2016 College Football Rankings Primer and Background

Well, California is the #1 team this week. Somehow, I think this will change. At the same time, with a week of college football in the books, I thought I’d debut my own system for ranking teams. A detailed primer is here, but the essential ingredients are the following:

  • Who did you beat? Teams that beat the best teams should get the most credit. Duh.
  • And how do you determine whether it mattered? We use an adjusted points margin to determine team “strength” and use that to weight wins and losses. Basically we take the principle if Team A beats Team B by 10 and Team B beats its opponents by 7 points on average, then Team A has +3 points of strength. (of course you can say Team B’s margin is based on its opponents, we iterate this out until we see no real change)
  • Treat every season independently. No preseason weights. Let the results determine the value of home field, as well as the value of an FCS team (we assume any non FBS team is part of one collective “Other” team)

As a result, every season only has that season as a context. Too often, the polls are dependent on the preseason poll — and teams ranked highly early can sometimes have an easier way to recover from a loss etc. Now fortunately, over time this tends to even out. Anyway, testing this with 2014 and 2015 data, below you can see how we did (Pre-Bowl Top 10s). I will also note how I’d put together a 4 and 8 team field.


2014

  1. Alabama (12–1, 100)
  2. Florida State (13–0, 98.9)
  3. Oregon (12–1, 94.3)
  4. Ohio State (12–1, 89.9)
  5. UCLA (9–3, 89.3)
  6. Mississippi State (10–2, 86.5)
  7. TCU (11–1, 85.7)
  8. Baylor (11–1, 84.6)
  9. Ole Miss (9–3, 84.4)
  10. Arizona (9–3, 84.0)

Analysis: Using the criteria the Selection Committee has used the first two go-rounds, let’s focus on the conference champs. Clearly the Top 4 all won their conferences outright, so I think the Committee got it fine. Alabama v Ohio State, Florida State v Oregon as the two semifinals. If you wanted to do a more representative 8-team event, I’d have given Baylor the Big 12’s automatic spot (beat TCU head to head and their raw scores were very close). Boise State at #19 earned the best of the “Group of 5”. UCLA and TCU would get the at-larges from my committee of one. UCLA’s fundamentals were too high to ignore and with TCU and Mississippi State being in a dead-heat, defer to a conference co-champ. The quarterfinals would have been:

  • (1) Alabama v (8) Boise State
  • (4) Ohio State v (5) UCLA
  • (2) Florida State v (7) TCU
  • (3) Oregon v (6) Baylor — lot of highlighter colours here

2015

  1. Clemson (13–0, 100)
  2. Michigan State (12–1, 95.1)
  3. Alabama (12–1, 93.5)
  4. Oklahoma (11–1, 92.6)
  5. Ohio State (11–1, 90.3)
  6. Notre Dame (10–2, 88.1)
  7. Stanford (11–2, 86.3)
  8. Iowa (12–1, 85.7)
  9. Northwestern (10–2, 82.2)
  10. Oklahoma State (10.2, 81.7)

Analysis: Again, this was a fairly easy job for the committee. The top 4 were fine. Personally I had Michigan State over Alabama because Michigan State played fewer FCS teams. Voters often do not count this, but I do to be fair. For an 8-team affair, Stanford would be added as an automatic bid, as would Houston (#18), as the best of the Group of 5. For at-larges, we go with Notre Dame (#6) and Iowa (#8).

  • (1) Clemson v (8) Houston
  • (4) Oklahoma v (5) Ohio State
  • (2) Alabama v (7) Iowa
  • (3) Michigan State v (6) Notre Dame

On ward to the real season.