College BlogPoll Ballot, The Real First Draft

by Hooper on Oct 5, 2009 10:12 PM EDT Comment 7 comments

UPDATE: If you're interested in seeing the numbers with the increased weight for strength of schedule, take a look here. That list is the next iteration in this week's draft, and may need a bit more tweaking. A small bump to the Win/Loss weighting may be warranted, but that's up to you all. -[Note by Hooper, 10/06/09 7:40 AM EDT ]

Ok, the numbers are finally in. The moves I anticipated earlier did pretty much happen, as well as a few I didn't notice. Midmajors appear to have a tough road ahead (mostly), especially if midmajors include "Big East". Everything is after the jump.

Rank	Team	Delta

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

Dropped Out: UCLA (#14), Houston (#20), North Carolina State (#22), Georgia Tech (#24).

More motion than I expected...

TCU??? They have numbers that are very considerably above average, with the exception of strength of schedule. But basically, the undefeated record and the considerably better third down offense offsets their SOS difference with Southern Cal.
Kansas and Iowa. I'd feel better with these reversed, which is easily doable with a tweak to SOS weight. This may happen, depending on how much it appears to improve things. More on that below.
South Florida, Boise State, Cincinnati would all move down further if SOS is increased. Virginia Tech would move upward in a big way.
Oklahoma ahead of Miami despite a worse W/L record and SOS because their actual on-field metrics are still far and away better. They're very close, though (see below), so a small SOS bump would reverse that.
Georgia Tech is the next one to climb up if SOS is bumped, even though they're behind NC State. The difference is small and Ga Tech overtakes before either climbs above a current top 25 team.

THE QUESTION TO YOU IS: SHOULD THE STRENGTH OF SCHEDULE BE WEIGHTED MORE HEAVILY?

Most of the oddness that I see would be fixed by increasing the importance of strength of schedule. But take a look at the numbers below. When you read the table, remember that positive numbers mean above average (i.e. better) and negative numbers mean below average (i.e. worse). If you see a factor that seems to consistently improve the ballot by increasing it or decreasing it, leave a comment.

Oh, and look at this graphical representation to see how closely most of the top 25 teams are rated.

Week_5_score_distribution_medium

Basically, TCU through Virginia Tech are all within a reasonable margin of error of each other. So don't hesitate to fire away. I have no compulsion to switch Alabama and Texas, even though they are close, though.

Ok, the numbers (explanation here):

Rank	Team	WL	SOS	PED	RD	3DO	TD	PEO	OPPG	TO	3DD	W SUM
1	Florida	1.629	-0.357	2.333	0.999	2.361	2.113	2.063	2.124	2.316	1.358	149.597
2	Alabama	1.629	0.004	1.401	1.623	0.644	1.968	1.472	1.199	1.319	1.496	124.574
3	Texas	1.629	-0.098	1.233	2.001	0.931	1.395	0.901	1.089	2.237	2.347	122.892
4	TCU	1.629	-1.096	0.964	2.007	0.105	1.698	1.428	1.153	0.701	1.330	93.377
5	Southern Cal	0.889	0.611	1.619	1.615	-1.371	1.716	0.200	1.950	0.813	1.168	89.237
6	Kansas	1.629	-0.925	-0.015	1.737	1.438	0.608	1.093	1.315	2.084	1.367	88.578
7	Iowa	1.629	0.581	1.563	0.350	0.380	0.733	-0.213	1.328	-0.262	-0.155	88.359
8	South Fla.	1.629	-1.399	1.423	0.734	0.712	1.341	1.932	1.846	0.546	1.232	86.329
9	Penn St.	0.889	0.016	0.998	1.139	1.990	1.301	0.097	1.561	0.370	0.758	85.545
10	Boise St.	1.629	-1.164	1.430	0.795	0.088	1.175	1.795	1.199	0.951	0.758	82.904
11	LSU	1.629	0.610	1.282	0.469	0.301	0.441	0.469	1.147	-0.900	-0.560	82.296
12	Ohio St.	0.889	0.589	1.255	1.205	0.116	1.501	0.060	1.535	-0.152	0.121	81.915
13	Cincinnati	1.629	-1.033	0.854	0.716	1.233	0.552	1.558	1.276	1.404	-0.110	81.056
14	Virginia Tech	0.889	1.891	1.104	-0.033	0.041	0.162	0.742	0.681	-0.029	0.977	78.140
15	Nebraska	0.704	-0.140	1.248	0.492	0.872	0.996	1.061	2.157	0.960	0.316	72.316
16	Auburn	1.629	-0.432	0.966	-0.337	0.800	0.094	1.298	-0.018	2.104	-0.114	72.312
17	Wisconsin	1.629	0.052	-0.454	0.223	2.001	-0.194	1.063	-0.122	0.854	-0.583	68.073
18	Connecticut	0.704	0.266	0.457	1.544	0.273	1.814	-0.537	1.121	-0.249	1.795	67.437
19	Arizona	0.704	0.978	0.679	0.960	0.051	0.954	-0.119	0.409	0.520	-0.554	66.644
20	Stanford	0.889	0.557	0.119	0.527	1.129	0.392	0.645	0.888	0.489	0.492	64.240
21	West Virginia	0.704	-0.180	0.572	1.109	1.860	0.282	0.957	-0.335	1.357	-0.381	62.433
22	Oklahoma	-0.220	0.712	0.740	1.599	-0.083	1.495	0.930	1.930	0.721	0.105	60.388
23	BYU	0.889	0.012	-0.252	0.209	2.462	0.162	1.394	0.215	0.844	-0.989	58.544
24	Miami (FL)	0.704	2.008	-0.468	-0.427	0.496	0.224	1.010	-0.238	-0.136	-0.011	56.202
25	Oregon	0.889	0.674	1.385	0.236	-1.130	0.853	-0.154	0.784	-0.114	0.632	55.628
26	North Carolina St.	0.150	-0.341	-0.011	1.570	1.325	1.707	0.920	0.759	0.341	1.326	53.800
27	Georgia Tech	0.889	0.748	-0.621	0.254	1.062	-0.062	1.499	0.085	0.420	-0.054	52.842
28	UCLA	0.704	0.595	1.451	0.856	-1.225	1.061	-0.730	1.056	-1.228	0.711	48.375
29	Missouri	1.629	-1.876	0.490	0.096	0.598	-0.022	1.696	1.056	1.165	-0.515	43.974
30	Utah	0.704	-0.523	1.432	0.069	-0.384	0.854	-0.040	0.603	0.847	2.601	42.607
31	Wake Forest	0.150	1.284	-0.203	0.161	1.071	-0.139	1.024	0.500	0.458	-0.645	41.856
32	Boston College	0.889	-0.309	0.450	0.817	-0.629	0.865	0.242	1.069	-0.796	1.297	38.831
33	Idaho	0.889	-0.001	-1.092	0.632	1.792	-0.016	0.997	-0.251	0.414	-2.029	38.739
34	South Carolina	0.889	-0.751	0.605	0.527	-0.238	1.320	-0.121	0.888	-0.105	0.153	38.456
35	Mississippi	0.704	-1.306	1.752	0.443	-0.132	1.115	-0.070	1.671	-0.084	1.962	38.363

Don't let the decimal points freak you out if you're not a numbers person. In the W SUM column (the actual blogpoll rankings), everything up to the decimal point is really all you need.

0 recs | Comment 7 comments

Story-email Email Printer Print

Comments

One other note: stability is on its way.

The delta dropped from 250 last week to 130 this week (so far). One thing to note is that some teams don’t have the delta on the table above – Wisconsin, Stanford, West Virginia, and Oregon. Without these, the delta calculates to 109.

FYI

by Hooper on Oct 5, 2009 10:24 PM EDT reply actions 0 recs

Bump the weight. It will screw us up in 2-3 cases. Boise and Cincy are too low already and don’t need to fall any further. Arizona doesn’t need to rise any further. But that’s less of a problem than USF, Kansas, Virginia Tech, and Miami. Close, but I’d go with a bump. If you can override in the cases of BSU and Cincy, that’d be nice, but all computer polls have some anomalies in them.

by Incipient_Senescence on Oct 5, 2009 10:43 PM EDT reply actions 0 recs

Overrides are always in play.

Thanks for the second opinion.

And I should have mentioned that I’ll post the final Tuesday evening, so more thoughts are certainly welcome.

by Hooper on Oct 5, 2009 10:58 PM EDT up reply actions 0 recs

Wow, that's solid evidence

That there really are only three great teams in college football this year.

Everyone where I live wants to know why the computers hate Virginia Tech, who appears to have the second best SOS among the teams listed.

Rocky Top Talk

by Will on Oct 6, 2009 12:09 AM EDT reply actions 0 recs

In this order:

(a) The win/loss record puts them at a disadvantage against the undefeateds.

(b) Their defense is good, but they’ve played a lot of great offenses and it’s hurt their numbers. (That will even out in the back half of the season.)

(cee) The bump to strength of schedule that I mentioned isn’t shown in this chart. I will publish the updated numbers tomorrow, and I guarantee that the V-Tech crowd will be pleased. (Hint: think Princess Bride.)

Just remember: the teams from 4 through 14 are really, really close together. The numbers in the W SUM category tell the story better than the ordinal rankings.

by Hooper on Oct 6, 2009 12:18 AM EDT up reply actions 0 recs