1. I start off by compiling computer metrics from three sources: ESPN's FPI, Sagarin's Pure Points, and Bill Connelly's S&P+. I have compared these ratings to results for the past three seasons, and have found that the most predictive formula is composed of 55% FPI, 25% Sagarin, and 20% S&P+. Once I compile this aggregate team rating, I use that to put a win percentage on each game, and then simulate the season 10,000 times.
2. Simulation results in hand, I compile counts of the number of times each power conference team loses 0, 1, and 2 games. I also calculate the average strength of schedule, using the simplest possible calculation. I also do this for Group of Five teams, but I only count the number of times these teams go undefeated. For the purposes of this model, Notre Dame counts as a power conference team, while all other independents do not.
3. Once I have these summary statistics, I plug the results into three separate logistic regression models - one for each "loss bucket." These models are built off of the population of teams from 1998-2017 that finished with the same number of losses, and either made the BCS/Playoff top four or did not (I include the BCS numbers because it gives me a bigger dataset, and the BCS criteria really weren't that different from what the committee does). I sum the results to get my preliminary playoff odds, and then run good ole Solver to get the final sum of playoff odds to 4.*
*The preliminary odds usually sum to lower than 4 because preseason odds tend to be conservative in nature. I've toyed with adding metric uncertainty to the mix with the purpose of alleviating this issue, but that makes the simulation run waaaay longer. If I ever come up with a clever way to do this, you'll be the first to know.
4. The "big" change to the 2018 model is one I actually implemented in late 2017. While Central Florida was making just about the best case for inclusion that a Group of Five team can, the committee was finding more and more comical ways to insult them. It's now clear that it's going to be basically impossible for a non-power team to make the Playoff. Hence I am now treating all undefeated Group of Five teams as two-loss power conference teams. And as basically no two-loss teams with mediocre or worse SOSs have made the top four, so too will no Group of Five teams.
The Rankings
I have included every team that finished at least one simulation with two or fewer losses (or undefeated if you're Group of Five). Sorry Illinois fans.
| Rank | Team | Agg Rank | POFF Prob |
| 1 | Clemson | 2 | 60.03% |
| 2 | Alabama | 3 | 56.09% |
| 3 | Ohio State | 1 | 44.11% |
| 4 | Oklahoma | 7 | 35.26% |
| 5 | Washington | 4 | 34.63% |
| 6 | Georgia | 5 | 28.59% |
| 7 | Notre Dame | 8 | 20.42% |
| 8 | Auburn | 6 | 16.37% |
| 9 | Penn State | 9 | 15.70% |
| 10 | Michigan State | 11 | 14.48% |
| 11 | Wisconsin | 10 | 10.92% |
| 12 | Miami (FL) | 13 | 10.86% |
| 13 | Michigan | 12 | 8.35% |
| 14 | Mississippi State | 15 | 6.58% |
| 15 | Stanford | 14 | 6.30% |
| 16 | Oklahoma State | 18 | 4.74% |
| 17 | Florida State | 16 | 4.15% |
| 18 | USC | 17 | 2.97% |
| 19 | Iowa | 23 | 2.54% |
| 20 | Texas | 19 | 2.52% |
| 21 | Oregon | 22 | 2.13% |
| 22 | Florida | 26 | 1.70% |
| 23 | Virginia Tech | 25 | 1.52% |
| 24 | Texas A&M | 20 | 1.18% |
| 25 | North Carolina State | 34 | 0.89% |
| 26 | TCU | 24 | 0.87% |
| 27 | Louisville | 29 | 0.66% |
| 28 | Utah | 27 | 0.64% |
| 29 | South Carolina | 28 | 0.58% |
| 30 | LSU | 21 | 0.57% |
| 31 | Missouri | 32 | 0.56% |
| 32 | Boston College | 33 | 0.51% |
| 33 | Arizona | 38 | 0.41% |
| 34 | California | 37 | 0.40% |
| 35 | Duke | 31 | 0.24% |
| 36 | Northwestern | 30 | 0.23% |
| 37 | Boise State | 35 | 0.18% |
| 38 | Georgia Tech | 36 | 0.13% |
| 39 | Ole Miss | 41 | 0.12% |
| 40 | North Carolina | 49 | 0.09% |
| 41 | West Virginia | 39 | 0.09% |
| 42 | Kansas State | 40 | 0.08% |
| 43 | Wake Forest | 47 | 0.08% |
| 44 | Washington State | 51 | 0.07% |
| 45 | Arkansas | 52 | 0.07% |
| 46 | UCLA | 43 | 0.06% |
| 47 | Baylor | 48 | 0.05% |
| 48 | Memphis | 42 | 0.05% |
| 49 | Iowa State | 45 | 0.04% |
| 50 | Texas Tech | 46 | 0.04% |
| 51 | Central Florida | 44 | 0.04% |
| 52 | Pitt | 55 | 0.02% |
| 53 | Minnesota | 62 | 0.02% |
| 54 | Nebraska | 50 | 0.01% |
| 55 | Purdue | 53 | 0.01% |
| 56 | Indiana | 61 | 0.01% |
| 57 | Tennessee | 56 | 0.01% |
| 58 | Marshall | 67 | 0.01% |
| 59 | Fresno State | 60 | 0.01% |
| 60 | Syracuse | 66 | 0.00% |
| 61 | Appalachian State | 77 | 0.00% |
| 62 | Maryland | 57 | 0.00% |
| 63 | Toledo | 69 | 0.00% |
| 64 | South Florida | 63 | 0.00% |
| 65 | Florida Atlantic | 65 | 0.00% |
| 66 | Kentucky | 58 | 0.00% |
| 67 | San Diego State | 59 | 0.00% |
| 68 | Troy | 85 | 0.00% |
| 69 | Ohio | 74 | 0.00% |
| 70 | Arizona State | 54 | 0.00% |
| 71 | Northern Illinois | 68 | 0.00% |
| 72 | Arkansas State | 83 | 0.00% |
| 73 | Western Michigan | 78 | 0.00% |
| 74 | Houston | 64 | 0.00% |
| 75 | North Texas | 88 | 0.00% |
| 76 | Utah State | 70 | 0.00% |
| 77 | Virginia | 71 | 0.00% |
| 78 | Colorado | 72 | 0.00% |
| 79 | Wyoming | 79 | 0.00% |
| 80 | Louisiana Tech | 76 | 0.00% |
| 81 | New Mexico State | 107 | 0.00% |
| 82 | Buffalo | 96 | 0.00% |
| 83 | UAB | 97 | 0.00% |
Alabama, Clemson, and Ohio State are all basically tied for first in the aggregate ranking, but the Buckeyes fall a bit behind the other two in Playoff odds. The reason for this is that they have a peculiar schedule. There are more stiff challenges (road games against three top 25 teams), which makes them more likely to lose two or more games than the other teams. At the same time the rest of their schedule is pretty soft, which means their overall SOS ranking falls below the Tide and the Tigers. Regardless, this group is clear of the field, and I would be surprised if at least one of them doesn't make the Playoff.
Further down the rankings we have Notre Dame coming in higher than you might think. They were also higher on this list than in public perception last year, and that worked out well for a while. Iowa and Florida get small bumps for playing in "easy" divisions, while LSU and Texas A&M get dinged for playing in the division of pain. Boise is the highest Group of Five team, largely because the Mountain West is easier than the American.
If a conference gets two teams, it's probably going to be the SEC again. In fact the SEC has better odds of placing a team in the Playoff than the Big 12 and Pac 12 combined.
| Conference | Exp Playoff Teams |
| SEC | 1.124 |
| B10 | 0.964 |
| ACC | 0.792 |
| P12 | 0.476 |
| B12 | 0.437 |
| IND | 0.204 |
| MWC | 0.004 |
| AMER | 0.001 |
| CUSA | 0.000 |
| SB | 0.000 |
| MAC | 0.000 |
Conference Title Races
We start with the ACC, where there are a couple of runaway favorites. And given that the second-place teams are shaky (coming off a bad year and injured as hell, respectively - and they play each other!), we have to really dream to get interesting races. So I'll just say that Duke is a top-40 team in all three computer rankings and is sitting at 7%. Dream no more.
Next is the American, which has decent odds of featuring a rematch of last year's crazy finale. If you're looking for usurpers, the West features the most balanced set of challengers, but I'm a little partial towards a Temple team that got better in a hurry last season.
Hey look another potential rematch! What fun! In all seriousness, I think the Big Ten will be a little more chaotic than it's been recently, as at least one of the second-year coaches should see a breakout. In Husker news, Scott Frost's first team will probably be something like the fourth best team in its division, but an impossible schedule takes an outside shot at a title game run off the table.
The Big 12 odds at the start of 2018 look almost the same as the odds at the start of 2017. Iowa State and Texas make small advances, and Oklahoma looks to be slightly less dominant, but there's nothing too shocking here. You may think West Virginia is underrated compared to the polls, but the team's depth is perilously thin, and I don't think the defense will be all that great.
The CUSA remains the most wide open conference. But instead of this being because there are no good teams, it's now because there's a few halfway decent teams. FAU is probably underrated here, but it should still be a coin toss with Doc Holliday's second major surge of talent during his Marshall tenure. On the other side, UAB has a 1 in 5 chance of being a division champ just 18 months removed from not being a team.
The MAC division odds look exactly as I would expect they would look. There is roughly a 1% chance of a Buffalo-Eastern Michigan title game...let's root for that.
Boise is the devil, Wyoming is going to bounce back after losing Josh Allen, and the West is a literal coin flip. This isn't vintage WAC or anything, but there's enough here to keep me interested.
We have a 50% chance of the obvious favorites meeting for the Pac 12 title. Outside of that, Utah is the only team to have never won the South, and has a reasonably good chance of pulling it off in spite of a difficult cross-division schedule. On the other side, Cal is the quietest 2% chance of the year.
Alabama and Georgia are here to murder you to death, but don't sleep on the mildly surprising third place teams. A&M and LSU would have a chance in any other league but not here. And Mizzou has a faint chance in spite of having to play Alabama and having to employ Derek Dooley.
The biggest complaint to my blog last year was that the Sun Belt didn't have divisions. So I put in a call to the league office and kaboom. You're welcome.
Weekends, Ranked
Now for my favorite part: previewing every weekend of the season, and telling you what's going to be great about them. As I've done previously, I will use a rather crude formula called "Playoff Teams Lost" that simply multiplies the odds of each team losing by 1/3 of their playoff odds. This means that middling games involving a top team will often be ranked ahead of a close-matched game between lower-ranked teams. You may disagree with that methodology, but since the Playoff is so central to the game (and my blog), I think it is appropriate. I have left out Week 0, as well as Championship week and the Army-Navy week, as none of those weeks feature anything currently scheduled that will have playoff implications.
Week 1 - 0.199 playoff teams lost (Rank: 8th of 13)
1. Washington (55% win probability) vs. Auburn - 0.082 playoff teams lost
2. Michigan at Notre Dame (60%) - 0.044
3. Alabama (87%) vs. Louisville - 0.025
A reasonably solid opening weekend features an impactful trio of games. The loser of Washington-Auburn won't be eliminated yet, while Notre Dame and Michigan have hard enough schedules such that neither can afford to give this game away. And Louisville is probably better than you think (but they still won't beat Bama).
Week 2 - 0.139 playoff teams lost (Rank: 12th)
1. Clemson (79%) at Texas A&M - 0.045
2. Georgia (78%) at South Carolina - 0.023
3. USC at Stanford (60%) - 0.014
What week 2 lacks in guaranteed marquee games, it makes up for in a deep slate of pure fun. Mississippi State goes to Manhattan. Penn State goes to Pitt. Michigan State is in the desert for some reason. And two of the best teams in the nation go on the road to sneaky good maroon teams. Plus there's El Assico.
Week 3 - 0.150 playoff teams lost (Rank: 11th)
1. Ohio State (82%) vs. TCU - 0.029
2. Washington (80%) at Utah - 0.025
3. Alabama (88%) at Ole Miss - 0.022
Home underdog week strikes early, as we also have Oklahoma at Iowa State and Miami at Toledo (?). This is probably the Rebel's best chance to stun everyone before their depleted roster falls apart, so let's all root for them.
Week 4 - 0.178 playoff teams lost (Rank: 9th)
1. Texas A&M at Alabama (86%) - 0.030
2. Clemson (87%) at Georgia Tech - 0.026
3. Georgia (80%) at Missouri - 0.021
Georgia's second road game against a feisty team. A&M's second game against a top-two team. Wisconsin-Iowa in a battle for the Big Ten West. Oregon-Stanford in a weirdly early battle. And Notre Dame playing at Wake the week their QB returns from suspension. This is a sneaky good week.
Week 5 - 0.212 playoff teams lost (Rank: 6th)
1. Ohio State (64%) at Penn State - 0.087
2. Stanford at Notre Dame (66%) - 0.037
3. Baylor at Oklahoma (89%) - 0.013
This may be the highest ranked weekend of the season to date, but it's largely because the second biggest game of the year happens in it. If you're going to skip a week, make it this one (but do DVR OSU-PSU).
Week 6 - 0.229 playoff teams lost (Rank: 4th)
1. Oklahoma (71%) vs. Texas - 0.040
2. Auburn (61%) at Mississippi State - 0.035
3. Notre Dame (68%) at Virginia Tech - 0.025
Now this is a weekend. ND plays another game they're mild favorites in. Mississippi State gets their best chance to challenge the powers in the West. Alabama goes to an Arkansas team that might have congealed a bit at this point. And Wake hosts another top team in Clemson. Add in the usual rivalries (Red River, FSU-Miami), and you have a full slate.
Week 7 - 0.209 playoff teams lost (Rank: 7th)
1. Michigan State at Penn State (58%) - 0.050
2. Washington (75%) at Oregon - 0.034
3. Wisconsin at Michigan (53%) - 0.032
Can I interest you in some Big Ten football? This is a shallow week, but these games (plus LSU-Georgia) should make for some fun. Also, my system thinks Indiana-Iowa is the 12th most exciting game of the weekend, so I'm throwing it out and starting from scratch.
Week 8 - 0.173 playoff teams lost (Rank: 10th)
1. Michigan at Michigan State (57%) - 0.037
2. Oklahoma (71%) at TCU - 0.036
3. NC State at Clemson (91%) - 0.020
This is the last underwhelming week before the season kicks it up to eleven, but anytime the fates make Ohio State go to Purdue and Penn State go to Indiana, you have to watch. Utah hosts USC if you're hoping for the Trojans to get embarrassed again.
Week 9 - 0.212 playoff teams lost (Rank: 5th)
1. Clemson (75%) at Florida State - 0.060
2. Georgia (79%) vs. Florida - 0.024
3. Washington (83%) at Cal - 0.020
Here's where the season gets going in earnest. Clemson plays their main division challenger. Georgia gets their third interesting test against a division rival away from home. Penn State and Iowa renew their weirdly exciting rivalry. And Miami goes to BC, which always makes me think of this.
Week 10 - 0.268 playoff teams lost (Rank: 3rd)
1. Stanford at Washington (75%) - 0.045
2. Penn State at Michigan (53%) - 0.041
3. Alabama (80%) at LSU - 0.038
Now here we go. The game to decide the Pac 12 North (probably). Another Big Ten coin flip. Manball Christmas. And ND takes a fun rod trip to Northwestern, and Georgia goes on the road again (to Kentucky but still), and Duke gets their chance against Miami. Should be good.
Week 11 - 0.393 playoff teams lost (Rank: 1st)
1. Ohio State (66%) at Michigan State - 0.082
2. Auburn at Georgia (58%) - 0.072
3. Mississippi State at Alabama (81%) - 0.054
4. Wisconsin at Penn State (56%) - 0.043
5. Oklahoma State at Oklahoma (75%) - 0.042
6. Florida State at Notre Dame (70%) - 0.030
7. Clemson (86%) at Boston College - 0.029
The best week of the year gets a few extra games just because. The quality drops off rapidly after this septet, but that should be enough games to fill the day with intrigue and drama. Can't wait.
Week 12 - 0.119 playoff teams lost (Rank: 13th)
1. Duke at Clemson (91%) - 0.019
2. Miami (60%) at Virginia Tech - 0.018
3. Ohio State (93%) at Maryland - 0.010
This week is...less good. But you'll need the rest, because...
Week 13 - 0.317 playoff teams lost (Rank: 2nd)
1. Auburn at Alabama (68%) - 0.097
2. Michigan at Ohio State (76%) - 0.056
3. Notre Dame (59%) at USC - 0.034
...the next week has the best game of the year. And all of the trappings of rivalry week. And TCU-Oklahoma State for some reason. Cool.
