently second in Def. S&P+, behind only Auburn — and that has helped to paper over some concerns stemming from a merely good offense (26th in Off. S&P+). But they’ve won four in a row since falling to Auburn in Atlanta to start the season, and they get what should be a pretty easy shot at UCLA this weekend before a true curse-of-No.-5 test, the October 13 trip to Oregon.

The week’s top movers (good)

1. Tulane (up 21 spots, from 97th to 76th)
2. Virginia Tech (up 18 spots, from 53rd to 35th)
3. UAB (up 17 spots, from 83rd to 66th)
4. SMU (up 16 spots, from 115th to 99th)
5. Arizona State (up 15 spots, from 68th to 53rd)
6. Ohio (up 15 spots, from 104th to 89th)
7. Fresno State (up 13 spots, from 38th to 25th)
8. Troy (up 13 spots, from 86th to 73rd)
9. Army (up 13 spots, from 84th to 71st)
10. Five teams up 12 spots

S&P+ had a great week of predictions, hitting 62 percent against the spread with a paltry absolute error (the average difference between projected scoring margin and actual) of 11.4. Of last week’s 58 games, it had the projected scoring margin within five points for 20 of them. Great week.

But there are still quite a few teams it can’t get a read on yet. College football is never boring.

This week’s most confusing team: Virginia Tech. In four games this season, the Hokies have 1) won comfortably at Florida State, 2) destroyed FCS William & Mary, 3) lost by 14 points at Old Dominion, and 4) destroyed previously unbeaten Duke in Durham. They started 34th in S&P+, fell to 53rd after ODU, and now they’re right back to 35th.

Top movers (bad)

1. BYU (down 19 spots, from 59th to 78th)
2. Memphis (down 18 spots, from 21st to 39th)
3. Buffalo (down 16 spots, from 52nd to 68th)
4. UL-Lafayette (down 15 spots, from 99th to 114th)
5. Coastal Carolina (down 15 spots, from 91st to 106th)
6. Toledo (down 15 spots, from 54th to 69th)
7. Duke (down 15 spots, from 27th to 42nd)
8. South Alabama (down 14 spots, from 95th to 109th)
9. North Carolina (down 12 spots, from 78th to 90th)
10. Three teams down 11 spots

Despite a statistically fluky loss to Navy in Week 2, Memphis had been one of the steadiest mid-majors, beating inferior Mercer, Georgia State, and South Alabama by a combined 106 points. So naturally, the Tigers went to New Orleans to take on a 1-3 Tulane that had lost at UAB and just gotten pummeled at Ohio State ... and lost, 40-24. Sure. Memphis remains a top-40 team ... but now it’s a top-40 team that’s 0-2 in AAC play.

FBS conferences, ranked by average S&P+ rating:

1. SEC (plus-12.7 adjusted points per game, down 2.2 points)
2. Big Ten (plus-7.6, down 1.1)
3. Big 12 (plus-7.6, down 1.5)
4. Pac-12 (plus-5.4, down 1.2)
5. ACC (plus-5.2, down 1.0)
6. AAC (minus-0.1, up 1.5)
7. Mountain West (minus-3.7, up 1.5)
8. Sun Belt (minus-7.5, up 0.3)
9. Conference USA (minus-8.7, up 1.7)
10. MAC (minus-8.8, up 0.5)

Again, the scoring curve is the primary reason for the top conferences falling and the bottom conferences rising, but there was still movement within this movement.

The biggest movement: Conference USA is no longer the lowest-ranked conference. Apparently S&P+ viewed this weekend’s series of surprise results — MTSU upsetting FAU, Louisiana Tech upsetting North Texas, Marshall barely avoiding upset against WKU, plus Southern Miss holding its own at Auburn — as a sign of depth. And now the MAC is bringing up the rear.

---

A reminder: I have made a few philosophical changes in this year’s S&P+ rankings.

When I get the chance (so, maybe in the offseason), I will update previous years of S&P+ rankings to reflect these formula changes, too.

1. I changed the garbage time definition. S&P+ stops counting the major stats once the game has entered garbage time. Previously, that was when a game ceased to be within 27 points in the first quarter, 24 in the second, 21 in the third, and 16 in the fourth. Now I have expanded it: garbage time adjustments don’t begin until a game is outside of 43 points in the first quarter, 37 in the second, 27 in the third, and 21 in the fourth. That change came because of a piece I wrote about game states at Football Study Hall.
2. Preseason projections will remain in the formulas all season. Fans hate this — it’s the biggest complaint I’ve heard regarding ESPN’s FPI formulas. Instinctively, I hate it, too. But here’s the thing: it makes projections more accurate. Our sample size for determining quality in a given season is tiny, and incorporating projection factors found in the preseason rankings decreases the overall error in projections. So I’m doing it.
3. To counteract this conservative change, I’m also making S&P+ more reactive to results, especially early in the season. If I’m admitting that S&P+ needs previous-year performances to make it better, I’m also going to admit that S&P+ doesn’t know everything it needs to early in a season, and it’s going to react a bit more to actual results.

Basically, I’ve added a step to the the rankings process: after the rankings are determined, I go back and project previous games based on those ratings, and I adjust the ratings based on how much the ratings fit (or don’t fit) those results.

The adjustment isn’t enormous, and it will diminish as the season unfolds.

Testing this process for past seasons improved performance against the spread a little and, more importantly, decreased absolute error (the difference between projections and reality) quite a bit. I wouldn’t have made the move if it didn’t appear to improve performance.