Santos computer model

[jay santos]

Here is my computer ranking right now. As you can see, the results are screwy. My model includes no preseason or prior season weighting. Please ignore all computer rankings until teams have played 5-6 games. Any computer model that doesn't look a lot like this has preseason or prior season weighting.



RANK TEAM W L SOS_RANK
1 Southern Cal 2 0 33
2 Wisconsin 3 0 17
3 Utah 4 0 21
4 Alabama 4 0 48
5 Wake Forest 3 0 14
6 Florida 3 0 79
7 Nebraska 3 0 73
8 Vanderbilt 4 0 51
9 Ball State 4 0 78
10 Georgia 4 0 80
12 Boise State 3 0 68
12 Texas 3 0 103
13 TCU 4 0 105
14 Oklahoma 3 0 111
15 Penn State 4 0 117
16 Kentucky 3 0 96
17 Tulsa 3 0 113
18 Missouri 4 0 110
19 Minnesota 4 0 98
20 LSU 3 0 108
21 South Florida 4 0 100
22 Connecticut 4 0 107
23 Oklahoma St 3 0 120
24 Texas Tech 4 0 121
25 Brigham Young 4 0 119
26 Northwestern 4 0 118
27 East Carolina 3 1 35
29 Oregon 3 1 59
29 Colorado 3 0 112
30 UNLV 3 1 16
31 California 2 1 34
32 Georgia Tech 3 1 74
33 Ohio State 3 1 60
34 Michigan State 3 1 82
35 Virginia Tech 3 1 52
36 Auburn 3 1 67
37 Air Force 3 1 70
38 Kansas 3 1 92
39 Miami FL 2 1 46
40 Marshall 3 1 47

[MikeWaters]

heaven forbid we use the past to account for the future.

[jay santos]

Originally Posted by MikeWaters

heaven forbid we use the past to account for the future.

I'm not saying it's innacurate to do so. I'm saying the discussion you guys are having about Auburn losing, BYU's relative SOS, whatever, is all missing the big picture--which is that the computer models have preseason weight right now and the discussion of the effect of actual game data is WAY premature.

[il Padrino Ute]

I'm not a statistician or computer ranking kind of guy, so could you explain how it's done - in layman's terms?

How did Utah get a number 3 ranking? It makes no sense to me.

[Travis Henry]

Originally Posted by jay santos

Here is my computer ranking right now. As you can see, the results are screwy. My model includes no preseason or prior season weighting. Please ignore all computer rankings until teams have played 5-6 games. Any computer model that doesn't look a lot like this has preseason or prior season weighting.



RANK TEAM W L SOS_RANK
1 Southern Cal 2 0 33
2 Wisconsin 3 0 17
3 Utah 4 0 21
4 Alabama 4 0 48
5 Wake Forest 3 0 14
6 Florida 3 0 79
7 Nebraska 3 0 73
8 Vanderbilt 4 0 51
9 Ball State 4 0 78
10 Georgia 4 0 80
12 Boise State 3 0 68
12 Texas 3 0 103
13 TCU 4 0 105
14 Oklahoma 3 0 111
15 Penn State 4 0 117
16 Kentucky 3 0 96
17 Tulsa 3 0 113
18 Missouri 4 0 110
19 Minnesota 4 0 98
20 LSU 3 0 108
21 South Florida 4 0 100
22 Connecticut 4 0 107
23 Oklahoma St 3 0 120
24 Texas Tech 4 0 121
25 Brigham Young 4 0 119
26 Northwestern 4 0 118
27 East Carolina 3 1 35
29 Oregon 3 1 59
29 Colorado 3 0 112
30 UNLV 3 1 16
31 California 2 1 34
32 Georgia Tech 3 1 74
33 Ohio State 3 1 60
34 Michigan State 3 1 82
35 Virginia Tech 3 1 52
36 Auburn 3 1 67
37 Air Force 3 1 70
38 Kansas 3 1 92
39 Miami FL 2 1 46
40 Marshall 3 1 47

Having not looked at the methodology, it looks like the model weighs heavily towards road wins. Utah has already had three road wins. I think road wins should play a part in all of the computer rankings. I also think margin of victory should be taken into consideration

[ERCougar]

Originally Posted by il Padrino Ute

I'm not a statistician or computer ranking kind of guy, so could you explain how it's done - in layman's terms?

How did Utah get a number 3 ranking? It makes no sense to me.

I'm curious too. I'm really fascinated by the computer rankings and I like seeing people's formulas. Also, how do you input the scores/outcomes each week?

[ERCougar]

Originally Posted by Travis Henry

Having not looked at the methodology, it looks like the model weighs heavily towards road wins. Utah has already had three road wins. I think road wins should play a part in all of the computer rankings. I also think margin of victory should be taken into consideration

I agree. MOV should count some, but in order to discourage RUTS, all 20+ pt victories should count the same.