2018 Championship Week: Football Bracketology

The calendar might say November, but to me it feels a bit like March.  I have always been fascinated by March madness.  The basketball is fun, of course, but for me even the structure of the tournament is beautiful.  As a kid, every year in March I would painstakingly create a hand-written bracket on Selection Sunday.  As I got older, I gained an appreciation for the subtleties of seeding like S-curve and geographic optimization while simultaneously avoiding rematches. More recently, I have been fascinated (or possibly obsessed) with mathematical analysis of the tournament, such as performance against seed expectation (PASE) and why 2-seeds get upset in the 1st round so much more often than 1-seeds.  I have a traditional of constructing my own bracket in the evening before Selection Sunday.  It is all a tremendously beautiful puzzle.

So, when the college football playoffs were instituted back in 2014, it immediately piqued my interest.  More specifically, I became instantly fascinated by the New Year's Six (NY6) Bowls.  Much like bracketology in March, selecting teams for the Playoffs and other NY6 Bowls is a shifting puzzle to solve, and it is one that pushes all my Sports Geek buttons. 

And so, with December's Selection Sunday coming up in just a few days, I have been crunching some numbers and analyzing the various scenarios that may come to pass on Saturday.  Somethings I can calculate and other things I can speculate, and today I would like to present my findings to everyone as something on which to masticate.

After Tuesday night's penultimate poll, we can get a pretty strong hint as to what the committee is thinking.  In addition, there are certainly several assumptions that need to be made as to the true pecking order of each team relative to each other.  Finally, since we know the Vegas spread for each game on Saturday, we can assign probabilities to each potential outcome to get a feel for the likely and unlikely scenarios.  If I take all things together, I can make a pretty good guess as to how things might play out.  In general, there are several burning questions and observations that we can make:

1) Are Clemson and Alabama already in the Playoffs?

Honestly, I think so.  Even if Pitt and/or Georgia win on Saturday, can you really see the Committee leaving out one or both of these teams?  Yeah, me neither.  While I certainly am not ruling it out, I think it would require a large point differential and something like a major injury to a key player for the Committee to get gun-shy.  Everyone want to see Clemson vs. Alabama in the final.  It has been that way since August. No, it has been that way since January.  One loss is not going to change that.

My only real question is if the committee will tweak the seeds in order to avoid pairing the two teams up in the semi-finals.  For example, if Georgia beats Bama, do you just drop Bama to #3 so you can pair up #1 Clemson and #4 Georgia?  Do you do something weird like raise Notre Dame up to #1?  What happens if Pitt beats Clemson?  Similarly, do you just knock Clemson down to #3?  Honestly, I would absolutely do this if I were on the Committee.  I do not want to see Alabama play either Clemson or Georgia in the semifinals.  I just don't. I am totally OK with rigging the seeds to make this happen.  My #1 mandate for the Committee is simple: create compelling and fair match-ups.

2) Who will that 4th team be?

Based on Tuesday night's poll, I see four or maybe five remaining candidates, and I can estimate the probability that each will make the playoffs:  Oklahoma (58%), Ohio State (19.8%), Georgia (19.3%), and UCF (2.9%).  First, I will comment briefly on the probabilities.

When it comes to analyzing all the different scenarios, there actually aren't that many different combinations.  There are five Power 5 Championship games, which each can have 2 possible outcomes. This leads to a total of 32 combinations (2^5).  However, the outcome of the Pac 12 doesn't really matter, as the winner will go to the Rose Bowl and that's it.  So, this cuts the number in half.  Also, Clemson is a large (-24.5) favorite over Pitt in the ACC Title Game, meaning there is only a 4% chance of an upset.  As a first pass, we can partially ignore this result as well, reducing the number of serious scenarios to only 8.  But, considering that result of the AAC Title Game (Memphis at UCF [-4.5]) will have some impact on the NY, this again increases the number of scenarios a bit.

Based on the most recent playoff poll, I think that it is reasonable to conclude that:

• If Georgia beats Alabama (-13), they are in (along with Clemson, ND, and Bama)
• If Oklahoma (-7.5) beats Texas, they are in over Ohio State
• The only way Ohio State get to the playoffs is if they win (and they are -12.5), but both Oklahoma and Georgia lose.

As for the 8 primary scenarios, this covers 7 of them.  The remaining scenario is the one where Alabama, Texas, and Northwestern all win this weekend, which has about a 4.5% chance of happening.  In this case, things get really interesting.  

In this case, what I would do would be to give UCF the 4th spot (assuming that they beat Memphis, which has a probability of 63%).  However, they are sitting at 8th place in the Playoff Poll, behind a certain team from Ann Arbor.  I am not too concerned about that, as it is easy for the Committee to move a team up a spot when they still have a game left to play.  It is much harder (in principle) to move a team around relative to other teams when they are "in the clubhouse" already.  So, my vote would be to give the "defending champs" a shot, even if it is with a back-up QB.  

If UCF were to lose to Memphis, the Boise State / Fresno State winner is going to claim the NY6 spot (almost certainly in the Fiesta Bowl due to geographic considerations) and there is clearly one less candidate for the last Playoff spot.  I would personally just give it to the team that was ranked the highest currently, in other words, Georgia.  However, there is a chance that the Committee could instead elevate Oklahoma, Ohio State, or even Michigan instead.  I have no way to handicap this possibility.

In any event, the follow table gives the breakdown of the 8 most likely scenarios (Cases 1a-8a) including my calculated odds (derived from the Vegas lines).  As I stated above, however, I am at least curious if the Committee will rig the dice a bit to manipulate the play-off match-up.  Below, I am assuming below that they don't... but I hope that they would.

In the subset of scenarios where UCF loses, Case 5 is different:

Finally, here is the set of additional cases where Pitt upsets Clemson.  The match-ups change a little and as you can see, the odds are very small.  I am also lumping in the cases where UCF loses, as that only would affect Case 13 and the odds are very small.

3) Wait. Go back a minute. So Michigan is still a possible Playoff team?

Yes, maybe.  As I stated above, in the one case where Alabama wins, but both Ohio State and Oklahoma lose, things get interesting.  At best, the odds of this happening are no more than 4.5%.  If we assume that UCF would get the nod over UofM if they beat Memphis (which may not be true), the odds would fall to 1.6%.  Then, if we argue that it would be, at best, a coin flip that UofM would get the nod over Georgia, Oklahoma, and Ohio State, the odds would fall to less than 1%.  That is essentially where I peg the odds, but they come with a lot of speculation.

4) Who is the "Last Team in?"

Based on the latest Playoff Poll, that team is Penn State.  They are currently ranked #12, and since there are 12 total spots, they, in principle, would be the last team selected for the NY6.  However, there is actually only one scenario where this happens, and it is Case 4, where Clemson, Ohio State, Oklahoma, and Georgia win.

In the other 7 scenarios, another team would necessarily pop up into the NY6 from lower in the rankings.  For example, if Northwestern beats Ohio State (4 of the scenarios), the Wildcat go to the Rose Bowl, knocking Penn State out.  Also, the Sugar Bowl is obligated to take a Big 12 team.  So, if Texas beats Oklahoma (2 additional scenarios), or if Oklahoma makes the Playoffs and the Sugar Bowl picks up Texas (1 more scenario), Penn State is out.  There is only a 9.7% chance that the Lions make the NY6.

The next team up appears to be #10 LSU.  (Note the Washington is #11, but as noted above they are either in the Rose Bowl or they will get replaced by Utah, so they don't matter for this analysis).  In the cases where more than one lower ranked team (Northwestern, Texas, or Pitt) makes the NY6, LSU would also get bumped out as well, but I calculate the odds that LSU makes it to be roughly 80%.

Interestingly, this does impact MSU's potential bowl opponent.  I have been assuming that LSU was going to play Penn State in the Citrus Bowl, but my math now says there is only a 20% chance that this happens. So, if I am still correct in my analysis that MSU is going to the Outback Bowl, there is most likely one less SEC team for these Bowls to selection from. 

If we go strictly by the Playoff poll, the remaining ranked SEC teams are:  #15 Kentucky, #18 Mississippi State, #19 Texas A&M, and #24 Missouri.  Those teams are very likely to be distributed to the Citrus, Outback, Gator, and Music City Bowls.  Going strictly by the rankings, Kentucky seems to be the choice for the Citrus Bowl, but the Wildcats have been fading a bit down the stretch and I am not convinced the Citrus Bowl will pick them up.  I have to wonder if Texas A&M or Mississippi State will get that bid instead, leaving Kentucky to play MSU in Tampa.  Honestly, MSU of UK seem a bit meant for each other this year, and the Outback Bowl does tend to take SEC East teams.  I still think that is how this will shake out.  If not, Texas A&M or Mississippi State or I suppose even Missouri (also an SEC East team) will be MSU's opponent.

Getting back to the NY6, there is are a few other very low probability scenarios where even more chaos occurs. Above I mentioned that LSU gets knocked out if 2 out of 3 lower ranked teams (Northwestern, Texas, and Pitt) make the NY6, but what happens if ALL THREE make the NY6?  In these cases (#11 and 13) Florida is the next team to get cut out.  However, the combined odds of this event are only 0.6%.

Finally, the most entertaining scenario is also the least likely, #16.  In this case, all hell breaks loose and all the underdogs win in the four Power Five championships.  In this scenario, I think the committee simply punts and takes the Top 4 are they stand now (Alabama, Clemson, Notre Dame, and Georgia).  If this were to be their decision, the Sugar Bowl would have to take Texas and Florida by contract, and Pitt, Northwestern, and the Group 5 Champ would also all get a NY6 slot.  This would leave only two available spots for Ohio State, Oklahoma, and Michigan.  While it is not assured, it is possible and perhaps likely that Michigan would be the team left out in the cold.

If this does happen, the "no repeat" rule could also prevent Michigan from playing in either the Citrus Bowl or the Outback Bowl.  They could, in principle, slide all the way to the Holiday Bowl.  If MSU does, in fact, wind up in Tampa, this would be perhaps, the most hilarious thing ever.  Alas, the odds of this occurring are no better than 0.04%

5) So, how will the whole NY6 shake out?

Based on the analysis above, I can project the full NY6 slate in each of the relevant scenarios, based on the most likely bowl pairings.  I will start with Cases 1a to 8a, where I assume Clemson beats Pitt and that UCF beats Memphis.  There is one other constraint that I imposed as well.  There are some indications that Georgia will not play in the Peach Bowl, as the SEC Title game will be played their and the fan base would generally rather have a trip to New Orleans and the Sugar Bowl assuming that they are not in the Playoffs.  I am not 100% convinced that this is true, but I made my initial projections based on this assumption.

If for some reason Georgia gets the Peach Bowl as a preference, some of the scenarios change a little:

If UCF losses to Memphis, the Mountain West Champ will get the NY6 bid, and I expect them to get shipped to the Fiesta Bowl.  Again, the match-ups would change a bit in the NY6:

Finally, here is the set of low-probabilities scenarios where Pitt upsets Clemson

In some of these cases, especially the ones where neither Michigan nor Ohio State is in the Playoffs or Rose Bowl, it becomes fairly arbitrary to assign the Buckeyes and Wolverines to either the Peach or Fiesta Bowl.  I would assume that in these cases Ohio State would draw the perceived tougher opponent, but you never know.  Most of those projections are toss ups.

6) Final Thoughts and Analysis

In general, the most likely (highest probability) scenario outcome is Case 1, with a total probability of 45% (including either result for UCF, i.e. Case 1a + Case 1b).  This has Oklahoma in the playoffs, Ohio State in the Rose Bowl, and Michigan playing either Georgia or Florida in the Peach Bowl or LSU in the Fiesta Bowl, depending on where Georgia really wants to go and if UCF wins.

For those that are generally curious about where Michigan will end up, it is really all over the map.  If I assume the Georgia and Peach Bowl rumor to be a 50% proposition, UofM has about a 48% chance to play in the Peach Bowl, a 32% chance to play in the Fiesta Bowl, and a 20% chance at Pasadena.  This all assumes that they do not somehow back into the Playoffs.

As for the Wolverine's possible opponents, I estimate:

Florida (20.2%)
Washington or Utah (19.8%)
LSU (19.6%)
Georgia (15.5%)
UCF (15.1%)
Boise / Fresno (6.7%)
Oklahoma (2.3%)
Pitt (0.7%)

In other words, it is all over the place.

7) Championship Game Breakdown and Predictions:

In case anyone is curious my spreadsheet picks for the week are shown here:



I have UCF covering (likely because my computer doesn't get an injury report) and I have Frenso State and Marshall winning in upsets. Overall, my weekly simulation expects about 3.5 upsets.

As for the individual games:

ACC Title Game: Pitt vs. Clemson (-24.5), 96% chance of victory

Bless you Coach Darduzzi, but I really have a hard time seeing an upset here. Actually, if it were to happen, it would be the second biggest upset of the year relative to the Vegas line (Old Dominion over VA Tech [-28] is #1). That said, I think I read that Pitt upset Clemson the last time they played and there was an interesting throw away line in a preview magazine last year or the year before. It was one of those anonymous quotes from another ACC coach that suggested Clemson struggled with teams like Pitt that actually huddle and don't call in the play from the sideline... perhaps suggesting that Clemson was stealing signals... So, take that for what it is worth. My spreadsheet likes Clemson to cover and I would tend to agree.

Pac 12 Title Game: Utah vs. Washington (-4), 61% chance of victory

As I recall, the Utes picked up several key injuries to skill players over the past couple of weeks and have limped to the finish line. While it would be cool for Utah to make their first Rose Bowl appearance, I expect Washington to roll them. The Huskies already beat them 21-7 on the road back in September. I don't think it will even be that close. My spreadsheet expects a barn burner, but I don't see it. Also, remember that Washington was a preseason Top 5 team. I have been thinking for a while that the Big Ten had a great chance to win the Rose Bowl this year, but I could easily see Washington beat any of the three possible candidates (Northwestern, OSU, or UofM)

SEC Title Game: Georgia vs. Alabama (-13), 82% chance of victory

This one is tricky. Most people seem to think Georgia is not as good as they were last year and Bama is perhaps better. So, the spread makes sense. However, the ESPN's FPI expects a much closer game. The one thing that gives me pause about Bama is QB depth and health. It seems like Tagovailoa is gettting just a bit banged up and if he goes down... who knows. But, at the end of the day, I think it is a safe bet to assume Bama will win every game until some team proves otherwise.

Big Ten Title Game: Northwestern vs. Ohio State (-12.5), 81% chance of victory

Call me crazy, but my gut feeling is that Northwestern wins this one. Sure, OSU looked great last week, but I tend to believe that their lack-luster body of work this year is who they really are. The main questions that I have are 1) Can Northwestern handle the moment, or will they choke? and 2) Will OSU actually respect Northwestern enough to come out with the proper focus? If Thorson comes out and plays like he has against MSU, I really like NW's chances. 

Big 12 Title Game: Texas vs. Oklahoma (-7.5), 70% chance of victory

Honestly, this spread seems a bit high. I mean, Texas already beat Oklahoma once this year, and the Sooners have been winning by the skin of their teeth for the last month. The only team that they have beaten my more than 3 points recently is Kansas, and they have given up 48 points per game doing it. Again, my gut likes the upset here, which is too bad, as a Sooner win would seal the deal to lock Michigan out of the Playoffs (which is likely already true) and the Rose Bowl. 

AAC Title Game: Memphis at UCF (-4.5), 62.6% of victory

UCF is favored, but once again, I like the upset. Memphis is the one team in the AAC that has really pushed the Golden Knights over the past two seasons. With UCF's starting QB out, I think the Tigers get this one.

Mountain West Title Game: Fresno State at Boise State (-3), 58% chance of victory

If UCF does go down, as I fear they will, this game will be for a NY6 spot. My spreadsheet loves Fresno, but I think Boise wins to claim a spot in the Fiesta Bowl.

Now, if you are paying close attention, you may have noticed that my gut feeling picks are exactly the nightmare scenario above (Case 5b) where Michigan could potentially back into the playoffs. I fear this will happen, and for 12 hours or so, there is going to be some freaking out. However, I also predict that Georgia will still get the 4th spot, which will then leave Ohio State and Michigan to play either Oklahoma in the Peach Bowl or Boise State in the Fiesta Bowl. That is what I am feeling.

Well, if you made it this far, I commend you!  I am certainly looking forward to see all this pans out.  Also, MSU still gets to play another game! Yeah! (I think).  Once the bowls are all announced, I plan to take a closer look at MSU's opponent, as well as the entire bowl landscape.  Until then and as always, Go State! Beat (a team to be named later)!