Merging Pandemic Obsessions – Building a Better College Football Playoff

So – with 6 weeks in the can, the 2021 College Football Rankings are starting to take shape. With the pandemic in full force last year, it was hard to do a computer model with almost no interconference play of consequence. The Big Ten and Pac-12 played tiny seasons! Given that, this year is a bit of a return to “normal”. One of the fun stories of the offseason were mouth noises from powers that be to expand the College Football Playoff. Of course, now with a hint of a plan, there comes resistance to having anything nice.

That said, idle minds wander – and got me thinking … what would a more inclusive playoff look like, and how could you do it in a reasonably fair way? To craft a perfect (to me) playoff, I had a few goals:

• It should reward conference champions. It’s the majority of a team’s schedule. The primacy of winning that makes sense to me.
• There should be diversity among regions and conferences. One of the tough parts of ranking teams is the general lack of nonconference games compared to other sports. It creates some wide error bars. Going strictly by a Bradley Terry model is not sufficient.
• Make sure the best teams in the country (by season accomplishment) are in the field

At the same time, with a year working at home and having children and weekend afternoon obligations often, European soccer became a real attraction. The games go on early in the States, only take two hours – and are exciting when good teams play. At the top of this is the annual UEFA Champions League (UCL), and its lower level competition the Europa League. What is interesting is that the Champions League – in broad brushstrokes – is a bit like the challenge of the College Football Playoff. You have a lot of teams from different leagues and you have to adjudicate who enters the competition. So the Champions’ League makes an obvious inspiration when looking at what a good College Football Playoff would look like

A 16 team UCL Inspired College Football Playoff

So, how do we implement this? UEFA uses 5-year performance to determine Coefficients for each League and allots spots accordingly. I’m not interested in the long long season for every team – we have to be practical. Instead, I used a 3-year club coefficient – basically the time for a particular class to be really productive over a career.

Since I’ve been ranking teams since 2008 – I have end of season rankings for all the teams. I assigned points based on post-Bowl rank.

• Top 5 – 8 points
• 6-10 – 5 points
• 11-25 – 3 points
• 26-40 – 2 points
• 41-50 – 1 point

Using these point values we calculate 3-year averages and then apply them to the conference affiliation the following year. This means that a conference gets the benefit of its new members for playoff spots. Using these coefficients we allocate the 16 playoff positions to the ACC, Big Ten, SEC, Big 12, Pac-12 and Other (every other conference and independents). The Top 4 seeds go to the Top 4 Conference Champs (and if an Independent cracks the Top 4) Finally, brackets are shifted as necessary to avoid intraconference first round matchups.

Example: Fitting the 2019 College Football Playoff

Using the 3-year formula, the 2016-18 ranks ended up with the following points and berths

• ACC: 24.7 points (2 berths)
• Big 12: 15.7 (2)
• Big Ten: 28.7 (3)
• SEC: 37.7 (4)
• Pac-12: 17.3 (2)
• Other: 26 (3)

Applying the relative weights to a 16 team field and using the “pre bowl” ranking, we get the following

• ACC – Clemson (3), Virginia (36)
• Big 12 – Oklahoma (4), Baylor (15)
• Big Ten – Ohio State (1), Penn State (6), Wisconsin (7)
• SEC – LSU (3), Georgia (5), Alabama (8), Auburn (9)
• Pac-12 – Oregon (14), Utah (17)
• Other- Notre Dame (12), Memphis (13), Navy (16)

The rankings result in the following first round matchups:

• (16) Virginia at (1) Ohio State
• (9) Notre Dame at (8) Alabama
• (12) Oregon at (5) Georgia
• (13) Navy at (4) Oklahoma
• (14) Baylor at (3) Clemson
• (11) Memphis at (6) Penn State
• (10) Auburn at (7) Wisconsin
• (15) Utah at (2) LSU

Using the implied odds from the ratings, a 10,000 trial Monte Carlo Simulation yields

• Ohio State wins 65.65% of the time
• LSU wins 29.72% of the time
• Clemson wins 2.72% of the time
• Penn State wins 0.28%
• Wisconsin wins 0.31%
• Alabama wins 0.73%
• Notre Dame wins 0.29%
• Auburn wins 0.14%
• Memphis wins 0.13%
• Baylor wins 0.02%
• Utah wins 0.01%

Rankings and fit through Week 6

Based on the 2019 rankings, the same berth allocation holds for 2021. How does it look through 6 weeks?

• ACC (2 teams): Wake Forest (9), Clemson (31)
• Big 12 (2 teams): Oklahoma (4), Oklahoma State (6)
• Big Ten (3 teams): Iowa (3), Michigan (5), Penn State (7)
• SEC (4 teams): Georgia (1), Kentucky (10), Ole Miss (15), Alabama (16)
• Pac 12 (2 teams): Arizona State (25), Oregon (26)
• Other (3 teams): Cincinnati (2), Notre Dame (11), SMU (12)

The mock playoff based on only the 6 weeks played so far?

• (16) Clemson at (1) Georgia
• (9) Kentucky at (8) Wake Forest
• (12) Ole Miss at (5) Michigan
• (13) Alabama at (4) Oklahoma
• (14) Arizona State at (3) Iowa
• (11) SMU at (6) Oklahoma State
• (10) Notre Dame at (7) Penn State
• (15) Oregon at (2) Cincinnati

As time allows, perhaps we’ll stroll down recent memory lane to see how this would have looked in days of relative yore.