Ranking College Football Teams by Strength of Record

The College Football Playoff has been controversial since its conception. By including only the top four teams in the entire FBS (just 3% of 130 teams), several deserving and skilled teams are left out annually. Furthermore, the seemingly inconsistent approach of the CFP Committee has left college football fans confused. Just earlier this year, the Committee had Oregon one spot ahead of Ohio State, citing the head to head victory as their reasoning despite several advanced metrics suggesting Ohio State was the better team, but they also had Michigan ahead of Michigan State despite a head to head loss. An even more pressing issue is that it not all teams have a realistic chance to make the playoff when the season starts. There have been two instances of an undefeated Group of 5 team missing the College Football Playoff, with UCF in 2017 and Cincinnati in 2020 both left out of the top four after going 13-0. To eliminating the inconsistency of the CFP Committee, we need a better, more objective way to choose the top 4 teams in College Football or an expanded playoff with more than 4 teams.

An undefeated Cincinnati team missed the CFP in 2020 despite being undefeated in the regular season

Strength of Record: An Objective Way to Rank Teams

Instead of having a committee choose teams for the College Football Playoff, there should be a concrete procedure for selecting the teams. One easy way to choose the top 4 teams would be to create a metric that captures the overall strength of a team’s wins and the weakness of their losses. This means that we obviously shouldn’t use win-loss record alone, but reward more for beating good teams and punish more for losing to bad teams. The best existing metric for this is ESPN’s Strength of Record. ESPN’s Strength of Record metric “reflects chance that an average Top 25 team would have team’s record or better, given the schedule.” This metric analyzes a team’s resume, uses only their schedule, wins and losses, and opponent strength. Here is the current top 15:

Obviously, Georgia ranks number 1, and the rest makes sense as they are followed by Michigan, Alabama, and Cincinnati. Some strengths of using strength of record to rank teams is that it is objective, relies only on schedule and record, and does not rely on a team’s rating (which is a predictive metric ranking teams based on per-play efficiency). However, there are weaknesses, most important of which is an inability to incorporate head to head results (like Ohio State’s win over Michigan State).

Using My Own Model

Ranking by strength of record is not only limited to ESPN’s FPI model, but rather can be utilized with any predictive college football model. For example, I have created my own strength of record metric using my own college football model. My model was created on a per-drive basis rather than a per-play basis, where I used the starting field position, ending field position, and the drive result (all adjusted for strength of schedule) to create offensive and defensive efficiency ratings. The rankings for my model preceding the Conference Championship games are shown below.

SAFER stands for Schedule Adjusted Football Efficiency Ratings

My model has several issues, like overrating some group of five teams, such as Appalachian State and Air Force, and underrating some good teams from this year, like Oklahoma and Cincinnati (both of which are highly ranked in almost every predictive model). In order to create my strength of record, I found the expected win-loss record of the average top 25 team based on the schedule for each team, then used the win percentage over expected to rank the teams. The rankings are shown below.

Compared to the current CFP Rankings (week 14), the top 4 teams are the same and are also in the same order. Outside of the top 4, however, my strength of record has Ole Miss, Michigan State, and Iowa ranked higher than the CFP Committee’s rankings, while Baylor, Ohio State, and Pittsburgh are ranked lower. Ole Miss is ranked in the top 5 despite 2 losses because they played Alabama, Tennessee, Mississippi State, and Auburn (each of which is in the top 22 of NCAA SAFER) all on the road. An average top 25 team would have had just an 11% to win against Alabama, 36% against Mississippi State, 46% against Tennessee, and 51% against Auburn. Meanwhile, Michigan State is rewarded for a very difficult game at Ohio State and challenges against Michigan (home) and Purdue (road), and Iowa was rewarded for winning 2 of 3 road games against Wisconsin, Nebraska, and Iowa State. Baylor is ranked low because they had 8 games which the average top 25 team would have an 80%+ chance to win (partially due to my model underrating Oklahoma) and Pittsburgh is low since their only challenge in the ACC was a home game vs Clemson (they avoided playing NC State, Wake Forest, and Louisville).

Both the CFP’s rankings and my model’s strength of record have Georgia as the number 1 team

2020 Predictions

Using my SAFER rankings and strength of record, I also looked at the results from the 2020 college football season. The SAFER rankings are less accurate as a result of scheduling abnormalities due to COVID-19, and the number of games played was not accounted for because I used win percentage in my strength of record, not total wins. The top 25 of the SAFER rankings for 2020 are shown below.

In 2020, Alabama, Ohio State, Clemson, and Notre Dame were selected to the College Football Playoff. My model had each team in the top 7. BYU was likely overrated because it was difficult to determine strength of schedule without a large sample of non-conference games. Additionally, undefeated (in the regular season at least) Coastal Carolina was ranked higher than undefeated Cincinnati (although this also could have been due to inaccurate strength of schedule). One positive things about my model is that it correctly predicted Alabama as the champion and it predicted Ohio State beating Clemson in the CFP semi-final. The strength of record from these rankings are displayed below.

The strength of record for 2020 correctly predicted the four teams in the CFP, although the order was incorrect. Ohio State was punished by the committee for playing far fewer games than ACC, SEC, and Big 12 teams, but they were not punished in strength of record. These strength of record ranking show a glaring issue with the College Football Playoff: its size. Coastal Carolina, San Jose State, and Cincinnati all went undefeated before bowl season, but none were even considered for the top four, even though they won all possible games. Therefore, the CFP needs to expand, preferably to 12 or 14 teams, to give all teams a chance regardless of whether they are a member of a Power 5 conference or not.

Coastal Carolina had no chance to win a championship last year despite going undefeated, showing that the CFP needs to be expanded

What a 14 Team Playoff Looks Like

Using my strength of record rankings, I looked at what a 14 team college football playoff would look like as of right now (before conference championship games). Some of my personal suggestions for an expanded playoff are that there should be first-round byes (similar to the NFL) and true home games for the higher seed in order to incentivize winning deep into the season. Therefore, there would still be dire consequences for a loss or several losses. Additionally, I would suggest that there are no automatic qualifiers. My reasoning for this is that conference champions should automatically qualify unless divisions within conferences are expunged. Currently, the division format gives the possibility for a very undeserving team to win the conference. Take the Big Ten, for example. Ohio State, Michigan, Penn State, and Michigan State are all in the East division, while the West division contains Wisconsin and Iowa.

Below are the matchups in a potential 14 team college football playoff. I used ESPN’s strength of record to create the rankings, but changed the order a little to avoid regular season rematches in the first round.

  • First-round byes: 1 Georgia, 2 Michigan
  • 14 Wake Forest at 3 Alabama
  • 13 Oklahoma at 4 Cincinnati
  • 12 Oregon at 5 Oklahoma State
  • 11 Iowa at 6 Notre Dame
  • 10 Baylor at 7 Michigan State
  • 9 Ohio State at 8 Ole Miss

An extended playoff would have numerous benefits. First, all teams would have a realistic chance to win a championship regardless of their conference. Second, the playoffs would have more excitement since more teams are included, and fans can see more standout players (like Oklahoma’s QB Caleb Williams, Ohio State’s QB CJ Stroud, Ole Miss’s QB Matt Corral, and Michigan State’s RB Kenneth Walker) playing meaningful games. It seems like after one or two losses, a team is eliminated from a chance at winning a championship, so their games don’t seem to matter as much. With an extended playoff, teams will have more meaningful games further into the season, and fans and players alike will be more engaged.

If the CFP is expanded, fans will get to see more NFL Draft prospects like Matt Corral play in more meaningful and difficult games

There are several issues with the current College Football Playoff. The CFP Committee is inconsistent with their rankings, and the playoff field is too small. The committee should be eliminated and replaced by a mathematical ranking system. This ranking system should rank teams based on their strength of record, using a composite of several predictive measures to determine the strength of opponents and expected win probability for an average top 25 team. Additionally, the playoff field should be expanded from 4 teams to 12 or 14 teams to increase entertainment levels and give all teams a chance. This would not decrease the importance of the regular season, but rather increase it since more teams have more meaningful games and there is an incentive to win in order to achieve home playoff games and a bye. In addition, there would be more top-tier matchups between top 15 teams during the playoffs. The sport of college football would be better off without a committee ranking teams and with a different playoff format than the one currently in place.


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 )

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