College BlogPoll Ballot: Week 6 Draft

by Hooper on Oct 12, 2009 9:00 AM EDT Comment 5 comments

Here is the first draft for Rocky Top Talk's ballot for the college blogpoll. Discussion of the numbers in the computer follows after the jump, including one significant change to the computer system that (I think) will make the numbers a little more reliable.

Rank	Team	Delta

1	Florida
2	Alabama
3	Texas
4	Virginia Tech	2
5	Boise State	2
6	TCU	3
7	Southern Cal	3
8	Iowa	3
9	Nebraska	7
10	Cincinnati	2
11	Ohio State	1
12	Oregon	12
13	Penn State	2
14	South Florida
15	Miami (Florida)
16	Brigham Young	9
17	LSU	7
18	Utah
19	West Virginia	4
20	Georgia Tech
21	Wisconsin	1
22	Arizona	4
23	Kansas	10
24	Oklahoma	2
25	South Carolina
Last week's ballot

Dropped Out: Auburn (#17), Connecticut (#19), Stanford (#21).

Quick Notes:

A lot of teams are tightly bunched together. As you can see in the numbers below, many teams are statistically equivalent, which is the computer's way of conceding that six games isn't enough for definitive separation. But hey, here we are.
Colley is out. After waiting far too long (again) for Colley to update his rankings, I imported the strength of schedule numbers from Sagarin instead. They operate in the usual manner: the 1-AA teams are filtered out, the scores are then scaled with an average of zero (i.e. positive is above average) and a standard deviation of 1. But Sagarin actually updates in a prompt and timely manner.
Sagarin appears to be more reliable. The one sticky point about Colley is that he lumps 1-AA teams into groupings. In other words, if Charleston Southern and Northwestern State get lumped together, then Florida and Baylor have their schedules intrinsically tied together through those schools. This makes little sense. And having compared the effect of Sagarin's and Colley's SOS metrics on the computer, I think I will probably stay with Sagarin going forward.
Weights are changed. Partly due to the switch to Sagarin and partly due to the increasing data, the weights are now: W/L = 25; SOS = 15; Pass Efficiency Defense (PED) = 9; Run Defense (RD) = 8; Third Down Offense (3DO) = 7; Total Defense (TD) = 6; Pass Efficiency Offense (PEO) = 5; Opponents' Points Per Game (OPPG) = 3; Total Offense = 2; Third Down Defense (3DD) = 1.

Computer Standings

For convenience, only the top 35 are shown. I will probably put the entire list out later on in a dedicated post.

Rank	Team	WL	SOS	PED	RD	3DO	TD	PEO	OPPG	TO	3DD	W SUM
1	Florida	1.721	0.262	2.251	1.166	2.039	2.359	1.976	2.324	1.742	1.780	156.901
2	Alabama	1.721	0.857	1.905	1.649	-0.040	2.077	0.741	1.515	1.065	2.072	136.874
3	Texas	1.721	-0.798	1.372	2.081	1.031	1.880	0.885	1.183	1.632	2.546	114.014
4	Virginia Tech	1.076	2.119	1.264	0.357	0.020	0.663	1.592	0.829	0.141	1.541	110.243
5	Boise St.	1.721	0.046	1.475	0.842	0.078	1.238	1.883	1.262	0.980	0.839	108.275
6	TCU	1.721	-0.254	1.112	1.255	0.390	1.625	0.913	1.156	0.623	1.351	101.722
7	Southern Cal	0.947	1.246	1.669	1.668	-1.468	1.792	0.241	2.032	0.837	1.271	100.567
8	Iowa	1.721	0.751	1.704	0.123	0.423	0.742	-0.171	1.073	-0.258	0.093	98.728
9	Nebraska	0.947	0.409	1.845	0.646	0.908	1.244	0.703	2.112	0.416	0.641	94.724
10	Cincinnati	1.721	-1.081	0.884	0.762	1.291	0.599	1.639	1.342	1.447	-0.075	84.621
11	Ohio St.	1.076	0.515	1.293	1.127	-0.146	1.269	-0.019	1.581	-0.658	0.332	80.898
12	Oregon	1.076	1.244	1.603	0.504	-1.264	1.127	-0.282	1.006	-0.304	0.608	78.465
13	Penn St.	1.076	-1.175	1.212	1.290	1.942	1.523	0.473	1.824	0.787	0.817	78.359
14	South Fla.	1.721	-2.196	1.467	0.779	0.739	1.408	2.025	1.926	0.562	1.339	75.956
15	Miami (FL)	0.947	1.699	0.071	-0.216	0.361	0.495	1.147	0.042	0.182	0.297	74.184
16	BYU	1.076	-0.150	-0.186	0.560	2.850	0.206	1.771	0.277	1.352	-1.111	73.982
17	LSU	1.076	1.100	1.119	0.231	-0.184	0.471	0.291	1.249	-1.361	-0.546	72.580
18	Utah	0.947	0.405	1.396	0.042	0.282	0.785	0.242	0.705	0.951	2.162	70.012
19	West Virginia	0.947	-0.370	0.562	1.228	1.394	0.668	0.981	0.042	1.139	0.460	67.347
20	Georgia Tech	1.076	1.212	-1.260	0.097	1.609	-0.515	1.930	-0.343	0.774	-0.929	64.110
21	Wisconsin	1.076	0.731	-0.341	0.379	1.369	0.303	0.439	-0.233	0.703	-0.275	64.018
22	Arizona	0.173	1.707	0.510	0.913	0.334	1.112	-0.011	0.042	0.695	-0.632	63.761
23	Kansas	1.721	-1.621	-0.150	1.082	1.385	0.023	1.409	0.785	2.277	0.772	62.433
24	Oklahoma	0.173	0.121	0.832	1.917	-0.046	1.518	0.695	2.059	1.265	0.516	62.272
25	South Carolina	1.076	-0.121	0.538	0.234	-0.075	1.135	0.285	0.741	-0.140	0.029	51.219
26	Notre Dame	0.947	0.814	-0.407	0.006	0.031	-0.799	2.090	0.015	1.476	-0.100	50.675
27	Idaho	1.076	-0.029	-1.022	0.916	1.884	0.028	0.710	-0.210	0.684	-2.314	49.282
28	Auburn	1.076	0.152	0.676	-0.551	0.449	-0.266	0.830	-0.432	1.797	-0.075	47.880
29	Central Mich.	1.076	-1.487	0.524	0.134	1.558	0.868	1.090	1.227	0.274	0.706	45.544
30	Missouri	0.947	-0.278	0.519	0.264	0.092	0.290	0.538	0.811	0.461	-0.670	42.035
31	Wake Forest	0.431	0.217	-0.235	0.456	1.363	-0.135	1.523	0.277	0.763	-0.765	41.355
32	Stanford	0.431	0.827	-0.152	0.353	0.526	0.073	0.494	0.476	0.409	0.048	41.354
33	Tennessee	-0.215	0.957	1.160	0.660	-0.275	1.266	-0.009	0.586	0.222	0.820	41.188
34	Pittsburgh	1.076	-0.305	-0.730	0.675	-0.049	0.261	1.667	0.675	0.160	0.004	40.831
35	UCLA	0.173	1.204	1.454	0.362	-1.076	1.052	-1.192	0.891	-1.558	0.911	39.062

Visually, here's the distribution of scores among all 1-A teams:

Week_6_score_distribution_medium

Here are just the top 25:

And a look at the conference averages:

There you go. Should any changes be made? If so, what? This is the draft, but the final ballot can have human overrides if there is sufficient reason to do so.

0 recs | Comment 5 comments | Share

Story-email Email Printer Print

Comments

I had a feeling that UT might jump into the top 25

Your top 12 only has a single difference from my Mumme Poll ballot: Nebraska over Miami. Given that Nebraska’s only loss is to the now dominating Virginia Tech (in a much closer fashion than Miami’s), it’s hard for me to argue that point. Also, at #15, Miami isn’t too terribly far down.

Wisconsin has still beaten a bunch of mediocre teams and lost (badly) to one good one. I guess a ranking is fair.

Similarly, an OU ranking can be argued, despite the lack of signature wins. They haven’t lost to a bad team, and their losses were without Bradford.

The only one that jumps out to me as highly problematic is Arizona. But because I defended Oklahoma, I can’t say too much here. Unless we make a Bradford-adjustment, those two have almost identical resumes.

by Incipient_Senescence on Oct 12, 2009 12:32 PM EDT reply actions 0 recs

That last subject line was a typo

I meant to say “top 35”

by Incipient_Senescence on Oct 12, 2009 12:32 PM EDT up reply actions 0 recs

Understood on UT.

But with the Colley SOS, they very nearly made top 25. I am really beginning to think that there’s a problem with the way that Colley handles 1-AA teams. Teams that played the 1-AA schools have an “opponent” on their schedule who is typically 0-6 right now (based on that conglomerating of schedules thing), which tanks their SOS. UT, on the other hand, didn’t play a 1-AA team, so they conversely get their SOS skewed toward the high end since SOS is a zero-sum game. Sagarin’s method fixes that.

But I’m pleased to hear that it’s coming in line with intuitive agreements. As for Arizona – their SOS comes out way high by any metric, which is seriously buoying them right now. I think that it’ll calm down over the next few weeks.

by Hooper on Oct 12, 2009 12:53 PM EDT up reply actions 0 recs

Colley blows. :)

Keep using Sagarin.

Eric Berry for sending the guy who wins the Heisman spinning 720 degrees in the air at the podium - or for intercepting it and returning it to where it rightfully belongs

by Graysnail on Oct 12, 2009 6:24 PM EDT up reply actions 0 recs

Also, one guess as to who I think is ranked too low.

Did the computers really love Missouri that much?

Eric Berry for sending the guy who wins the Heisman spinning 720 degrees in the air at the podium - or for intercepting it and returning it to where it rightfully belongs

by Graysnail on Oct 12, 2009 6:28 PM EDT reply actions 0 recs