Creating a formula for a top 25: The final result for 2022

Dec 8, 2022 - 3:06 PM

What is a college football “top 25”? Why 25 teams?

The Associated Press has been ranking college football teams since 1936, and it wasn’t until 1989 that they even ranked 25 teams. For much of its existence, the AP ranked 20 teams...except for that weird little seven year gap from 1961 to 1967 where they ranked 10...the ‘60’s were experimental for all it seems.

The AP Poll is comprised of 63 voters, all being sportswriters and broadcasters from across the country, and their importance and influence on college football had been as powerful and influential as anyone’s. As many of you know, it wasn’t up until 1998 that the top two teams in the country actually played each other for the National Championship under the BCS system, and that ran all the way through 2013 before the playoff took over. We used to be a proper country.

The AP Poll was used from 1998 to 2004 in the BCS formula alongside the Coaches Poll, and its rankings prior to 1998 are widely accepted by many as national title claims.

Since 2014, however, the a college football top 25 has lost its, let’s say dignity, as the CFP Committee’s introduction as the end-all be-all top 25 has been about as imperfect of a ranking as there’s ever been.

Alas, after this long winded intro, it’s time to explain what I’m doing here. Prior to the season, I created my own formula for a top 25. Not because of the current state of the rankings, but because it’s fun! For years, I tried to decipher a way to create a numerical value to a win and a loss, and I think I figured out one way to do just that.

What is the value of a win?

There is no definite number one can place on a win, but it never hurt to try. Let’s start with the basics in this formula. It is very simple. Winning is good, losing is bad. Circumstances within that is where the value hides.

Zero serves as the boundary between a win and a loss. Beating an FBS team ranked below 100 as the favorite by one score at home versus beating a top-10 team as the favorite at home obviously hold very different values. For the formula, the difference of a win against the 5th ranked team versus the 108th is 0.60, nearly eight times more valuable. I placed a 0.20 increase between a win over a team in the 11-to-25 range and a win over a top-10 team as I find those wins to be the most impactful for a résumé.

As you can see, a value has been placed on being the underdog and being the favorite. This is done, not because I’m a gambler, but because oddsmakers have infinite data in front of them that determine not if a team is better than another, but if in that specific game, would team X beat team Y at home, on the road, or on a neutral field.

I specifically used Bill Connelly’s SP+ rankings for head to head matchups as he value teams as how much better they would be against the average or “zero” college football team. For reference, Georgia ranks first in his rankings with a 36.7 value, saying that they would be 36.7 points better than the average college football team while UMass ranks last at -30.1, saying they would lose to the average college team by that many points.

I recognize that a win or loss in week one over, say, a top 25 team can have its value change by week seven both positively or negatively. I will change the team’s value if the team they won or lost against drops more than 30 spots or into a different tier. However, I won’t make that change if a team goes from ranked 25 to 26 and subsequently switches tiers. Clearly that team’s value has barely changed, and they moved down usually because a team below them ascended up the ranking.

I also placed a value on conference. I used strength of conference based on FPI and created multipliers. A win against an SEC team was more valuable than any other conference in 2022, followed closely by the PAC-12 and then the Big 12, Big Ten, and ACC to round out the Power-Five conferences.

The AAC was the most valuable Group-of-Five conference in 2022, narrowly followed by the Sun Belt. Conference-USA, the Mountain West, and the MAC were very distant, and beating a team in those conferences were less valuable.

End lesson: It pays to win, it pays extra to win as an underdog, on the road, or to just bury your opponent.

What is the value of a loss?

The value of a loss is a net negative under every circumstance. College football, more than any other team sport in the world, holds such value in winning and perfection, and losing is imperfect and should always be treated as such.

This is where I think this formula works at its best. A loss is not an arbitrary value. It is definite and permanent, and it should be seen as substantially less valuable than a win. Let’s use the Alabama versus Tennessee game this season. Tennessee’s upset win value was +1.063625 while Alabama losing as the road favorite has their loss value at -0.054375.

Contrary to what the CFP Committee wants us to think, that loss holds a substantial value loss compared to the value of Tennessee’s win. Big wins matter a lot, losses matter no matter what.

End lesson: Don’t lose. Losing is bad.

How I decide my Top 25

My top 25 takes four weeks into the season to determine. This is because it allows me to gather enough data points to put together a coherent ranking.

My top 25 in week six went 1) Alabama, 2) Ohio State, 3) Georgia, 4) Michigan, 5) Tennessee, 6) USC, 7) TCU, 8) Kansas State, 9) Clemson, 10) Penn State, 11) Ole Miss, 12) Oklahoma State, 13) UCLA, 14) Texas, 15) Mississippi State, 16) Kansas, 17) Oregon, 18) Syracuse, 19) Wake Forest, 20) Purdue, 21) NC State, 22) James Madison, 23) Cincinnati, 24) Utah, 25) Minnesota.

The top 25 is determined on a team’s average weekly value. Let’s look at two examples. We’ll use Tennessee and Alabama. Tennessee’s average weekly value ended up at 0.461065 while Alabama’s ended up at 0.406542. On average, Tennessee was roughly 0.060 points better than Alabama this season, a close value on the surface, but I’ll give some more examples to show that it’s really not.

The difference between 18th ranked Oregon State and 32nd ranked Illinois was about 0.065.

Georgia’s end value wound up being roughly 0.060 points higher than second place Michigan and 0.10 higher than Tennessee. Georgia was comfortably the number one team in the country.

This is my final top 25 prior to bowl season:

The biggest discrepancies here versus the final CFP ranking are:

• Alabama being lower than both Tennessee and Kansas State
• I have Clemson at 11 versus 7 in CFP
• I have Penn State at 8 versus 11 in CFP
• I have NC State at 43 versus 23 in CFP
• Ole Miss at 24 versus unranked in CFP
• I have LSU at 13 versus 17 in CFP and higher than Tulane
• I have Oregon over Oregon State
• I have Purdue at 25 versus unranked in CFP
• I have UTSA at 27 versus 25 in CFP
• Final thoughts

All in all, I created this formula because I love numbers and college football. It was merely a fun exercise to see if I could create something that would accurately represent what I value in this great sport. Let me know if you’ve done something similar, what your rankings are, your thoughts on the formula, and whether you agree or disagree with the final rankings!