RTT BlogPoll Computer Rankings: Week 11 Draft

by David Hooper on Nov 15, 2009 12:00 PM EST in College Football

Here is the rough draft for RTT's BlogPoll Ballot. As always, the rough is taken straight from the computer rankings. We will then use feedback and human judgment to make adjustments - and there are some adjustments that will be virtually mandatory this week.

Rank	Team	Delta

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

Dropped Out: Houston (#16), Utah (#20), West Virginia (#25).

Wow. Lots of movement. Topics of discussion follow, as well as all of the computer data.

Despite yesterday's trouncing, Alabama moves above TCU. I can go either way with this, honestly, but I don't see that there's a reason to change things. Looking at the scores below, you'll notice that the top four are way out in front. At least in my poll, if TCU wins out, they'll be one of those four.
Georgia Tech has the Strength of Schedule. They move up above Cincy and Boise becuase their SOS is more than compensating for their one loss. The numbers are all close, but about the only way to make the computers change things is to weight W/L to near-absurdity. But I wouldn't mind overriding here and keeping the undefeateds up. It's not like Cincy and Boise look bad.
CPUs love Big Ten. Because the Big Ten plays defense, man. That, or offenses in the conferences are so bad that their defensive metrics skew. And since offenses are underrated (an offseason fix), it does help Big Ten teams. Scan through and make arguments for adjustments.
Oklahoma? Oddly enough, their numbers are sky-high except for wins. The Sooners still obliterate lesser foes, which really helps their metrics, but lose to tougher teams. Should they slide down? If so, how far?
Stanford and Southern Cal Ok, I'm not about to go all AP Poll on you and leave that one alone. Stanford clearly deserves the higher ranking. Even without feedback, I will bump Stanford above USC. The rest, however, will depend on feedback. North Carolina is #26 and very close. Should I bring them back in and drop USC out? How far up should Stanford go? Deep questions.
I'm sure there are others. Those are the ones that caught my eye during a quick scan. Feel free to open up discussion on other teams.

The Data

UPDATE: Here are the numbers in Excel 2003 and Excel 2007 forms.

Untitled Document

Rank	Team	WL	SOS	PED	RD	3DO	TD	PEO	OPPG	TO	3DD	SCORE
1	Florida	1.968	0.618	2.172	1.224	0.690	2.251	1.613	2.064	0.891	2.114	153.446
2	Texas	1.968	0.231	1.707	2.253	1.159	2.279	0.817	1.762	0.896	2.525	150.033
3	Alabama	1.968	0.936	2.110	1.728	-0.424	2.157	0.425	2.007	0.694	1.713	144.787
4	TCU	1.968	0.259	1.612	1.390	0.643	2.060	1.843	1.718	1.581	1.890	141.955
5	Georgia Tech	1.584	1.271	-0.753	0.687	2.114	0.291	1.940	0.183	1.263	-0.380	106.578
6	Cincinnati	1.968	-0.074	0.728	0.317	0.488	0.478	2.381	1.141	1.766	-0.036	101.589
7	Boise St.	1.968	-0.697	1.380	0.790	-0.258	1.174	2.344	1.141	1.168	1.133	100.400
8	Ohio St.	1.201	0.236	1.607	1.471	0.353	1.827	0.010	1.796	-0.166	1.295	98.301
9	Virginia Tech	0.702	1.811	1.872	-0.023	-0.057	1.019	1.195	1.185	-0.010	1.037	95.858
10	Penn St.	1.201	-0.478	1.354	1.233	1.663	1.588	0.473	1.900	0.443	0.963	93.810
11	Pittsburgh	1.546	0.035	0.184	1.027	0.950	0.845	1.737	1.041	0.568	0.532	91.660
12	Oregon	1.124	1.617	1.394	0.363	-0.743	0.887	0.123	0.565	0.645	0.629	89.079
13	Iowa	1.201	0.661	1.673	0.417	0.142	1.198	-0.356	1.140	-0.589	0.188	82.904
14	BYU	1.124	-0.334	-0.362	1.008	3.024	0.449	2.215	0.464	1.237	-0.793	81.377
15	Oklahoma	0.280	0.471	1.572	1.465	0.364	1.755	0.230	1.834	1.013	1.141	77.587
16	LSU	1.124	0.747	1.157	0.460	-0.009	0.896	0.164	1.603	-1.215	0.396	76.913
17	Arizona	0.562	1.294	0.337	0.975	0.914	1.012	0.147	0.374	0.900	0.861	75.650
18	Miami (FL)	0.702	1.463	0.279	0.570	0.827	0.619	0.727	0.089	0.476	0.278	75.556
19	Oklahoma St.	1.124	0.033	0.690	1.334	0.522	0.511	0.849	0.608	0.393	0.790	73.877
20	Nebraska	0.702	0.060	1.943	1.210	-0.029	1.447	-0.117	1.992	-0.446	0.963	73.527
21	Oregon St.	0.702	1.289	-0.328	1.034	1.255	0.137	0.997	0.204	0.519	0.365	72.618
22	Wisconsin	1.124	0.060	-0.133	1.099	1.367	0.972	0.404	0.320	0.563	0.403	69.759
23	Clemson	0.702	0.851	1.360	0.340	-0.141	1.113	-0.088	0.997	-0.129	0.492	66.355
24	Southern California	0.702	1.773	0.942	0.256	-0.943	0.282	0.067	0.449	0.386	0.299	64.836
25	Stanford	0.702	1.391	-0.412	0.188	1.115	-0.350	1.225	0.089	1.191	-0.186	62.351
26	North Carolina	0.702	0.374	1.402	1.228	-0.457	1.658	-1.202	1.242	-1.431	1.426	60.044
27	Utah	1.124	-0.164	1.330	-0.105	-0.006	0.678	0.497	0.651	0.357	1.526	58.614
28	West Virginia	0.702	0.426	0.512	0.598	0.833	0.375	0.473	0.521	0.261	0.732	57.537
29	Mississippi	0.702	-0.001	1.364	0.324	0.116	0.875	-0.269	1.286	0.481	2.116	54.387
30	Navy	0.817	-0.146	0.114	0.295	1.505	0.434	0.741	0.537	-0.445	0.695	49.250
31	Air Force	0.434	-0.941	1.511	0.212	0.914	1.704	0.362	1.665	-0.310	1.835	45.268
32	California	0.702	1.126	-0.480	0.994	-0.582	-0.166	0.181	0.348	0.409	0.299	44.551
33	Boston College	0.702	0.523	0.565	0.865	-1.081	0.638	-0.275	0.911	-0.775	1.115	42.701
34	Central Mich.	1.124	-1.524	0.457	0.642	1.265	0.389	0.980	0.983	0.409	0.213	42.663
35	Auburn	0.434	0.602	0.838	-0.544	0.256	0.071	1.063	-0.315	1.117	0.924	40.506
36	Tennessee	-0.141	0.871	1.349	0.080	-0.302	0.844	0.385	0.521	0.167	0.751	36.823
37	South Carolina	0.050	1.246	0.519	0.033	-0.172	0.949	-0.299	0.589	-0.387	0.185	35.876
38	Georgia	0.280	1.269	-0.502	0.614	-0.087	0.508	0.155	-0.156	-0.611	0.707	35.390
39	South Fla.	0.562	-0.291	0.905	0.296	-0.449	0.755	0.898	0.823	-0.335	0.102	34.526
40	Notre Dame	0.280	0.993	-0.521	-0.157	0.121	-0.474	1.615	0.219	1.308	0.142	31.420
41	Missouri	0.280	0.446	-0.066	1.017	-0.551	0.382	0.504	0.233	0.229	-0.416	28.329
42	Texas Tech	0.280	-0.067	0.473	0.274	0.203	0.174	0.948	0.190	1.487	-0.329	28.240
43	Rutgers	1.030	-1.592	0.691	1.132	-0.661	1.020	-0.191	1.335	-0.580	1.249	26.891
44	UCLA	-0.141	1.599	0.643	-0.047	-0.961	0.389	-0.575	0.493	-0.580	0.921	24.481
45	Nevada	0.702	-0.573	-1.838	0.959	2.132	-0.862	0.521	-0.199	2.521	-0.041	20.808
46	Michigan St.	0.050	0.261	-0.413	0.952	-0.263	0.215	1.065	0.183	0.627	-0.645	18.512
47	Arkansas	0.280	0.826	-0.950	0.324	-0.813	-0.848	1.791	-0.099	1.271	0.275	17.457
48	Northwestern	0.434	-1.050	0.045	0.487	1.395	0.315	-0.046	0.314	0.051	0.668	15.477
49	Fresno St.	0.280	0.057	0.087	-1.226	1.224	-0.562	1.231	-0.084	1.193	0.102	15.348
50	Kentucky	0.280	0.175	1.162	-0.729	0.061	0.349	-1.306	0.449	-0.484	-0.190	12.899
51	Temple	1.124	-1.728	0.207	0.921	-0.442	0.729	-0.503	0.521	-0.902	0.416	12.786
52	Florida St.	-0.141	1.791	-1.317	-1.094	1.533	-1.279	1.079	-0.834	1.053	-0.423	12.772
53	Houston	1.124	-0.720	-0.805	-1.566	1.548	-1.720	1.504	-0.863	3.266	-0.816	10.746
54	Arizona St.	-0.563	0.979	0.665	0.961	-1.272	1.077	-1.197	0.550	-0.691	1.426	9.320
55	East Carolina	0.093	0.085	-0.079	0.554	0.557	0.055	-1.146	0.422	-0.637	0.013	7.173
56	UCF	0.280	-0.349	-0.661	1.524	-0.365	0.057	0.088	0.392	-0.972	-0.735	5.845
57	Kansas	-0.141	-0.282	0.139	0.450	0.376	0.207	0.247	-0.012	0.770	0.093	4.683
58	Southern Miss.	0.280	-1.368	-0.109	0.872	0.323	0.019	1.088	0.190	0.650	0.259	2.997
59	Wake Forest	-0.717	1.400	-0.202	-0.623	0.647	-0.231	0.649	-0.106	0.288	-0.605	2.868
60	Northern Ill.	0.702	-2.593	-0.111	0.874	0.858	0.844	0.456	1.113	-0.493	-0.048	0.383
61	Idaho	0.434	-0.288	-1.518	-0.197	1.601	-0.884	1.285	-1.324	1.166	-2.726	-0.932
62	Minnesota	0.050	0.747	0.008	-0.201	-0.409	-0.323	-0.545	-0.132	-1.313	-1.510	-1.408
63	Connecticut	-0.376	0.835	-0.803	0.293	-0.153	-0.049	-0.114	0.181	0.540	0.416	-2.035
64	La.-Monroe	0.280	-1.538	0.286	0.928	0.437	0.154	-0.032	-0.300	0.444	0.105	-2.644
65	Washington	-0.985	2.568	-1.409	-0.452	1.441	-1.013	-0.304	-0.892	-0.360	-0.868	-5.151
66	Troy	0.702	-0.912	-0.867	0.171	-0.146	-0.820	0.620	-0.618	1.407	0.057	-5.406
67	Ohio	0.702	-1.728	1.074	-0.030	-0.966	0.344	-0.416	0.565	-0.998	0.804	-6.431
68	Purdue	-0.717	0.720	0.625	-0.583	0.516	-0.096	-0.251	-0.721	0.329	-0.193	-7.513
69	Kansas St.	0.050	-0.453	-0.349	0.954	-0.752	0.291	-0.292	0.131	-0.426	-0.340	-8.428
70	Duke	-0.141	-0.191	0.073	-0.220	-0.307	0.356	0.294	-0.185	-0.312	-0.380	-9.393
71	Middle Tenn.	0.702	-1.730	0.810	-0.016	-1.048	0.007	-0.545	0.075	0.912	0.707	-10.498
72	Texas A&M	-0.141	0.091	-0.676	-0.429	1.347	-1.130	0.280	-1.324	1.312	0.105	-10.952
73	Marshall	-0.141	-0.108	0.310	-0.230	-0.097	-0.274	-0.888	0.464	-0.869	-0.336	-14.369
74	Mississippi St.	-0.563	1.233	0.130	-0.267	-1.530	0.038	-1.182	-0.055	-0.195	0.695	-15.817
75	North Carolina St.	-0.563	-0.399	-0.657	0.363	1.029	0.306	0.600	-0.877	0.616	-0.682	-16.188
76	Bowling Green	-0.141	-0.322	0.356	-1.214	0.415	-0.308	0.006	-0.315	0.265	0.877	-16.456
77	Iowa St.	0.050	-0.252	-0.332	-0.465	0.358	-0.691	-1.023	0.471	-0.300	-0.045	-18.791
78	SMU	0.280	-0.327	0.548	-0.408	-1.937	-0.688	0.081	-0.661	-0.003	-0.702	-19.980
79	Virginia	-0.985	1.438	1.172	-0.448	-1.451	0.315	-1.896	0.118	-1.983	0.591	-20.811
80	Baylor	-0.563	0.516	0.232	-0.480	-0.559	-0.587	-0.289	-0.300	-0.548	-0.785	-24.383
81	Tulsa	-0.376	-1.215	-0.302	0.593	-0.679	-0.345	1.132	0.214	0.481	0.373	-30.596
82	Louisville	-0.563	0.428	-0.912	-0.101	-0.551	0.208	-0.881	0.017	-0.693	0.080	-30.789
83	Wyoming	-0.141	-0.060	-0.296	-0.138	-0.941	-0.416	-1.383	-0.257	-1.085	-0.066	-33.598
84	Louisiana Tech	-0.985	-0.028	0.198	-0.670	0.488	-0.178	-0.319	-0.156	-0.315	0.262	-35.453
85	San Diego St.	-0.563	-0.446	-0.286	-0.490	0.680	-0.284	0.024	-0.747	-0.690	-1.413	-35.963
86	Michigan	-0.333	-0.666	-0.448	-0.427	0.106	-0.642	-0.059	-0.472	0.229	0.121	-37.050
87	Buffalo	-0.985	-0.982	0.099	0.239	0.554	0.113	-0.016	-0.199	0.629	-0.737	-39.697
88	Kent St.	-0.141	-1.818	0.397	0.549	-1.593	0.139	-0.600	0.507	-0.588	0.100	-44.103
89	Hawaii	-0.563	-0.752	-0.838	-1.120	1.230	-0.799	0.351	-0.819	1.517	-0.859	-45.144
90	Illinois	-0.985	0.353	-0.547	-0.314	0.017	-0.644	-0.965	-0.171	-0.154	-1.041	-45.912
91	Indiana	-0.717	0.028	-0.533	-0.369	-0.291	-0.770	-0.281	-0.564	-0.395	-1.741	-46.336
92	Colorado	-0.985	0.437	-0.385	-0.218	-0.658	-0.185	-1.514	-0.560	-1.186	0.760	-49.207
93	Syracuse	-0.985	0.541	-1.700	1.109	-1.766	0.167	-0.580	-0.257	-1.300	0.513	-49.455
94	Colorado St.	-0.985	0.697	-1.157	0.014	-1.124	-0.557	-0.238	-0.920	-0.526	-3.136	-54.106
95	UAB	-0.141	-0.749	-1.393	0.049	-0.842	-1.861	0.703	-0.992	0.495	-1.554	-54.325
96	Toledo	-0.563	-0.729	-0.226	-0.705	-0.210	-1.215	0.318	-2.088	0.946	-1.475	-56.429
97	Vanderbilt	-1.484	0.722	0.575	-1.196	-0.709	0.103	-2.362	0.314	-1.248	0.264	-59.388
98	La.-Lafayette	-0.141	-1.723	-0.538	-0.670	0.488	-0.702	-0.756	-0.733	-0.569	-1.067	-59.955
99	Western Mich.	-0.333	-1.382	-0.594	-0.796	0.282	-1.272	-0.199	-0.459	0.045	-0.260	-60.449
100	Maryland	-1.407	0.907	-1.155	-0.049	-0.475	-0.637	-0.984	-1.137	-1.195	-0.190	-62.240
101	Utah St.	-0.985	-0.363	-0.321	-1.273	-0.689	-1.264	0.280	-0.848	1.038	-0.823	-68.439
102	UNLV	-0.717	0.195	-1.513	-1.985	1.179	-1.634	-0.550	-1.206	-0.557	-2.176	-68.752
103	Army	-0.563	-1.929	0.140	0.120	-1.537	1.167	-1.684	0.219	-1.694	-0.084	-68.869
104	Fla. Atlantic	-0.844	-0.978	-1.209	-1.980	1.150	-1.362	0.404	-1.452	1.178	-0.214	-77.559
105	Arkansas St.	-1.313	-1.500	-0.477	0.681	-1.237	0.198	-0.592	0.454	-0.811	-0.816	-81.076
106	Miami (OH)	-1.867	0.453	-0.456	-0.834	-0.638	-0.199	-1.011	-1.245	-0.687	-0.437	-82.625
107	Tulane	-0.985	-0.620	-2.027	-1.284	0.852	-0.857	-0.280	-1.685	-0.762	-1.750	-88.098
108	UTEP	-0.985	-0.738	-1.062	-1.469	0.312	-1.758	-0.766	-1.483	0.666	-1.222	-90.797
109	North Texas	-1.407	-1.448	-0.315	-1.026	0.671	-0.947	-0.088	-1.901	0.717	0.066	-90.813
110	Akron	-1.407	-0.817	-0.051	-0.680	-1.552	0.026	-1.160	-0.574	-1.621	-1.233	-93.860
111	Memphis	-1.407	-0.718	-1.394	-0.855	-0.730	-1.309	-0.486	-1.180	-0.024	0.261	-103.762
112	New Mexico St.	-0.985	-1.120	0.111	-1.071	-1.346	-0.639	-2.834	-0.646	-2.563	-0.790	-104.036
113	FIU	-0.985	-0.991	-0.603	-2.156	-0.674	-2.221	-1.193	-1.238	-0.991	-0.237	-113.719
114	San Jose St.	-1.782	1.237	-0.591	-3.330	-1.575	-1.931	-0.887	-1.741	-1.765	-1.582	-117.676
115	Washington St.	-1.829	1.931	-1.609	-1.931	-2.778	-2.581	-1.300	-2.045	-2.052	-0.303	-121.803
116	Ball St.	-1.829	-1.475	-1.322	-0.201	-0.571	-0.056	-1.975	-0.545	-1.270	0.146	-122.945
117	New Mexico	-2.250	-0.058	-1.484	-0.272	-1.535	-1.022	-1.066	-1.512	-1.135	-1.687	-127.599
118	Rice	-1.829	-0.053	-2.164	-1.055	-1.300	-1.490	-1.407	-2.406	-1.302	-0.876	-136.055
119	Eastern Mich.	-2.250	-0.688	-1.047	-3.127	-0.664	-1.267	-1.763	-1.959	-1.834	-0.982	-163.724
120	Western Ky.	-2.250	-0.962	-2.585	-2.838	-1.522	-2.492	-1.299	-2.506	-0.936	-2.539	-198.389

Tweet Comment 14 comments | 0 recs |

Do you like this story?

Comments

Wow. The computer is hormonal this week. First thoughts:

Big problems:

*Stanford, as you mentioned, has beaten (at the time) top 10 teams two weeks in a row by a combined 45 points. They’re plenty worthy of 15-20. Is there any way to more heavily weight recent performance?

*Virginia Tech should NOT shoot up 9 spots after beating Maryland. How does that even happen? I know their SOS is still sky high, but it was sky high last week too. They should be hanging out with Miami back near where they were last week.

Smaller problems:

*Undefeateds should jump GT

*Iowa and Penn State should switch. Iowa won head to head and vastly outperformed PSU vs. their only common top 20 opponent.

*Arizona will stay. We’ll see this week whether they’re really not that good or whether it was a look ahead to Oregon.

*I’d be fine with UNC jumping USC. They’ve just beaten VT and Miami (both top 20 teams). That seems worthy of top 25 to me.

by Incipient_Senescence on Nov 15, 2009 1:25 PM EST reply actions

This year, recent performances can't be weighted.

It’s a possibility going forward, once I get a chance to revamp this thing, but I’m leery of that kind of adjustment. It’s one of my bigger gripes with human polls that they tend to be far more forgiving of early losses than late ones, all other things being equal.

But even without that, Stanford feels low. They’ll move to #24 on my own adjustment, but anything higher will take advisement.

As far as VTech: they gained about 0.2 in nearly every category (except SOS), which explains the 12-point jump in score. It’s odd to see that much jump that late, and I wonder how much is reflective of other teams losing ground. That, and like always, their weakest metrics (offense) are somewhat masked by the weights.

by David Hooper on Nov 15, 2009 3:14 PM EST up reply actions

Re: recent performance

I see your concern, and I believe I have the solution, if you’d like to consider it for next year. The polls weight late losses much higher than early losses. This is a problem if a team is playing well late and just loses a nailbiter that may be unlucky. However, the computers’ impartiality poses some problems. Remember when Nebraska (2001) and Oklahoma (2003) played in NC games despite getting destroyed in their last game? That is not good. My solution: don’t weight recent WL at all (this is the highest weighted metric, and it prevents us from doing what human polls do); do weight performance metrics (this allows us to rank higher teams that have actually improved over the course of the season without necessarily penalizing teams that lose a tight game late). What are your thoughts?

by Incipient_Senescence on Nov 15, 2009 3:39 PM EST up reply actions

Definitely worth considering.

I’ll first have to see how the reorg pans out, but that’s definitely an idea. I’d first need a good expression for what an improving team looks like, metrically, in a way that separates it from, say, an inconsistent team. But that might not be unreachable.

by David Hooper on Nov 15, 2009 3:45 PM EST up reply actions

Consisttent positive-trending mean performances for the season?

Does that make sense?

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 Chris Pendley on Nov 15, 2009 4:03 PM EST up reply actions

It does.

It’ll just be something I’ll have to be able to mathematically capture and define.

by David Hooper on Nov 15, 2009 4:16 PM EST up reply actions

I wonder if I'm being unclear:

By all that, I just mean that, if I tell the computer to search for a type of pattern, I’ll need to be able to say why that particular pattern, then state why the subsequent adjustment is appropriate, mathy-speaking.

by David Hooper on Nov 15, 2009 4:18 PM EST up reply actions

Do you have access to enough stats to weight the individual game metrics?

Game 1: 1.0
Game 2: 1.05
Game 3: 1.1
Game 4: 1.15
.
.
.
Game 12: 1.55

Or some such? If you have managed to incorporate SOS into the metrics themselves, then adjusting for SOS isn’t a problem.

by Incipient_Senescence on Nov 15, 2009 4:12 PM EST up reply actions

by Chris Pendley on Nov 15, 2009 6:20 PM EST up reply actions

I think it's probably everyone else bombing out, to be honest.

When data points are that tightly clumped, small movements can produce big changes, especially if they’re all upstream. I’d guess something similar happened with VT.

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 Chris Pendley on Nov 15, 2009 4:04 PM EST up reply actions