Semi-Objective CFP Ranking System: 2017 Week 10
The last couple of years I've posted this semi-objective ranking model and haven't this year yet since it's not as fun for a Michigan fan. However, the discussion about the CFP rankings made me curious about how they compare to the model.
This is is the week 10 college football team rankings produced by a semi-objective model as described previously here. The model was inspired by Seth's post that proposed a point system to determine bowl eligibility. These rankings aim to be a relatively objective starting point from which to apply considerations such as the eye test, margin of victory, and head-to-head results.
The Rules
- +3 points for a conference championship.
- +4 points for a win over a top 10 team.
- +3 points for a win over a top 25 team (not in top 10).
- +2 points for a win over a winning P5 team (not in the top 25).
- +1 point for a win over a winning G5 or a losing P5 team.
- +0 points for a win over a losing G5 or any FCS team.
- -1 point for a loss to a top 10 team.
- -2 points for a loss to a top 25 team (not in top 10).
- -3 points for a loss to a winning P5 team (not in top 25).
- -4 points for a loss to a losing P5 or any G5/FCS team.
Top 10 and top 25 status are determined by this ranking system and the model is solved iteratively until it converges. In scenarios where oscillatory states lead to a failure in convergence, the average points for oscillatory states are used.
Rank | Team | Points |
---|---|---|
1 | 'Notre Dame' | 15 |
2 | 'Georgia' | 14 |
3 | 'Clemson' | 13 |
4 | 'Alabama' | 11 |
5 | 'Wisconsin' | 11 |
6 | 'Michigan State' | 10 |
7 | 'Miami (FL)' | 10 |
8 | 'USC' | 10 |
9 | 'Penn State' | 10 |
10 | 'TCU' | 9 |
11 | 'Oklahoma' | 8.5 |
12 | 'UCF' | 7 |
13 | 'Washington' | 7 |
14 | 'Washington State' | 6 |
15 | 'Iowa' | 5.5 |
16 | 'Ohio State' | 5 |
17 | 'Virginia Tech' | 4.5 |
18 | 'Auburn' | 4 |
19 | 'Mississippi State' | 4 |
20 | 'Oklahoma State' | 3.5 |
21 | 'Michigan' | 3 |
22 | 'Northwestern' | 3 |
23 | 'Iowa State' | 1.5 |
24 | 'Memphis' | 1 |
25 | 'NC State' | 1 |
26 | 'Stanford' | 1 |
27 | 'West Virginia' | 0.5 |
28 | 'Boise State' | 0 |
29 | 'LSU' | 0 |
30 | 'South Carolina' | 0 |
31 | 'Boston College' | 0 |
32 | 'Arizona' | 0 |
33 | 'Toledo' | -2 |
34 | 'San Diego State' | -3 |
35 | 'South Florida' | -3 |
36 | 'Arizona State' | -3 |
37 | 'Wake Forest' | -3 |
38 | 'Troy' | -4 |
39 | 'FIU' | -4 |
40 | 'Texas A&M' | -4 |
41 | 'Texas' | -4.5 |
42 | 'SMU' | -5 |
43 | 'Kentucky' | -5 |
44 | 'Fresno State' | -5 |
45 | 'Georgia Tech' | -5 |
46 | 'Louisville' | -6 |
47 | 'Ohio' | -6 |
48 | 'Army' | -6 |
49 | 'Florida Atlantic' | -6 |
50 | 'Houston' | -6 |
51 | 'Virginia' | -6 |
52 | 'Utah' | -6 |
53 | 'California' | -6 |
54 | 'Florida State' | -6 |
55 | 'Arkansas State' | -7 |
56 | 'North Texas' | -7 |
57 | 'Navy' | -7 |
58 | 'Wyoming' | -7 |
59 | 'UCLA' | -7 |
60 | 'Kansas State' | -8 |
61 | 'Oregon' | -8 |
62 | 'Maryland' | -8 |
63 | 'Syracuse' | -8 |
64 | 'Texas Tech' | -8 |
65 | 'Nebraska' | -8 |
66 | 'Florida' | -9 |
67 | 'Georgia State' | -9 |
68 | 'Tennessee' | -9 |
69 | 'Arkansas' | -9 |
70 | 'Indiana' | -9 |
71 | 'Duke' | -9 |
72 | 'Colorado' | -10 |
73 | 'Northern Illinois' | -10 |
74 | 'Marshall' | -10 |
75 | 'Western Michigan' | -10 |
76 | 'Vanderbilt' | -10 |
77 | 'Ole Miss' | -10 |
78 | 'Colorado State' | -10 |
79 | 'Purdue' | -10 |
80 | 'Alabama-Birmingham' | -11 |
81 | 'UTSA' | -11 |
82 | 'Rutgers' | -11 |
83 | 'Minnesota' | -11 |
84 | 'Central Michigan' | -12 |
85 | 'Missouri' | -12 |
86 | 'Appalachian State' | -12 |
87 | 'Pittsburgh' | -13 |
88 | 'Southern Miss' | -14 |
89 | 'Akron' | -14.5 |
90 | 'Utah State' | -15 |
91 | 'Louisiana-Lafayette' | -15 |
92 | 'Lousiana Tech' | -16 |
93 | 'Western Kentucky' | -16 |
94 | 'Temple' | -16 |
95 | 'Louisiana-Monroe' | -16 |
96 | 'Air Force' | -17 |
97 | 'UNLV' | -17 |
98 | 'Middle Tennessee' | -18 |
99 | 'New Mexico State' | -19 |
100 | 'Miami (OH)' | -19 |
101 | 'Tulane' | -19 |
102 | 'North Carolina' | -19 |
103 | 'Cincinnati' | -20 |
104 | 'Connecticut' | -21 |
105 | 'South Alabama' | -21 |
106 | 'East Carolina' | -21.5 |
107 | 'Old Dominion' | -22 |
108 | 'Illinois' | -22 |
109 | 'Oregon State' | -22 |
110 | 'Buffalo' | -23 |
111 | 'New Mexico' | -24 |
112 | 'Idaho' | -24 |
113 | 'Hawaii' | -24 |
114 | 'Baylor' | -24.5 |
115 | 'Massachusetts' | -25 |
116 | 'BYU' | -25 |
117 | 'Kansas' | -25 |
118 | 'Eastern Michigan' | -26 |
119 | 'Tulsa' | -26 |
120 | 'Texas State' | -27 |
121 | 'Bowling Green' | -27 |
122 | 'Kent State' | -28 |
123 | 'Nevada' | -28 |
124 | 'Charlotte' | -30 |
125 | 'Ball State' | -31 |
126 | 'Rice' | -31 |
127 | 'Coastal Carolina' | -32 |
128 | 'UTEP' | -33 |
129 | 'Georgia Southern' | -34 |
130 | 'San Jose State' | -35 |
November 8th, 2017 at 3:28 PM ^
OK....so Michigan is ranked #20 instead of like #26-30. Feel better?
November 8th, 2017 at 3:33 PM ^
I'm not trying to feel better. I don't have a problem with Michigan not being ranked by the committee.
November 8th, 2017 at 3:48 PM ^
November 8th, 2017 at 3:55 PM ^
They'll be in it next week after beating Miss. State. If you look at who Alabama has played so far this year, you can see why they're outside the top 4 here though. They have played one game against a team with fewer than 4 losses. Freso State.
November 8th, 2017 at 4:01 PM ^
Also the conference championship boost. You could imagine that the conference championship could have something like +10.
November 8th, 2017 at 4:05 PM ^
The winner of the conference championship game is also usually winning a +3 or +4 type game. So it ends up being a substantial boost for conference winners (+6 or +7). +10 makes it nearly impossible for a non-conference winner to qualify even if P5 conference winners are 3 or 4 loss teams.
November 8th, 2017 at 3:52 PM ^
Nevermind, I didn't read it all.
November 8th, 2017 at 4:05 PM ^
I am not too sure if I understand the iterative model correctly. At the first pass, how do you assign points for quality of wins/losses? Do you start with an initial ranking? If so, how do you pick the initial ranking? If so, how robust are the results with some change up in the initial ranking?
November 8th, 2017 at 4:14 PM ^
No initial ranking for the first iteration. Top 25 and top 10 wins/losses only come into play in iterations 2 and beyond.
November 8th, 2017 at 4:48 PM ^
Funny that Georgia and the team they beat are 1 and 2, and that the former Georgia coach sits at 3 and has an opportunity this weekend to beat that #2 team that Georgia beat.
November 8th, 2017 at 5:04 PM ^
I like how it is iterative. Only suggestion would be some incorporation for site-of-game (road wins valued more than home wins).
November 9th, 2017 at 10:22 AM ^
Thanks for the suggestion. I actually had a previous version that accounted for home vs. away wins and losses but it produced some counter-intuitive results and I axed it because tweaking it to look right felt too arbitrary. It might have just been an uncommon set of circumstances, however and I may revisit it in the future.
November 8th, 2017 at 5:16 PM ^
How does your model account for beating or losing to a team that has a record of .500? Neither winning nor losing record. Just curious.
November 9th, 2017 at 9:33 AM ^
.500 counts as winning in this case.
November 8th, 2017 at 5:35 PM ^
How do you get to 12 points with -4 points from the loss to losing-record Syracuse?
- Kent +0
- AUB +3
- Lou +3
- BC +1
- VT +3
- Wake +1
- 'Cuse -4
- GT +0
- NCS +3
That totals 10 points - am I missing something?
November 8th, 2017 at 7:57 PM ^
What stands out to me is that for all the ragging on Wisconsin's strength of schedule, Alabama really hasn't played anyone better. LSU and Texas A and M are the only winning P5 wins, and they don't grade into the top 25, which Northwestern does. Granted, the rest of 'Bama's season is more impressive, if they survive it.
November 8th, 2017 at 8:56 PM ^
November 9th, 2017 at 9:40 AM ^
I like this
November 9th, 2017 at 10:51 AM ^
Better buckle up, boys.
November 10th, 2017 at 9:16 AM ^
November 10th, 2017 at 10:30 AM ^
Shouldn't Michigan have 3 points?
We have wins against 5 losing P5 teams, UF, Purdue, IU, Rutgers, and Maryland so +5,
We have wins against 2 losing P5 teams, AFA and Cincy, so +0
Then we have 2 losses against top 10 teams, MSU and PSU, so -2
+5 and -2 equals 3. Am I missing something?
November 10th, 2017 at 1:04 PM ^
Good catch. There was a weird bug where Texas Tech was getting Air Force's nonP5 tag and vice versa so our Air Force win erroneously counted as a losing P5 win (+1) instead of losing G5 win (+0). This fix caused some shakeup to the rankings.
November 10th, 2017 at 1:33 PM ^
That makes sense.
I like this type of analysis, but I'm not a fan of the linear scale that is based solely on wins and losses. So Iowa's domination over OSU counts the same as Northwestern's OT win against Iowa.
It seems like if the goal is to award a point value to the winner and loser of a game you could somehow include a point differential to reward big wins more generously and not penalize close losses.
Either way I enjoy reading this so thank you for doing it.
November 10th, 2017 at 6:39 PM ^
I wish I had time, but I just feel there isn't enough of a jump in the different tiers of rewards here. Its a good start though.
November 13th, 2017 at 1:33 PM ^
What would you suggest? It's trivial for me to adjust those numbers and run it again.
The problem I've found with a a larger disparity in the numbers is that the top teams are simply the ones with the hardest schedules.
Comments