Catallarchy’s College Football Top 25 - Week 8

For all games up to 10/22, here are the updated rankings:

Rank Team Week 7
1 Texas (1)
2 USC (2)
3 VaTech (3)

4 Georgia (5)
5 UCLA (7)
6 Wisconsin (9)
7 Alabama (10)
8 LSU (13)
9 WVU (8)
10 Miami (11)
11 Texas Tech (4)
12 Florida State (14)
13 Colorado (16)
14 Auburn (6)
15 Cal (17)
16 Oklahoma (20)
17 Louisville (18)
18 Toledo (21)
19 Texas A&M (27)
20 Ohio State (24)
21 Oregon (22)
22 Boston College (19)
23 Penn State (23)
24 Northwestern (29)
25 Iowa (12)

Dropped Out:
28 Nebraska (15)
38 Virginia (25)

Other Catallarchy Teams:
31 Georgia Tech (32)
62 Miami (OH) (70)
68 Western Michigan (79)
106 Kentucky (103)

Commentary below, full rankings on page 2, and my subjective reordering of the given Top 25 on page 3 *PLUS* newly added Strength of Schedule ratings (and an alternate Top 25 now reflecting SOS).

I was away this weekend at a wedding, so I didn't get to see many of the games. Boo! Though I was at the Maryland-VT game, and woo, that was some tension until the 4th quarter.

What was up with that UNC-UVA game? 7 to 5? :???:

The big thing is that I figured out how to work strength of schedule into the rating system. It was a long, hard slog typing in each team's schedule, but it is done, and it seems to work pretty well. What I did was take the wins for each team on a given team's schedule and take the average. This average is then compared to the median of all schedule averages divided by the range. For Division 1-AA teams, I put a zero. My SOS does not recognize wins from 1-AA teams, thus teams that so schedule are at a programmed disadvantage.

Thankfully there weren't any surprises in the "rankings with SOS", though I was surprised at how weak VT's SOS is comparatively. Oklahoma has the strongest schedule in the land (unless I mistyped something) which is also surprising. Here is how the rankings go with SOS added in:

Rank - Team - Rank w/o SOS
1 Texas (1)
2 USC (2)
3 VaTech (3)
4 Georgia (4)
5 Wisconsin (6)
6 Oklahoma (16)
7 Alabama (7)
8 UCLA (5)
9 WVU (9)
10 Miami (10)
11 LSU (8)
12 Colorado (13)
13 Texas Tech (11)
14 Ohio State (20)
15 Florida State (12)
16 Northwestern (24)
17 Penn State (23)
18 Texas A&M (19)
19 Michigan (27)
20 Auburn (14)
21 Cal (15)
22 Louisville (17)
23 Iowa (25)
24 Oregon (21)
25 Missouri (29)

Dropping Out:
26 Boston College (22)
29 Toledo (18)

Oklahoma's massive schedule strength equals much love when SOS is added in. The Big 10 benefits quite a bit as well. Question to the readers: Should I go forward in weeks 9-12 with the SOS version of the poll, or keep the SOS-less version (or run both)?

Full Rankings:

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

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

Rank Team Computer
1 Texas (1)
2 USC (2)
3 VaTech (3)
4 Georgia (4)
5 Miami (10)
6 Alabama (7)
7 UCLA (5)
8 Wisconsin (6)
9 LSU (8)
10 WVU (9)
11 Texas Tech (11)
12 Florida State (12)
13 Colorado (13)
14 Auburn (14)
15 Cal (15)
16 Oklahoma (16)
17 Louisville (17)
18 Toledo (18)
19 Texas A&M (19)
20 Ohio State (20)
21 Oregon (21)
22 Boston College (22)
23 Penn State (23)
24 Northwestern (24)
25 Iowa (25)

Conference Ranks by Wins Per Team:

Ranking Conf WPT
1 Big 12 - 4.67
2 Big Ten - 4.64
3 Pac-10 - 4.30
4 ACC - 4.08
5 SEC - 4.00
6 MWC - 3.89
7 Big East - 3.88
8 Notre Dame - 3.58
9 Conf USA - 3.25
10 MAC - 3.08
11 WAC - 2.67
12 Sun Belt - 2.38
13 I-A Indep - 1.67

Conference Ranks by Average Team Ranking

Ranking Conf Average Team Rank
1 Big 12 (34)
2 Big Ten (41)
3 ACC (44)
4 Pac-10 (49)
5 Big East (49)
6 SEC (50)
7 Notre Dame (54)
8 MWC (68)
9 Conf USA (77)
10 MAC (79)
11 WAC (79)
12 I-A Indep (93)
13 Sun Belt (96)

Ranking by Individual Team Strength of Schedule:

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

Share this

Why is Penn State being

Why is Penn State being ranked lower than Ohio State, a team we beat? It's even worse in the non-SOS version, where we're ranked behind two teams we beat already.

Sorry, the with-SOS version.

Sorry, the with-SOS version.

Stormy- It comes down to

Stormy-

It comes down to preseason bonus and Ohio State's stronger schedule strength. Penn State was mediocre last year and so didn't "finish in the money" so to speak (#84 out of 117) while Ohio State finished #28. If the final positions in 2004 were reversed, at this point Penn State would be #9 and Ohio State would be #27. So already the poll is showing love to Penn State relative to OSU.

And while PSU's SOS is currently #13 in the nation (by my calculation, which is continuous and not iterative like the other computer polls), Ohio State's is #4. Reverse the SOS and you reverse their positions in the SOS poll.

I have to think that by the

I have to think that by the time you get this late in the season, you should phase out last season's results. That may be useful for pre-season and early week rankings, but perhaps you should simply phase out last season's results and replace it by SOS?

A good idea might be a sliding scale where SOS and last season's rankings become more and less important as the season goes on. I.e. at week 1, SOS has no effect on rank, and last-seasons has full effect. As the season progresses, about the middle each effect are at half-strength, and by the end of the season last-season's result has no effect, and SOS has its full effect?

Brad- Good idea! I was

Brad-

Good idea! I was thinking the same thing when i was mulling over the early season effects- SOS would have an insane whipsaw effect at first until enough games are played to make things stable, and that worried me.

Right now, I had the preseason negatives phase out after week 4 so that anyone who wasn't positive had no millstone about their neck (I've already taken out preseason minuses in the course of trying to fix the Notre Dame 'problem', or at least remove anti-ND bias). I'll fiddle about in the test sheet and see what a sliding scale does (e.g. seeing what the current Top 25 would look like with diluted preseason bonuses). If I did your way though, should I keep the preseason negatives in all season or keep the status quo of removing them by week 5? I need the negatives in to anchor the poll at first, but I agree that last year's (bad) performance shouldn't be determinant.

Is SOS based on the entire

Is SOS based on the entire schedule, or only the portion played so far? e.g. does Penn State's SOS include Wisconsin now, or not until we actually play them?

Stormy- As I mentioned in

Stormy-

As I mentioned in the other thread, it includes Wisconsin now even though PSU hasn't played them yet. It is for this reason that I suspect that putting SOS in at the beginning of the year would tend to increase volatility, since it includes everyone, but upon further reflection I think that the equation I have in place compensates for big or small numbers, so it may not be an issue. I think Brad's addition of phasing out preseason effects as the year goes on (and phasing SOS effects in) has a lot of promise. Without the preseason bonus, OSU may well fall behind Penn State, given OSU has 2 losses to PSU's 1.

Notre Dame behind Rutgers?

Notre Dame behind Rutgers? Weak.

Luca- Not as weak as Notre

Luca-

Not as weak as Notre Dame's schedule. (snerk)

Seriously though, the teams on ND's schedule have 3.58 wins on average, which is less than the Division 1-A average. The only teams so far that are worth a crud on their schedule, they've lost to. I'll grant that they have heart and tenacity and are playing way above their heads, but I've removed every bit of anti-ND bias in the system and still there's no love. I'm pretty sure now that the data and performance thus far do not warrant a high ranking for ND. They also have 2 losses.

Rutgers' SOS is only 3 places behind Notre Dame's, and Notre Dame's conference (also its SOS) is weaker than the Big East. Both are on even playing fields as far as preseason bonus goes (neither have one), so voila, Rutgers has a higher ranking. I'll have to go home and check the spreadsheet to see how close it is; I imagine its pretty close given the records of the teams. [Edited to add- the difference is 0.08]

And finally, the rankings are not predictive. :) ND vs. Rutgers heads up, I pick ND in a heartbeat. The rankings are backward looking based on performance.

The Maryland game convinced

The Maryland game convinced me that the hokies are for real. Even after not scoring from 1st and goal from the 1 twice, missing a short FG, Vick throwing 3 INTs, getting a couple of penalties that extended possession for MD, we still blew them out. Right now we're favored by 14 over BC which is astonishing, almost as much as us being 6-1 against the spread this year. There's just something about this team that makes me think we can't lose. I haven't had that feeling since '99. Ordinarily, I'd be a tad nervous about playing BC, but they don't stand a chance despite their ranking. It'll be a magical night in Blackburg Thursday night.

I still think there's a decent chance that if we beat BC, Miami, and FSU, we will catch up with Texas enough in the human polls for the computers to put us over the top.

I think you should keep the SOS in there. Every computer ranking should factor in SOS; the only question is how much.

I agree about the SOS. Oh,

I agree about the SOS. Oh, and I wouldn't discount BC. We'll see how it goes =)

Stormy, you should like

Stormy, you should like this-

I ran a test using Brad's suggestions, though I tweaked it thus: The negatives are removed by week 5 (when there is an average of 5 games played or more throughout Div 1-A), but the bonuses drop by an 1/11th every week. SOS is phased in by 1/11th of their "real" value each week, with no effect preseason and barely any in game 1. I found that my hunch was correct- leaving SOS undiluted from week 1 caused weird volatility.

What I found was that the behavior of USC and Texas was unchanged- after Oklahoma's opening loss, Texas has been atop of USC. The interesting bit was that VT was #4 until Week 4, and then in Week 5 drops back to #4 as Wisconsin jumps over, then back to #3 the next week after Wisky loses.

But Penn State jumps into the Top 25 at Week 5 to #15 (from #26), showing the release of the last bit of negative is decisive. Then PSU goes to #8 in Week 6, Down to #17 in Week 7 after the loss to Michigan, and then back up to #9 in Week 8. Also note that in Week 7 they're still ahead of Ohio State and Northwestern, two of your peeves with the current model. ;)

Notre Dame remains at #39, 3 spots behind Rutgers. [/Luca]

Here is the Top 25 for week 8 with the Brad model in place:

Rank Team
1 Texas
2 USC
3 VaTech
4 Georgia
5 UCLA
6 Alabama
7 Wisconsin
8 Miami
9 Penn State
10 LSU
11 Texas Tech
12 WVU
13 Oregon
14 Ohio State
15 Boston College
16 Colorado
17 Florida State
18 Northwestern
19 Oklahoma
20 Missouri
21 TCU
22 Texas A&M
23 Michigan
24 Stanford
25 Cal

David, I agree with

David,

I agree with Jonathan; I'm not discounting BC, I'm just factoring in the following:

1. Thursday game. Tech is awesome on Thursdays.
2. Night Game. Tech plays much better at night.
3. Home Game. Tech plays MUCH better at home.
4. Revenge Game. BC embarassed VT with a last second go ahead score on Senior Day in '03.
5. VT comes off of shaky play in MD. Coaches & team will refocus.
6. BC coming down cocky. Lost only to FSU in close game.
7. BC is slow, even on astroturf.
8. VT is fast, on a heavily crowned & moist grass field no less. (see MD game)
9. VT's field is flat and grass - disadvantage for BC, major plus for VT

1, 2, and 3 all multiply for VT. 4, 5, 8, 9 add for VT. 6, 7, and 9 subtract from BC.

The result can be summed up by a paraphrase of Austin 3:16 - "[The Hokies] just whupped your ass!"

Re: BC. The last time VT

Re: BC. The last time VT lost on Thursday night was 1995 against BC. It was the only home game that I didn't attend during my four years as I was busy studying for a thermodynamics test the next day. Knowing the game was scheduled the night before the exam, I asked the professor on the first day of class if we could change the date, and she said it didn't make any sense to do so because we could always study extra earlier in the week. I couldn't find fault with her argument even though I didn't like it. Who was she to try to encourage us to plan ahead?

I ended up watching it on TV anyway. It was Jim Druckenmiller's coming out party. Cornelius White dropped a pass on the BC 3 on 4th down that would have given VT a chance to win.

Brian, I think the "Brad

Brian,
I think the "Brad model" is working well... I don't follow too many teams outside the Big Ten, but I can't say I'd find any of their rankings to be highly "wrong", and Penn State should be up around the #10 spot, not #23, so I think that helped them out quite a bit. Of course, when my Boilermakers pull off the tremendous upset over PSU this weekend, we'll have to see what that does to them :lol:

For that suggestion, can I get a nice underline for Purdue? Not like anyone cares this year, but I'm hoping for big things next year!

Brad, Sure I'll underline

Brad,

Sure I'll underline Purdue, if they ever crack the top 25 again. :D

I figure I'll go with the modified SOS version with the caveats that there is a major methodological(?) break from Week 8 to 9.

I like the new system. I

I like the new system. I asked for an inch and got a yard.

Now if I can just get the Big Ten Refs to give me two seconds back. ;>

Stormy- Glad you approve. :)

Stormy-

Glad you approve. :) The odd jumping of PSU is likely an artifact; I didn't bother (nor did I have) to put the week-to-week changes in historically, but rather just added a full week's worth of results at a time. This is important because in the normal week-to-week, some teams have played more games than others, and that will have a non-trivial effect on their point total. So PSU probably wouldn't have done the bungee jump bit up and down but rather been a bit smoother in the drop and rise (I don't think *that* many people lost ahead of PSU). As long as my ratings get the Top 5 reasonably correct (and the top 3 now mirror the BCS poll, bwahaha), I'm happy, with the rest of the top ten to a slightly lesser extent. Beyond that it gets highly subjective anyway.

The VT defense looked

The VT defense looked impressive tonight, especially at the end. Granted, the BC quarterback made two really bad decisions near the end of the 4th quarter, but he was forced into tough situations by the VT line.

Yeha, they just got better

Yeha, they just got better as the night progressed. What impressed me the most was the final rushing stat- 27 yards net for BC! Andthat's with their genetic mutant offensive line. I was expecting them to rip off a few big ones early in the game before getting shut down, but the VT D-line really got it done. Equally impressive was almost 500 yards of total offense against one of the nation's top D's. Marcus and the receiving corps were on fire last night. Hallelujah, we've been waiting for a legitimate passing game for lo these long years, and finally we have one that goes all over the field for short, medium, and long yards (:grin:).

Now for the big one. Miami. :sweat:

Jonathan, I recalled this

Jonathan,

I recalled this comment from the third week of football:

"Why midwestern football is boring: lack of team speed. They simply don’t have the athletes like those in the SEC and ACC. It’s boring trenches-centric football. Plus, I think the game venues and atmospheres in the southeast are much better."

You have to be surprised so far with the offensive production coming out of the Big Ten. Northwestern and Michigan State have two of the best QB's in the nation. The spread offense is killing most Big Ten defenses, and OSU is really the only defen team in the Big Ten this year. Wisconsin, MSU, Northwestern (and ND) all have more yards and points per game than any east coast teams. I'm really happy with the way this year has panned out. Brad may be upset with his Boilermakers, but Big Ten football has been very exciting to follow.