College Football Playoff Predictor - Week 12

As always explanation here, data below, thoughts after.

Rank	Team	Agg Rank	CFP Prob
1	Alabama	1	92.44%
2	Ohio State	2	68.55%
3	Clemson	3	48.13%
4	Oregon	12	36.34%
5	Northwestern	14	32.99%
6	Wisconsin	4	27.27%
7	Notre Dame	7	24.75%
8	USC	21	16.00%
9	Miami (FL)	22	14.37%
10	Texas A&M	13	8.85%
11	Washington	24	6.74%
12	Cincinnati	10	4.15%
13	BYU	11	3.37%
14	Indiana	15	2.81%
15	Florida	6	2.72%
16	Marshall	32	2.00%
17	Oklahoma	5	1.43%
18	Georgia	8	1.40%
19	Colorado	60	0.81%
20	Oklahoma State	25	0.76%
21	Iowa State	17	0.63%
22	North Carolina	19	0.61%
23	Coastal Carolina	42	0.57%
24	Maryland	71	0.40%
25	Iowa	9	0.33%
26	Western Michigan	66	0.31%
27	Texas	20	0.31%
28	Purdue	30	0.30%
29	San Jose State	84	0.28%
30	Buffalo	64	0.17%
31	Nevada	79	0.09%
32	Auburn	16	0.09%
33	Kent State	91	0.02%

Diehard Blogacz fans may have noticed I haven't updated the Playoff odds for a while.  The main reason for this was that the Pac-12 was making everything whack.  Last week's numbers had an 0-1 Arizona State all the way up in the top ten.  Normally this sort of oddity would take care of itself soon enough but not this year.  Because my model only looks at 1) number of losses and 2) strength of schedule, it is not able to account for a team that simply doesn't play a lot of games.  A Pac-12 runner-up could finish something like 3-1 with a "decent" strength of schedule and end up in the top four when it's pretty clear that wouldn't reflect reality.

Luckily I was able to come up with a quick and dirty workaround: I simply disregard any Pac-12 team with a loss.  So for the four remaining teams, their playoff odds simply reflect their odds of going undefeated.  And while I don't think a 7-0 USC would necessarily be a shoe-in, this is at least closer to reality that the previous numbers this thing was spitting out.  Good enough for 2020.

ND fans may wonder why the Irish are so low (well behind a Clemson team they beat), but it's pretty straightforward when I peak under the hood.  The model has ND with roughly a 20% chance of winning out, in which case they would definitely be in.  But if they do lose, their SOS is quite bad and my model punishes them for that.  On the other side of the ACC, the Tigers have about a 50% chance of finishing with one loss.  Because their SOS is much better, my model would consider Clemson a near-lock at that point, hence the Playoff odds in the same neighborhood.

Conference	Favorite	Perc		Runner-Up	Perc
ACC	Notre Dame	93.4%		Clemson	89.2%
AMER	Cincinnati	99.9%		Tulsa	81.1%
B10E	Ohio State	99.6%		Maryland	0.3%
B10W	Northwestern	93.0%		Wisconsin	5.8%
B12	Iowa State	74.0%		Oklahoma	67.0%
CUSAE	Marshall	96.7%		Charlotte	2.0%
CUSAW	UAB	88.0%		Louisiana Tech	6.2%
MACE	Buffalo	66.7%		Kent State	16.9%
MACW	Western Michigan	93.3%		Toledo	2.7%
MWC	Boise State	83.6%		Nevada	53.7%
P12N	Oregon	74.3%		Washington	24.8%
P12S	USC	77.5%		Colorado	13.1%
SECE	Florida	95.6%		Georgia	4.4%
SECW	Alabama	99.7%		Texas A&M	0.2%
SUNE	Coastal Carolina	98.5%		Appalachian State	1.2%
SUNW	Louisiana-Lafayette	100.0%

The conference title grid reflects the impact of Week 12, which clarified virtually every conference/division race.  There are still some pivotal games in the next three weeks (Oregon-Washington, USC-Colorado, Buffalo-Kent State, Iowa State-Texas), but almost every conference features one or two teams in the catbird seat.

Week 13 Preview

Home	Away	Home Win Prob	Playoff Teams Lost
North Carolina	Notre Dame	35.5%	0.033
Alabama	Auburn	91.3%	0.028
Clemson	Pitt	92.6%	0.018
Michigan State	Northwestern	15.8%	0.017
Oregon State	Oregon	13.2%	0.016
USC	Colorado	81.0%	0.012
Texas A&M	LSU	74.1%	0.011
Illinois	Ohio State	5.0%	0.011
Wisconsin	Minnesota	91.9%	0.011
Washington State	Washington	34.9%	0.008

Lol the last game on this list was cancelled as I was typing this.  Cool.