theknightswhosay

Early Preview of Computer Ratings

In College Football, General LSU, Preview, Rankings Commentary on September 22, 2015 at 3:36 PM

I decided to do my first preliminary computer rankings now that most teams have actually played someone in the FBS who in turn played other teams who have played FBS opponents and it’s possible to get a computer rating on everyone.

Before I reveal more, a couple of disclaimers: I don’t start using a full or almost full computerized system until early October, so my official rankings are still the ones that were posted on Sunday; but I thought it would be interesting to see how the teams shaped up at this point. I like to try to get a smooth transition from subjective to objective. Usually I will still move a couple of teams for the first or second top 25 list (although the ratings on my ratings site are always 100% objective).

The transition is not always easy. I got someone mad at me when my attempted transition backfired with Missouri in 2013. I kept thinking the Tigers would lose (and I also thought the added points they got for beating Vanderbilt would go away, but Vandy had one of their best seasons in recent memory instead), so I kept them down around #20. Then the weekend where I fully moved to the computer rankings, they got another big win (this time over previously unbeaten Georgia, who had already defeated four P5 opponents in the first five weeks) and shot up to #2. So the transition process involves some guessing and gambling, but I still think it’s better than going from 100% subjective to 100% objective in one week.

This was really surprising to me since they usually don’t align early on, but the prelimary #1 is Ohio St., the same as my subjective #1. The prelimary 24th and 25th teams are BYU and U. Miami, the last two teams in my subjective top 25 (but in different order). So to that extent I’m encouraged with my attempts at objectivity.

There many discrepancies in the middle of the two rankings though. One example is Florida St., who is 11th in my rankings but is tied with Arizona, Clemson, and Houston for 48th  in my computer system.  This is because although the Seminoles have beaten three FBS opponents, none of THOSE have beaten an FBS opponent. Boston College did beat a couple of FCS teams, and that would normally result in FSU getting points by extension, but it doesn’t because those FCS teams haven’t beaten ANYONE in FBS or FCS.

In the next month, the Eagles will play Northern Illinois, Duke, and Clemson, so they may turn out to be a strong win for FSU in the near future; they just aren’t now. For its part, FSU only plays Wake Forest in the next two weeks, so they mostly have to rely on teams they’ve already played to win to pick them up or they may fall out of the top 25 when I move to the purely objective system.

Northern Illinois fumbled away a chance at a huge upset in a sloppy game on Saturday, but the Buckeyes still look good on paper.

The top four in total rating are:

Ohio St.

Michigan St.

Texas A&M

Notre Dame

However, it looks a little different when you divide the teams by playing weeks:

Ohio St.

LSU

Michigan St.

West Virginia

Anyway, without further ado, here is the full computer top 25 as it would look if I did the fully objective system now:

Team rating rating/week subjective
1 Ohio St. 0.3874 0.1291 1
2 Mich. St. 0.3615 0.1205 3
3 Notre Dame 0.3309 0.1103 12
4 Texas A&M 0.3294 0.1098 7
5 N’western 0.3032 0.1011 18
6 Oklahoma 0.2908 0.0969 16
7 TCU 0.2890 0.0963 6
8 Iowa 0.2797 0.0932
9 Alabama 0.2669 0.0890 8
10 LSU 0.2583 0.1291 4
11 Utah 0.2468 0.0823
12 Indiana 0.2362 0.0787
13 Ole Miss 0.2335 0.0778 2
14 W. Virginia 0.2245 0.1123
15 Florida 0.2237 0.0746
16 Missouri 0.2121 0.0707
17 UCLA 0.2045 0.0682 9
18 Syracuse 0.1980 0.0660
19 Temple 0.1898 0.0633
20 N.C. St. 0.1879 0.0626
21 Texas Tech 0.1738 0.0579
22 Ohio U. 0.1561 0.0520
23 Toledo 0.1437 0.0718
24 BYU 0.1431 0.0477 25
25 U. Miami 0.1333 0.0444 24

Obviously LSU is higher (they’re 10th the other way) because their Week 1 game was cancelled, but the difference between 2 games and 3 games is statistically much larger than the difference between 12 games and 13 games will be. So right now, I think that’s definitely worth considering. Also, just to note, you have to go down five decimal places to separate tOSU and LSU under the average-week calculation. Almost every other pair of consecutive teams is separated in the second or third decimal place.

West Virginia is sort of a statistical fluke at the moment. That’s the short version anyway. They beat Georgia Southern, who beat Western Michigan. I don’t think Ga. Southern or Western Michigan are good teams, but the former looks like a good win for WVU right now. Western Michigan is the only FBS opponent the Eagles (that’s Ga. Southern’s mascot) have played, and the Broncos played Michigan St., one of the best teams, so they seem to have a really good strength of schedule.

I will mostly be discussing the overall score though.

I also mentioned in my last rankings blog that looking at this year’s results alone, not factoring in last season or any personal perceptions, Texas A&M should be in the top five, so I’ve confirmed that as well.

I’ve laid out several reasons I’m skeptical of the Aggies, but they won’t get a ton of points before playing Alabama in about a month, so their computer rating should come closer to aligning with the subjective ranking anyway.

I know I don’t have Notre Dame nearly that high, but as far as my ratings know, the Irish beat Virginia by 70 instead of 7, so that’s one reason I have them a good bit lower. Maybe they’re even better with their third QB in just a handful of games going back to the end of last regular season. Sort of like Ohio St. was in the playoff in January.

Notre Dame-UMass SHOULD be dull, but that Notre Dame-Clemson game in two weeks will be huge in determining who goes where in the first computer ratings.

There is a bit of an issue with certain FCS opponents counting for too many points right now, but I’ll discuss that immediately afterward and then add some further comments.

Something else worth considering is that wins over unbeaten FCS teams (who have actually beaten a Division I opponent anyway) count for a good bit of points right now. The only teams that show up in the top 25 who are influenced by this are West Virginia (so that’s another reason WVU is so high in the weekly average), Iowa, North Carolina St., Ohio U., and U. Miami. This also helps #26 Baylor.

Cal-Berkeley and Missouri also get high points from their FCS opponents that will diminish over time. So if you just want take out those points, the top 25 would conclude thusly:

10 Iowa
11 Utah
12 Indiana
13 Ole Miss
14 Florida
15 UCLA
16 Missouri
17 Syracuse
18 W. Virginia
19 Temple
20 Texas Tech
21 N.C. St.
22 Toledo
23 BYU
24 Ohio U.
25 Penn St.

I swear I didn’t rig the system to make LSU’s upcoming opponent (Syracuse) look good this week. Wake Forest is otherwise undefeated, and Central Michigan has a very strong schedule right now having played Oklahoma St. as well.

As an aside I found amusing, Les Miles made a point of talking up Central Michigan (and the MAC in general) during his Monday press conference because apparently his wife is a graduate of CMU. After Syracuse, who beat the Chippewas in overtime last week, LSU will play MAC opponent Eastern Michigan, so that may have been a secondary motivation of bringing up the MAC’s success.

These things will sort themselves out though. If Wake Forest wins more than a few games, I’ll be surprised, and it’s hard to have a good strength of schedule out of the MAC. This year might be an exception for the MAC though after the way teams like Toledo, Northern Illinois, and Bowling Green have performed against major programs.

Just to go down the list though, Oklahoma and TCU are about right. Waiting for TCU to do something though. Oklahoma may go down a bit if Tennessee loses to Florida, so we’ll see about that. The Sooners also have a bye this coming week, so that will allow some teams to pass them up so that should help make it a more natural transition.

The Horned Frogs play Texas Tech, which looks good statistically right now due to some early competition.

I’m not a believer in Iowa. They’re getting a lot of points for their FCS win. Iowa St. probably will never count as a good win this year (likely a very weak one since the Cyclones went winless in the Big XII last season), and Pittsburgh will probably be mediocre.

Alabama and LSU are about right, with the proviso I mentioned about the latter having only had two playing weeks.

Utah has a good-looking schedule right now, and Michigan helped them out by beating two FBS opponents. The Wolverines are just out of the top 25 at #29.

Indiana is another fluke. The SIU and FIU wins will lose the lustre, I’d expect. Western Kentucky might be all right though.

Then we get to Ole Miss. It’s not exactly a secret that the Rebels didn’t play anyone before Alabama. Again, the fact that they won 70-7 or whatever doesn’t matter.

Unbeaten Florida knocked off unbeaten Kentucky on Saturday. No guarantee that will mean anything in a couple of weeks, but congrats to the Gators.   It’s something to have a 29-game winning streak over another SEC program regardless.

Missouri, UCLA, and Temple are other teams who got away with getting close wins, although there is a very small adjustment when you win a close game (≤3) at home.

North Carolina St. is another fluke. I don’t think we’ll look back at the end of the season and remark at the greatness of their non-conference wins over Troy and Old Dominion. Same thing with Ohio U.’s wins over Marshall and Idaho. Old Dominion and Marshall may turn out to be all right, but they may lose to a few more teams apiece instead of staying otherwise undefeated.

I think it’s fairly clear why Toledo, BYU, and Miami are there. Arkansas does have two losses now, but they got that FBS win and the Hogs’ losses are to two undefeated teams (I mentioned Texas Tech earlier). Nebraska also has two losses from apparently good teams (BYU and U. Miami).

Upcoming blogs

Finally, I wanted to announce something.  Other than the weekly rankings, this blog hasn’t had too much structure to it.  I believe next week I’m going to start a series called “SEC Wednesdays”.  This will be a time when I will go into more detail about past and upcoming SEC games rather than having it clog up space in more general blogs.  Of course there were a few big games last weekend and I mentioned them already.   But after the games this week, I will keep my comments short when it’s an SEC game until the SEC Wednesday blog.

The basic structure of the week will be Sundays (starting in October) the new ratings will come out, with the top 25 rankings and discussion blog coming out Monday generally.  Then on Thursday or Friday I usually do some kind of preview of the upcoming week for other teams or perhaps I might do something specific to LSU.

The last couple of weeks I did “conference reports”.  I will probably do that one more time next week before that series will go on hiatus.  There aren’t enough inter-conference games to sustain it after the first few playing weeks.  I try to do a final one for the regular season before the conference championship games and then another after the bowl games.

Advertisements
  1. […] mentioned previously I started doing my preliminary computer calculations.  They account for approximately 50% of the […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: