Before I begin, I wanted to mention that I’ve updated my LSU/Florida Rivalry entry. LSU is now one game shy of a .500 record at home against the Gators. Also, feel free to submit your top 25 rankings. If I don’t know who you are, leave a link to your blog and mention who your favorite team is.
These are probably best understood by starting with the strength of schedule. Ole Miss, Tennessee, and Georgia have top-8 schedules, and there are six other SEC teams among the top 30 schedules. Schedule strengths are just based on opponents’ records and opponents’ opponents’ records, so that’s pretty straightforward.
So that’s why the top 3 teams are all in the SEC. Georgia has two losses, but the #8 schedule despite that means the teams they’ve played (other than North Texas anyway) hardly had any other losses, and the ones they had were very high quality.
LSU has the #37 schedule now, but only having one loss, that gives them a lot of credit. It’s also the fact that LSU hasn’t had a bye week or FCS (I-AA) opponent yet.
This is the top 5 if you average the ratings by playing week:
2. Florida St.
Stanford is in a similar situation to LSU’s. If you average their rating by playing weeks, they fall to #9.
As the season progresses, Stanford’s loss to Utah, for instance, might hurt them a lot more. But at the moment, the Utes have only lost to an undefeated team and a one-loss team. I have the system set up so that a team like Florida or LSU who finishes with one loss would be ahead of a team like Boise St. (or this year, maybe Fresno St.) with no losses at the end of the year. With only roughly half of the season played, the differences can be a lot more pronounced, as teams like Oregon have most of their quality opponents still to play. The Ducks did move up from #19 last week.
Putting aside the spectacle of the Oregon offense, human voters look at a team like Oregon and know that they play in a competitive conference, so they don’t really worry about the fact that they haven’t played the most difficult teams yet. But before they played Washington, their rating was roughly equal to Fresno St.’s. Human voters never would have put Fresno St. so high because they would have felt obligated to keep them there.
So that brings up one of the weaknesses with the human rankings: they are less willing to move a team down without a loss. If Team A is right behind team B and Team A beats a better team than Team B three weeks in a row, Team A will move ahead. But human voters will often wait for Team B to lose before making such a change. Beating a team they’re favored against like Oregon did against Washington doesn’t move a team up nearly as far in the human polls. Of course, Oregon was too high in the human polls to move up so far, but imagine they were Oregon St. with the same schedule and outcomes so far instead but with less impressive scores.
Another example of the differences is that LSU will play Furman on October 26, followed by a bye week. In my ratings, that will be an opportunity for a lot of teams to pass them. Human rankings usually have the opposite effect. Those are two weeks when they’re relatively safe from moving down in the rankings.
I’ve had some people familiar with my ratings conclude they don’t mean very much until later, but that’s not true. They mean just about as much as they mean at the end of the season—the most accomplished teams at this point will be higher just like the ones at the end of the season—but they just don’t factor in any information about what might happen later on. There isn’t as much of a bell curve at this point. Some teams are severely untested, and others have been tested a lot. There will only be a few teams on the extremes at the end of the season, and the teams with the best schedules are usually about 8-4 at best. The teams with the worst schedules will either have good records (with relatively poor ratings in comparison) or they’ll be among the worst teams in the ratings.
The real weakness earlier is not inaccuracy but volatility. For instance, Washington would have been in the top 5 had I started the computer ratings a week earlier; but after two losses and Arizona’s loss to USC (Arizona is Washington’s best win), the Huskies are not even in the top 25. Let’s say, for instance, South Carolina wins the rest of their regularly scheduled games (to get into the top 5) before losing the SEC Championship and a bowl game. They’re still going to be in the top 25 in the last rating.
I tried to minimize the big jumps in going from subjective to objective rankings, and it turns out there were only a few. I had initially moved Washington up a couple of weeks ago thinking that would help transition, so I created the need for a big correction in the other direction. I also should have moved Missouri up more incrementally, but I didn’t think they would beat Georgia, to be honest. The others are pretty normal at this point, and movement should be more gradual as the season goes on.
rank / team / prior
1 Alabama 1
2 Missouri 22
3 LSU 9
4 Clemson 3
5 Florida St. 6
6 Ohio St. 4
7 Stanford 2
8 Oregon 7
9 Va. Tech 14
10 Miami 10
11 Baylor 21
12 Georgia 5
13 UCLA 17
14 Oklahoma 11
15 Auburn 20
16 Louisville 18
17 Texas Tech 13
18 S Carolina 15
19 Fresno St. 19
20 Houston —
21 TX A&M 16
22 N. Illinois 24
23 Michigan 12
24 Mich. St. —
25 Utah —
Out of rankings: (8) Washington, (23) Florida, (25) N’western