Preseason Rankings

So, taking a statistical approach to comparing teams, I made this predictive model over the course of the past few weeks.  Basically, I take a bunch of stats for each team, provide a weighting of how important that stat is to a team's success, and the output comes out.  The stats I looked at were: average weight of OL and DL, wins from 2009-2011 (if team was in FCS during this period, I halved their win total for the years in FCS), wins in 2012 (also halved win totals if in FCS), # of QB starts in the program, # of OL starts in the program, the Head Coach's career winning % (in the case where a school has a first-year coach with no prior head coaching experience, I took the team's combined winning % over the past 7 seasons), # of players from last year's roster returning, # of defensive starters returning, # of offensive starters returning, conference affiliation (I ranked the conferences 1-11), where a team finished in its conference (where divisions existed I combined the divisions and used conference record to determine ordering, used overall record to break a tie, then head-to-head to break if still a tie, and, if still a tie, gave both teams the same "finish" in the conference), how many freshmen are on their 2-deep (including redshirt freshmen), recruiting class rank, a formula accounting for freshmen expected to contribute and the class ranking, and how many seniors are on their 2-deep.  I gathered that data for all 126 teams in FBS (including Georgia State, Old Dominion, Massachusetts, and UTSA which joined FBS either this season or last season) and here it is in a nice package for you!

It still needs some tweaking in terms of using it to make predictions about individual game outcomes and total win/loss record, but I am somewhat satisfied with the results.  The main point was to come up with a way to quantify my gut feelings about the upcoming season.  I feel like this does a pretty good job.  Some of the teams that I thought might surprise this season ended up being ranked higher than other preseason rankings that I've seen (Ole Miss, Miami), though admittedly this puts them a little higher than I think they probably belong.  Others that I think are over-rated are ranked lower than other preseason rankings (Louisville, Texas A&M, USC).

This initial posting is simply how highly the teams rate based on the weighted average of the 16 statistical categories.  I placed the teams' strength of schedule next to them so you can envision what their season might play out like.  For example, according to my model, Stanford is the 5th best team in the country, however, they play the 3rd toughest schedule, so it is less likely that they'll finish ranked 5th at the end of the season.  Nebraska and Florida State, on the other hand, are the 8th and 10th best teams, respectively, playing the 85th- and 75th-toughest schedule, respectively, so the model would predict that they would probably finish higher than 8th and 10th.  Notre Dame, Florida, Arizona State are in a similar situation with Stanford: they are probably better teams than their final records will indicate.

Again, I am still playing around with some aspects of the model as I am not 100% satisfied with the predictive results.  When I combine quality of team with ease of schedule, some interesting things happen: mostly, it thinks there are 6 potential non-AQ BCS busters.  Basically any non-BCS team in the top 60 with an SOS lower than 60 jumps into the top 25, with two teams in the top 10.  Some non-BCS teams jump into the top 25 even if their predicted records are 9-3.  When was the last time a MAC team was even getting votes at 9-3, let alone getting ranked...

The model really just says, on paper, based on criteria that I selected to use for comparison, how good is each team.  Last I checked though, Ole Miss didn't have a cardboard field.  I could buy Texas as the 4th most talented team in the country, but running through a round-robin gauntlet Big XII schedule with two top 40 opponents in the non-conference (including one on the road), it seems unlikely they will stay up that high.

The model does predict that Alabama, clearly the best team in the country coming into the season by any measure, will defeat all of their opponents by double digits, excluding a 9-point win against Ole Miss.  The game against Georgia in the SEC Title game (the model projects those two winning their divisions), is a statistical draw, suggesting a one-point Alabama win.

I am very interested in your thoughts on this, so call, e-mail, or post a comment!

Rank	Team	SOS
1	Alabama	38
2	Ohio State	84
3	Georgia	11
4	Texas	23
5	Stanford	3
6	Oregon	39
7	Ole Miss	27
8	Nebraska	85
9	Notre Dame	7
10	Florida State	75
11	LSU	10
12	Miami	48
13	Oklahoma	19
14	TCU	40
15	OK State	42
16	Oregon State	33
17	Clemson	61
18	Florida	1
19	Wisconsin	66
20	Michigan State	65
21	Washington	37
22	Louisville	93
23	USC	15
24	Boise State	106
25	ECU	104
26	NIU	125
27	Arizona State	6
28	Virginia Tech	44
29	Texas A&M	30
30	South Carolina	22
31	Kansas State	51
32	Utah State	92
33	Bowling Green	119
34	Georgia Tech	46
35	Tennessee	14
36	Ohio	123
37	Fresno State	117
38	Auburn	20
39	Rice	122
40	BYU	16
41	Arizona	45
42	UCLA	21
43	Mississippi State	5
44	Michigan	32
45	Cincinnati	69
46	Missouri	18
47	NC State	82
48	Utah	8
49	Toledo	100
50	Rutgers	77
51	Vanderbilt	35
52	North Carolina	25
53	Lafayette	114
54	Northwestern	49
55	ULM	99
56	Minnesota	52
57	Baylor	41
58	Arkansas	4
59	MTSU	118
60	Texas Tech	43
61	Marshall	110
62	Penn State	59
63	Houston	63
64	San Diego State	78
65	Tulsa	74
66	Temple	83
67	Boston College	53
68	Pittsburgh	34
69	Purdue	17
70	San Jose State	73
71	Wake Forest	50
72	USF	64
73	WKU	103
74	Washington State	9
75	Navy	107
76	Arkansas State	111
77	Ball State	124
78	Kentucky	13
79	West Virginia	55
80	Connecticut	87
81	Duke	58
82	UCF	70
83	Indiana	36
84	Virginia	29
85	Texas State	120
86	Iowa	26
87	North Texas	90
88	UTSA	88
89	Colorado State	98
90	Hawaii	54
91	Buffalo	102
92	Syracuse	57
93	SMU	56
94	Troy	86
95	UNLV	94
96	Nevada	62
97	South Alabama	108
98	California	2
99	Iowa St	28
100	Illinois	24
101	Kent State	81
102	ODU	126
103	CMU	101
104	Memphis	76
105	Colorado	12
106	Southern Mississippi	67
107	Army	109
108	Georgia State	115
109	Wyoming	89
110	UAB	79
111	Louisiana Tech	121
112	Maryland	60
113	Kansas	31
114	Air Force	95
115	New Mexico State	97
116	Tulane	113
117	New Mexico	95
118	WMU	105
119	FAU	71
120	Miami OH	91
121	UTEP	112
122	FIU	116
123	Akron	72
124	Idaho	47
125	EMU	68
126	U Mass	80