College Football Playoff Predictor - Year 7

With a "normal" season of college football around the corner, it is time for me to offer my reasonably good prognostications for the Playoff and more.  With respect to the CFP, my current setup is only relevant for the current four-team iteration, so please enjoy it while it lasts.  Once the sport progresses to something else, the simple utility of my model will no longer hold, and I will likely discontinue this exercise.  As a wise man once said, this isn't meant to last - this is for right now.

Regarding my model, when I say simple I mean it.  Whereas other models make use of a number of factors like previous committee rankings, conference title odds, and fancy SOS calculations, mine explicitly does not.  All I look at is a) number of losses, b) major/minor conference distinction (an undefeated minor conference team is treated as a two-loss major conference team), and c) the basic SOS calculation used by the NCAA.  I do this in part because it's easier, but also because I want to drive home that the committee's logic really is that simple.  This simple model has worked well enough to predict* all but one CFP participant in the first seven years of the Playoff Era.  I will gladly take that extremely mild failure in exchange for being able to make my larger point (that the committee is dumb).

*Saying that the end of season is top four is the four I am "predicting" to make the Playoff is imprecise (ie. I re-do the logit equations and use Brier scores to double check my model each offseason), but it's not a bad gut check and it's much easier to explain briefly than all the other stuff.

To work backwards from the end of the season, I run 10000 simulations each week, using an amalgam of prediction systems (the "Agg Rank" column below) to set the odds for each game.  This gives me an idea of how many times each contender will finish with 0, 1, or 2 losses, which I in turn apply my model to proportionally.  The end result is the percentage odds below, which I round to a total of 400% using an algorithm that adjusts the logit models for the season in question (there's also an adjustment in there that accounts for the wackiness of the season that I introduced here.  Put briefly, if we're having a chalky season, this will serve to disproportionally decrease the overall odds of a 1-loss or 2-loss from making the Playoff).

Rank	Team	Agg Rank	CFP Prob
1	Alabama	1	64.79%
2	Clemson	2	59.38%
3	Oklahoma	3	58.28%
4	Ohio State	4	50.47%
5	Georgia	5	26.35%
6	Texas A&M	6	17.40%
7	Notre Dame	14	13.59%
8	Wisconsin	10	12.05%
9	Florida	8	9.55%
10	Oregon	12	9.30%
11	Iowa State	7	9.19%
12	Penn State	13	8.59%
13	North Carolina	15	8.35%
14	USC	16	8.31%
15	Miami (FL)	9	8.20%
16	Washington	18	6.84%
17	Utah	22	4.65%
18	Texas	11	3.65%
19	LSU	17	3.00%
20	Arizona State	24	2.98%
21	Iowa	20	2.28%
22	Ole Miss	23	2.01%
23	Indiana	27	1.60%
24	Oklahoma State	21	1.34%
25	Michigan	26	1.27%
26	Auburn	19	1.17%
27	Virginia Tech	29	1.05%
28	North Carolina State	35	0.52%
29	UCLA	31	0.42%
30	TCU	28	0.40%
31	Pitt	36	0.36%
32	Boston College	52	0.31%
33	Minnesota	30	0.30%
34	Mississippi State	32	0.26%
35	Northwestern	43	0.25%
36	Louisville	55	0.18%
37	Missouri	46	0.15%
38	Kentucky	40	0.13%
39	Wake Forest	58	0.11%
40	Virginia	47	0.11%
41	Cincinnati	25	0.10%
42	Florida State	44	0.08%
43	Nebraska	37	0.06%
44	Arkansas	38	0.05%
45	Appalachian State	34	0.05%
46	Michigan State	56	0.05%
47	Tennessee	51	0.05%
48	Washington State	59	0.05%
49	Baylor	41	0.05%
50	West Virginia	39	0.04%
51	California	62	0.04%
52	UCF	33	0.03%
53	Boise State	49	0.03%
54	Louisiana	42	0.03%
55	Maryland	60	0.03%
56	Texas Tech	50	0.02%
57	Coastal Carolina	48	0.02%
58	Stanford	54	0.01%
59	BYU	45	0.01%
60	Memphis	53	0.01%
61	Toledo	63	0.01%
62	Purdue	61	0.01%
63	Western Michigan	72	0.00%
64	Wyoming	81	0.00%
65	UAB	77	0.00%
66	Georgia Tech	68	0.00%
67	SMU	57	0.00%
68	Liberty	75	0.00%
69	Kansas State	64	0.00%
70	San Diego State	73	0.00%
71	South Carolina	70	0.00%
72	Colorado	66	0.00%
73	Buffalo	76	0.00%
74	UTSA	90	0.00%
75	Rutgers	82	0.00%
76	Troy	78	0.00%
77	Houston	67	0.00%
78	Marshall	87	0.00%
79	Oregon State	74	0.00%
80	Florida Atlantic	86	0.00%
81	Central Michigan	79	0.00%
82	Illinois	84	0.00%
83	Ball State	71	0.00%
84	Air Force	94	0.00%
85	Nevada	80	0.00%
86	Eastern Michigan	91	0.00%
87	Army	97	0.00%
88	North Texas	107	0.00%
89	Hawaii	102	0.00%
90	Duke	95	0.00%
91	Western Kentucky	106	0.00%
92	Ohio	92	0.00%
93	Arizona	96	0.00%
94	Syracuse	103	0.00%
95	Navy	114	0.00%
96	Southern Miss	109	0.00%
97	Colorado State	101	0.00%
98	Miami (OH)	99	0.00%
99	Louisiana Tech	112	0.00%
100	South Florida	104	0.00%
101	Vanderbilt	108	0.00%
102	Kansas	115	0.00%
103	Tulsa	69	0.00%
104	Kent State	98	0.00%
105	Tulane	65	0.00%
106	Utah State	113	0.00%
107	Georgia Southern	89	0.00%
108	East Carolina	100	0.00%
109	UNLV	122	0.00%
110	Georgia State	85	0.00%
111	Arkansas State	93	0.00%
112	San Jose State	88	0.00%
113	South Alabama	121	0.00%
114	Temple	111	0.00%
115	Rice	117	0.00%
116	ULM	123	0.00%
117	Bowling Green	126	0.00%
118	Middle Tennessee	110	0.00%
119	New Mexico	120	0.00%
120	Texas State	118	0.00%
121	Old Dominion	127	0.00%
122	FIU	116	0.00%
123	Fresno State	83	0.00%
124	UTEP	124	0.00%
125	Akron	125	0.00%
126	Charlotte	119	0.00%
127	Northern Illinois	105	0.00%
128	UConn	128	0.00%
129	UMass	129	0.00%
130	New Mexico State	130	0.00%

As you might expect, the exact four teams you would expect to be in the top four are in the top four.  What's more is that all four teams have greater than 50% chance of making the field.  But I still think there's reason for optimism if you're a fan of chaos (or at least relative chaos).  The distance between the top four and the next bunch of teams is a little smaller than you might expect, and all of them have to face at least one of the teams right behind them in the standings.  Outside of the top four, Notre Dame outpunches its weight due to a) a manageable schedule and b) no conference title game, while the other Big 12 contenders take a hit since they will mostly likely have to play Oklahoma twice and beat them at least once to make the Playoff.  If you're looking for a fun long shot, Arizona State at 3% is juicy.

Weekly Preview

My favorite part of this post is previewing each week to get you excited for the ins and outs of the season.  Per my "Playoff Teams Lost" metric, there are 73 games with at least a 1% (ie. 0.01 or greater in the metric) chance of flipping a Playoff spot.  Those are ordered by week below.

Week 1 - 0.316 playoff teams lost (ranked 1st of 13)

1. Clemson (63% win probability) vs. Georgia - 0.129 playoff teams lost
2. Alabama (83%) vs. Miami FL - 0.060
3. Penn State at Wisconsin (59%) - 0.033
4. Ohio State (84%) at Minnesota - 0.028
5. Oklahoma (93%) at Tulane - 0.014
6. Notre Dame (71%) at Florida State - 0.013
7. North Carolina (60%) at Virginia Tech - 0.013

The opening weekend of the season is not particularly deep (Iowa-Indiana is the only must-watch not on this list), but boy howdy look at that list of games.  Nine top-twenty teams playing away from home!  I'm not sure if this means that the Playoff race will actually get shaken up right away, but it's as good of a chance as you'll get in the age of timid scheduling.

Week 2 - 0.139 playoff teams lost (ranked 13th)

1. Oregon at Ohio State (80%) - 0.059
2. Iowa at Iowa State (73%) - 0.014
3. Washington at Michigan (50.4%) - 0.014

We go directly from the best weekend of the year to the worst.  The good news is that all three of these games are at different times, so at least you can microdose quality football throughout the day.  Miami-App State, Texas-Arkansas, and the Holy War fill out a subpar but still palatable weekend.

Week 3 - 0.166 playoff teams lost (ranked 9th)

1. Alabama (76%) at Florida - 0.076
2. Nebraska at Oklahoma (93%) - 0.013
3. Auburn at Penn State (64%) - 0.013

The next week is a little better by my metric in part because Alabama gets one of its biggest challenges, and in part because we have a few more major-conference matchups (Miami-Michigan State is a game I guess?).  Still, this is probably the most skippable weekend of the year.  Enjoy your lovely mid-September weekend.

Week 4 - 0.140 playoff teams lost (12th)

1. Wisconsin (53%) vs. Notre Dame - 0.043
2. Clemson (89%) at NC State - 0.024
3. West Virginia at Oklahoma (94%) - 0.013
4. Texas A&M (82%) vs. Arkansas - 0.011

Another lowly rated week before conference play ramps up, but at least this one gives us a fun matchup in Chicago and also Clemson's hardest (?) road trip.  Further down the ticket, USC-Oregon State has some fun upset potential.

Week 5 - 0.178 playoff teams lost (7th)

1. Ole Miss at Alabama (92%) - 0.024
2. Cincinnati at Notre Dame (67%) - 0.015
3. Michigan at Wisconsin (71%) - 0.015
4. Oklahoma (93%) at Kansas State - 0.014
5. Indiana at Penn State (71%) - 0.012
6. Boston College at Clemson (95%) - 0.011
7. Mississippi State at Texas A&M (84%) - 0.010

Now here is a week of College Football.  There's no headlining games to be found, but there is a lot of potential goofiness.  Oklahoma and Penn State lost these games in upsets last year and Clemson struggled with BC as well, so perhaps the encores will be just as entertaining?  Add in Georgia-Arkansas, Ohio State going to New Jersey, Washington going to Corvallis, and a TCU-Texas grudge match, and you have yourself one of the deepest weekends of the year.

Week 6 - 0.275 playoff teams lost (2nd)

1. Alabama (72%) at Texas A&M - 0.102
2. Oklahoma (78%) vs. Texas - 0.053
3. Georgia (69%) at Auburn - 0.030
4. Notre Dame (61%) at Virginia Tech - 0.020
5. Utah at USC (64%) - 0.020
6. Penn State (51%) at Iowa - 0.018

Look at those three SEC games at the top of the second-biggest weekend of the year.  Wow what a conference.

Week 7 - 0.165 playoff teams lost (11th)

1. Miami FL at North Carolina (53%) - 0.028
2. TCU at Oklahoma (90%) - 0.021
3. Alabama (91%) at Mississippi State - 0.021
4. Florida (54%) at LSU - 0.020
5. Arizona State at Utah (58%) - 0.012
6. Texas A&M (81%) at Missouri - 0.012

After two momentous weeks, it's nice to catch our breath.  The battle for the ACC Coastal, Cincinnati-UCF, and a potentially important game in the Pac-12 South not involving USC?  That's a nice low-stakes weekend before we get geared up for the stretch run.

Week 8 - 0.172 playoff teams lost (8th)

1. Ohio State (80%) at Indiana - 0.038
2. USC at Notre Dame (59%) - 0.035
3. Clemson (89%) at Pitt - 0.023
4. Oregon (65%) at UCLA - 0.012
5. Oklahoma State at Iowa State (73%) - 0.011

A weird weekend that could be fun?  If the Irish have survived everything else to this point, their battle with the Trojans will be one of the last big hurdles to a Playoff berth.  Ohio State and Clemson on the road against weird upsetty teams could be something as well.  Not to mention an under the radar App State-Coastal matchup that will be incredible.

Week 9 - 0.241 playoff teams lost (4th)

1. Penn State at Ohio State (80%) - 0.057
2. Georgia (64%) vs. Florida - 0.052
3. North Carolina at Notre Dame (58%) - 0.035
4. Iowa at Wisconsin (67%) - 0.018
5. Florida State at Clemson (95%) - 0.011
6. Texas Tech at Oklahoma (95%) - 0.010

The metrics indicate this is one of the best weekends of the year.  And sure, if Florida and Penn State can get through their front-loaded schedules in good shape, then the top two on this list will be as important as advertised.  But I'm just not feeling it otherwise.  A good time to trick or treat.

Week 10 - 0.210 playoff teams lost (5th)

1. LSU at Alabama (90%) - 0.031
2. Oregon at Washington (51%) - 0.027
3. Ohio State (86%) at Nebraska - 0.023
4. Texas at Iowa State (64%) - 0.019
5. USC (51%) at Arizona State - 0.019
6. Auburn at Texas A&M (73%) - 0.019
7. Clemson (92%) at Louisville - 0.016

Now here we go.  A battle to be second-best in the Big 12.  A fight for the Pac-12 South.  A couple weird road trips for contenders.  And a bunch of other weird stuff (Indiana goes to the big house, Georgia-Mizzou, and Ole Miss-Liberty for some reason).  I'm all-in on this as maybe the best weekend of the year.

Week 11 - 0.166 playoff teams lost (10th)

1. Texas A&M (64%) at Ole Miss - 0.025
2. Oklahoma (90%) at Baylor - 0.020
3. Arizona State at Washington (63%) - 0.015
4. Notre Dame (72%) at Virginia - 0.013
5. Georgia (87%) at Tennessee - 0.012
6. Michigan at Penn State (69%) - 0.012
7. North Carolina (66%) at Pitt - 0.010

Can it be a road test weekend if Alabama, Clemson, and Ohio State are all at home?  Maybe?  This week could go either way, depending on what happens before it.  But even if the top of this weekend ends up being boring, there's some good stuff buried (Marshall-UAB, Boise-Wyoming) that should make it worthwhile regardless.

Week 12 - 0.184 playoff teams lost (6th)

1. Iowa State at Oklahoma (78%) - 0.067
2. Oregon (54%) at Utah - 0.023
3. Arkansas at Alabama (95%) - 0.010
4. Michigan State at Ohio State (94%) - 0.010

A weird week that has a strong overall rating because of one big game and a deeeep slate of games that could be fun if things break right.  Mizzou-Florida, Oregon State-Arizona State, Cincinnati-SMU, USC-UCLA and more suggest a second straight weekend that will either be awesome or meh.  All the more reason to root for chaos early on.

Week 13 - 0.262 playoff teams lost (3rd)

1. Oklahoma (80%) at Oklahoma State - 0.043
2. Ohio State (79%) at Michigan - 0.039
3. Alabama (84%) at Auburn - 0.038
4. Texas A&M (59%) at LSU - 0.030
5. Wisconsin (65%) at Minnesota - 0.015
6. Notre Dame (73%) at Stanford - 0.012
7. North Carolina (65%) at NC State - 0.011
8. Clemson (95%) at South Carolina - 0.010

Rivalry week where everyone is on the road?  Sign me up.

Conference Races

I'm not going to do the conference-by-conference breakdown, but I will post the summary grid.  I'm doing this because most division races shape up as one of two things (either a two-horse race, or a formless blob of mediocre teams), both of which are kind of boring.  Here's hoping that a few upsets shake up some of these.