Catallarchy\'s College Football Top 25 - Week 4

For all games up to 9/24, here are the updated rankings:

Rank - Team - (Previous)
1 Texas (1)
2 USC (2)
3 VaTech (7)
4 Wisconsin (8)
5 WVU (9)
6 Cal (6)
7 Georgia (11)
8 Oklahoma St. (10)
9 LSU (5)
10 Virginia (13)
11 Texas Tech (12)
12 Auburn (16)
13 Florida State (15)
14 Nebraska (17)
15 Michigan State (20)
16 Colorado (3)
17 Louisville (4)
18 Minnesota (22)
19 Alabama (23)
20 Ohio State (30)
21 UCLA (21)
22 Az. State (26)
23 Utah (32)
24 Iowa (18)
25 Oklahoma (24)

Dropped Out:
26 Tennessee (25)
28 Purdue (14)
33 Georgia Tech (19) (Sorry about that! :twisted:)

Other Catallarchy Teams:
66 Miami (OH) (70)
97 Kentucky (91)
103 Western Michigan (109)

Commentary below, full rankings on page 2, and new for this week is my subjective reordering of the given Top 25 on page 3.

Another great week, though being that I was at the VT-GT game this saturday I didn't get to watch the rest of the field play. USC spotted Oregon 13 and then rolled over them. Each week I get less skeptical, though they still haven't played a legitimate offense yet. Colorado and Louisville, by virtue of losing, are now more in line with where I think they should be.

Minding the point in the comments from last week, I went in and changed the mechanics slightly for Notre Dame. Since I came up with my poll to automatically mimic the most famous biases of the sports writers, I figured it's in keeping to have a special rule for Notre Dame put into the formula. I made the conference for ND (and the other independents) to be all of D1-A rather than simply a pseudoconference of themselves. Second, I wiped the negative preaseason penalty from ND. That only gets ND up to #44, but thats also a 20 point or so jump. That's the extent of my special rulings for ND.

Two things puzzle me- why Oklahoma is still hanging around in my poll, and why my poll seems to hate Florida. Preliminary answers: Oklahoma is riding its preseason ranking and high conference power to keep it afloat. Florida was 7-4 last year, which killed it preseason in my poll, and the SEC is next to last in conference power among BCS conferences. So, both for the same reason in opposite directions.

I do like that Louisville and Colorado lost but didn't drop out. Ahh, the poll has ripened! Still, I think you have to be Top 10 or 15 to take a loss and stay ranked, still. I also like that VT is back to #3, in league with the human voters, and stand by my Texas #1 ranking. USC looks really good, but I'm still skeptical. I'm also pleased that my earlier prediction of GT not being too big a test for VT came true; I still think WVU is the biggest test so far, the peak since NC State. Speaking of those chokers, I have this sinking feeling they're going to the Pitt of the ACC- they will play us tough and hard and whether or not they beat us, they'll lose out in the end, screwing us in SOS or with a loss either way. Weasels.

And poor Clemson. Taking everyone to OT but rarely coming out on top.

This week's improbables can be noted in my subjectively resuffled Top 25 on page 3. :)

Rank	Team
1	Texas
3	VaTech
4	Wisconsin
6	Cal
7	Georgia
8	Oklahoma St.
10	Virginia
11	Texas Tech
12	Auburn
13	Florida State
14	Nebraska
15	Michigan State
16	Colorado
17	Louisville
18	Minnesota
19	Alabama
20	Ohio State
22	Az. State
23	Utah  
24	Iowa
25	Oklahoma
26	Purdue
27	Miami
28	Tennessee
29	Iowa State
30	Texas A&M
31	Toledo
32	U Conn
33	Georgia Tech
34	Florida  
35	Michigan
36	Northwestern
37	Clemson
38	Cincinnati
39	Boston College
40	Penn State
41	South Carolina
42	Missouri
43	Washington State
44	Notre Dame
45	Kansas  
46	Kansas State
48	Syracuse
49	Oregon
50	New Mexico
51	Baylor
52	Maryland
53	South Florida
54	Rutgers
55	Pitt
56	Indiana
57	UNC
58	Wyoming
59	Arkansas
60	Boise St.
61	Navy
62	Stanford
63	NC State
64	Vanderbilt
65	UAB
66	Miami (OH)
67	Akron
68	Oregon State
69	TCU
70	Bowling Green
71	Illinois
72	Fresno State
73	Ole Miss
74	Northern Illinois
75	North Texas
76	Memphis
77	Southern Miss
78	Air Force
79	BYU
80	Wake Forest
81	Hawaii
82	Mississippi State
83	Marshall
84	Arizona
85	Nevada
86	Tulane
87	Colorado State
88	Troy State
89	Ohio  
90	La Tech
91	Tulsa
92	SMU
93	Duke
94	Eastern Michigan
95	Kent State
96	San Diego State
97	Kentucky
98	Utah State
99	Middle TN State
100	Washington  
101	La Monroe
102	La Lafayette
103	Western Michigan
104	Arkansas State
105	Central Michigan
106	UNLV
107	Houston
108	Army
109	UCF
110	ECU
111	New Mexico State
112	Rice
113	San Jose State
114	Temple
115	Ball State
116	Buffalo
117	Idaho
118	Fla Intl
119	Fla Atlantic

Brian's Subjectively Reordered Top 25

Assuming my poll has it right with who the Top 25 should be but not exactly where, here's my opinion:

Rank Team
1 Texas
3 VaTech
4 Georgia
5 Michigan State
6 Cal
7 Oklahoma St.
9 Texas Tech
10 Virginia
11 Auburn
12 Florida State
13 Wisconsin
14 Nebraska
15 WVU
16 Minnesota
17 Ohio State
18 Alabama
19 Colorado
20 Louisville
22 Az. State
23 Utah
24 Iowa
25 Oklahoma

Share this

Do comment writers count?

Do comment writers count? Cause I went to Penn State.

Ryan, The poll doesn't take


The poll doesn't take into account MOV, how one did vs. the spread, or any of those other fancy predictive things. :) It works primarily on wins vs. losses. Win more, you move up, win less you don't. It also has a strong (in the early weeks) preseason factor. Louisville won big last year and until this weekend hadn't lost this year, and is playing in the Big East which oddly enough still has an edge in Strength of Conference (wins per team compared against the median WPT of all conferences- above the median is a plus, below a minus). Also, despite the horrible loss, it remains to be seen if its a fluke or if Louisville has indeed been unmasked. In light of that uncertainty, it would be premature to say "out with Louisville" as I think they're better than most CFB teams, thus should be in the Top 25. Where exactly remains to be seen.

I'll recap the formula in general terms:

Each week's win has progressively more points potential.

Win points(* % of teams that won that week) + Relative wins (more wins than average = +, less = -), + Relative losses (less losses than average = +, more = 1) + Strength of Conference (conference's WPT relative to the median, +/-) + preseason bonus/penalty (based on final finish in '04) * Games Played adjustment (if a team has played more games than average their score is discounted, less than average their score is enhanced, to try and minimize the implicit penalty for taking a bye week).

Roll it all together and then sort by highest total. For next year I'm thinking about lowering the preseason bonus, it seems to be a bit too much bias put into the system early.

I came up with this way of rating CFB teams when there was the controversy about SOS and MOV in the computer polls for the BCS. I wanted to see if I could make a ratings system that had neither but could still give realistic results, et voila.

I'm a hardcore (HARDCORE)

I'm a hardcore (HARDCORE) Louisville fan, and frankly, I'm surprised we're still in your poll. I'm not 100% sure how you've got this thing working, but we got beat down (lost by 31 points) to a team that we were twenty point favorites over. That is very, very bad.

Stormy- The Catallarchy


The Catallarchy teams are the (to my knowledge) alma maters of Catallarchy authors that play Division 1-A football. Miami of Ohio is where some of my homies are at, studying Economics, so I throw that one in. :)

How does a team get declared

How does a team get declared a "Catallarchy Team"?

Greg (3), I'm toying around

Greg (3),

I'm toying around with a separate measure of conferences, which would be by average rankings in the poll, though to put it into the point totals would be a circular equation. The SEC is not #5 in terms of average ranking, I believe.

Greg (2) - As for why the

Greg (2) -

As for why the SEC is low, its Wins Per Team is just slightly better than the PAC-10. Since conference play is going to be a wash in the end, its mostly driven by OOC play. The Big 10 and Big 12 have been loading up on OOC the first few weeks and won most of them, so they've got a very big lead at the moment. However at week 4 the differences are not so big anymore so the SOC component is starting to fade as a major driver.

Greg, I changed the 05


I changed the 05 formula to have the win pts discounted by the % of teams who won that week, to reduce some of the churn from teams winning (I want to reward the winners, but my poll is already super harsh on losing and I wanted to reign that in a bit). I'm now noticing that the preseason effect is now far too powerful in relation to the points the teams can win on the ground- last year the proportion of the preseason bonus to total points was a max of around 10% to 6.7%, which I thought was reasonable on the margin. I did a test last night of the system by assuming every team continued doign what they did on week 3 (i.e. just copied the columns over :grin:) and then looked to see what the totals would be. Since I've discounted the win points the total for the #1 team at the end was only 26% of last year. With the preseason bonus unchanged, that makes the preseason effect massive (for the #1 team it was a bit less than 25% of the total) and thats a bit too much bias, as we're seeing re: Florida, who is definitely a Top 25 team regardless of the lack of option attack in Meyer's Triple Option O. (UF's got good D, and you gotta respect that; besides, they can go vertical too).

The problem I saw in the test though was that if I, say, halved the preseason effect (which is roughly what multiplying the win points by the win % per week gets me) the "compensation for games played variance" factor goes way up and LSU is #1 by dint of not having played any games, and if I fiddled with that other oddities popped up. If I put the win points back up to full, Florida still didn't make it (other teams popped up), so it looks like its not a simple fix at the moment.

The bottom line is that, within reason, undefeated teams should be above 1 loss, 1 over 2, 2 over 3, etc. The extra considerations (last year's record, relative wins, strength of conference) are there to help tweak within the various categories and elevate some 1 loss teams above undefeateds, etc. However, the marginalia of the poll are starting to become the core drivers, which is mostly giving the right results at the top but there are still the serious deviations to deal with.

Guess I should've read the

Guess I should've read the comments first. But SEC last in conference power? How can that be, when Georgia, Florida, LSU, Tennessee, and Alabama are all top 20 teams, and even Vandy will be 5-0 on Saturday?

And where the hell is

And where the hell is Florida? You've got the top 3 teams, but as soon as #4 comes around, we're nowhere to be found. UF gets no respect.

Brian, How does the

How does the preseason and last-year's record affect things going forward? Looking at the games I've seen, Michigan State and Minnesota look a lot stronger than Wisconsin, and they're in the same conference to boot. I'm sure when Wisconsin loses a few games, they'll drop behind, but I'm wondering whether the preseason effect is designed to go away later in the year on its own...

SEC next to last? are you

SEC next to last? are you kidding?
glad your poll has nothing to do with anything at all...

Will- Not kidding at all. As


Not kidding at all. As of Week 5, the SEC's Wins Per Team is 2.75, making it tied for 4th place with the ACC (both conferences have 33 wins total).
The average rating for the SEC is in a 3 way tie with the Pac-10 and the Big East at #44 (the average rank of teams in the SEC) with the ACC at #42.

For comparison, the Big 12's average rank is an eye-popping #25, with 3.08 wins per team (37 wins for the conference), 2nd place being the Big 10 with an average rank of 30 and 36 wins for the conference (11 members means its WPT is higher than the B12).

Subjectively speaking, there are some very, VERY bad teams in the SEC. Kentucky is awful, Arkansas is worse than awful. Miss. State, Southern Miss, USC, are all not very good. What you have are the usual suspects: Georgia, Auburn, Tennessee, Alabama, and Florida and LSU as somewhat suspect-but-top 25 material. The bottom half of the SEC is pretty much garbage this year. Whereas the Big 12 and Big 10's teams have been winning their OOC games.

I stand by my rating system's conclusions re: the SEC right now.