The Ideal College Football Playoff – The 2018 Football Rankings Primer

Well, we are 3 weeks away from the beginning of peonage football – an area this space has obsessed over in the past.  Certainly, after collecting data and attempting to rank teams more perfectly over 10 years, there is no reason to stop.  That said, this year the rankings and ideas will be less about predicting the College Football Playoff – a silly exercise by a committee which has been consistently inconsistent about – and ranking teams in terms of an ideal playoff.

It seems inevitable that the College Football Playoff will expand to 8 teams.  The success of the playoff has been huge as expected, and for the most part the title games themselves have been extremely exciting.  But there are obvious flaws with the current system:

  • Only a subset of the games the 130 FBS teams play matter to the National Championship.  The Sun Belt, Mountain West, Conference USA – have no real say in the playoffs.  It is like they are playing their own subseason.
  • The criteria for being a Final 4 team is amorphous at best.  Is winning a league important?  Do losses matter?  Alabama won the title after getting in primarily because they did not have to play a title game.  Does that make sense?

So I make a modest proposal for a real 16 team playoff – which makes every conference relevant, while giving conference championships significant extra weight.  Now, I think the large 14 team conferences should just abandon counting cross-division games – but that is a different conversation.  So how would the Best CFP play out?

  • The 10 conference champions automatically qualify (Big Ten, ACC, Pac-12, Big 12, SEC, C-USA, Mountain West, Sun Belt, MAC, American)
  • There are 6 at-large teams.  The final rankings determine this.  I will use a modified Bradley Terry Model to rank the teams.
  • The bracket is populated as follows:
    • The conference champions from the Big Ten, ACC, Pac-12, Big-12 and SEC all automatically get priority seeding
    • If the best of the Group of 5 Champions finishes higher ranked than any of the Power 5 winners, that team also gets priority seeding
    • Any Independent that finishes in the Top 6 gets priority seeding
    • The priority seeds get ranked and seeded first
    • The remaining teams get seeded afterward

If we look at a Bradley Terry model of 2017’s teams (entering bowl season) the seeding would have worked likes this:

  • The following teams get priority seeds
    • SEC – Georgia (#1)
    • ACC – Clemson (#2)
    • Pac-12 – USC (#13)
    • Big 12 – Oklahoma (#3)
    • Big Ten – Ohio State (#8)
  • Central Florida (#4) would have also received priority seeding.  There are no independents in the Top 6
  • The priority ranking goes: Georgia, Clemson, Oklahoma, UCF, Ohio State, USC
  • The remaining bids
    • Wisconsin (#5, at-large)
    • Auburn (#6, at-large)
    • Alabama (#7, at-large)
    • Penn State (#9, at-large)
    • Notre Dame (#10, at-large)
    • Miami-FL (#11, at-large)
    • Boise State (#39, Mountain West Champ)
    • Toledo (#43, MAC champ)
    • Florida Atlantic (#58, C-USA champ)
    • Troy (#69, Sun Belt Champ)
  • The S-Curve
    • Georgia
    • Clemson
    • Oklahoma
    • Central Florida
    • Ohio State
    • USC
    • Wisconsin
    • Auburn
    • Alabama
    • Penn State
    • Notre Dame
    • Miami-FL
    • Boise State
    • Toledo
    • Florida Atlantic
    • Troy
  • The first round – if we assume the Round of 16 is a home game – and we move things around to prevent teams in the same conference from meeting as long as possible (and use divisions if possible to work around it)
    • ATLANTA REGION: (1) Georgia v (4) Troy, (2) Wisconsin v (3) Notre Dame
    • GLENDALE REGION: (1) Central Florida v (4) Boise State, (2) Ohio State v (3) Miami-FL
    • MIAMI REGION: (1) Clemson v (4) Florida Atlantic, (2) Auburn v (3) Penn State
    • DALLAS REGION: (1) Oklahoma v (4) Toledo, (2) USC v (3) Alabama

A quick inspection of this idea shows a few things:

  • Winning the conference is a big deal.  For the lower teams it means actually qualifying for the National Championship.  For the Power 5, it is the difference in drawing a home game against a team like Florida Atlantic, or having to go on the road against a real opponent.  Alabama’s fate is particularly impacted here.
  • Using the actual Bradley Terry rankings and running a quick Monte Carlo simulation …
    • Georgia 27.26%
    • Clemson 21.32%
    • Oklahoma 12.25%
    • Central Florida 12.05%
    • Alabama 6.30%
    • Wisconsin 5.79%
    • Ohio State 5.48%
    • Auburn 5.08%
    • Penn State 2.80%
    • Notre Dame 1.00%
    • Miami-FL 0.47%
    • USC 0.28%
    • Boise State 0.01%
  • So 13 teams won the title over 10,000 trials.  USC’s high seed did not help much since they were paired with Alabama.  At the same time, Alabama’s seeding made winning the title hard – the penalty for not winning the SEC is considerable.

——————————————————————————————————————–

For good measure, here are how the seeds would have worked out for the previous years

2016: Clemson’s stirring win over Alabama.  One of the blind spots of the model is that it is backwards looking – so if Clemson visibly did not give a crap (as they didn’t) it underrates their chances.  That said, Alabama was the strongest team by far at least based on the games played.

  • NEW ORLEANS REGION: (1) Alabama v (4) San Diego State, (2) Michigan v (3) USC
  • DALLAS REGION: (1) Penn State v (4) Temple, (2) Western Michigan v (3) Stanford
  • PASADENA REGION: (1) Washington v (4) Western Kentucky, (2) Ohio State v (3) Colorado
  • MIAMI REGION: (1) Clemson v (4) Appalachian State, (2) Oklahoma v (3) Wisconsin
    • Alabama 69.21%
    • Washington 10.98%
    • Ohio State 7.26%
    • Clemson 3.67%
    • Michigan 2.51%
    • Penn State 2.27%
    • Wisconsin 1.13%
    • USC 0.86%
    • Western Michigan 0.82%
    • Colorado 0.62%
    • Stanford 0.38%
    • Oklahoma 0.27%
    • Temple 0.01%
    • Appalachian State 0.01%

2015: Alabama beats Clemson in the first chapter of their three year rivalry.  The model has it 3 to the post between Alabama, Clemson and Oklahoma.

  • NEW ORLEANS REGION: (1) Alabama v (4) San Diego State, (2) Iowa v (3) TCU
  • PASADENA REGION: (1) Michigan State v (4) Bowling Green, (2) Stanford v (3) Houston
  • GLENDALE REGION: (1) Oklahoma v (4) Arkansas State, (2) Notre Dame v (3) Ole Miss
  • ATLANTA REGION: (1) Clemson v (4) Western Kentucky, (2) Ohio State v (3) Oklahoma State
    • Alabama 28.43%
    • Oklahoma 22.58%
    • Clemson 16.61%
    • Michigan State 8.84%
    • Ohio State 6.81%
    • Stanford 3.46%
    • Notre Dame 3.46%
    • Iowa 3.35%
    • TCU 2.63%
    • Oklahoma State 1.91%
    • Houston 1.08%
    • Ole Miss 0.48%
    • Western Kentucky 0.24%
    • Bowling Green 0.12%

2014: Ohio State makes the Final Four controversially then goes out and beats Bama and Oregon to win it all.  The road would have been tougher in the 16.

  • ATLANTA REGION: (1) Alabama v (4) Louisiana-Lafayette, (2) Ole Miss v (3) Georgia
  • MIAMI REGION: (1) Florida State v (4) Marshall, (2) Ohio State v (3) Auburn
  • GLENDALE REGION: (1) Oregon v (4) Northern Illinois, (2) TCU v (3) LSU
  • DALLAS REGION: (1) Baylor v (4) Memphis, (2) Mississippi State v (3) Boise State
    • Alabama 35.10%
    • Oregon 21.60%
    • Mississippi State 10.08%
    • TCU 9.90%
    • Baylor 7.47%
    • Ole Miss 5.22%
    • Florida State 4.41%
    • Ohio State 2.97%
    • Auburn 2.52%
    • Georgia 0.27%
    • LSU 0.18%
    • Marshall 0.18%
    • Boise State 0.09%

2013: The Florida State-Auburn collision course

  • MIAMI REGION: (1) Florida State v (4) Louisiana-Lafayette, (2) Arizona State v (3) Oklahoma State
  • DALLAS REGION: (1) Stanford v (4) Fresno State, (2) Missouri v (3) Central Florida
  • PASADENA REGION: (1) Michigan State v (4) Rice, (2) Alabama v (3) Oregon
  • NEW ORLEANS REGION: (1) Auburn v (4) Bowling Green, (2) Baylor v (3) Ohio State
    • Florida State 85.45%
    • Missouri 5.45%
    • Alabama 4.76%
    • Michigan State 1.96%
    • Auburn 0.84%
    • Arizona State 0.56%
    • Stanford 0.42%
    • Ohio State 0.42%
    • Baylor 0.14%
Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s