The Math
CFB Power Digits is a completely unbiased mathematical way of ranking teams in College Football.
The primary objective of this ranking is to not necessarily predict future winners, but to determine who has the best resume for the College Football Playoffs.
The Rankings were built to give a non-biased approach to teams’ strengths. By defining the path to success before the season, no name-bias can help a team’s ranking. The Power Digits methodology may not be universal, but ensures consistency from year-to-year.
There is no *right* way to rank college football teams. This is simply one way to do it, which will sometimes agree with the Pollsters and the CFP Committee, but often times not.
Every FBS 1-A team is given a Power Digits Rating, a number between 100 and 0, which determines their rank. 100 is the best possible value a team can have, 0 being the worst.
The algorithm is broken up into five different components. These are Team W-L Rating, Strength of Wins, Strength of Losses, Strength of Schedule, and Top 3 Wins.
The base of the algorithm is in the Team W-L Rating, which makes up 55% of the final score. This weighted win percentage comes from giving more (or less) than 1 into a win/loss column. Based on conference and location of the opponent, these values change coming from the strength of the conference the opponent belongs to.
Conference strengths come from the win percentage of the conference over the last three & current seasons (for 2024, consider 2021-2024) in all FBS-FBS games (including bowls), removing in-conference and FCS games. The base value is the win percentage plus 0.5.
This number becomes the value of a neutral site win against a team from that conference. If the win was on the road, 0.03 points are added, if it was at home, 0.03 points are subtracted. (It is harder to win on the road in CFB!)
For losses, the conference strength is 1 divided by the base win value. Similar to the wins, 0.03 is value added or removed based on the location of the game.
An approximation of the conference strengths, which are dynamic with more game results
Another 25% of the Power Digits score comes from the same weighted wins/losses formula, except based on the results of all opponents (in games not vs the team in question). Teams are benefitted when their previous opponents win games.
Note: Opponent strength is not counted until the teams play each other. There may be a known late season rivalry vs a juggernaut, but they aren’t part of the resume just yet.
5% comes from the strength of wins, which is the combined win-loss rating of the teams that were beaten. If the team is winless, they receive a 0.
5% come from the combined win-loss rating of the teams lost to. If a team is undefeated, they receive a 1.
The last 10% comes from the average win-loss ratings of the 3 best teams beaten. This is meant to reward those who have played multiple strong opponents, and not to be carried by playing (and losing to) tough opponents.
These ratings are not meant to say “Team X is better than Team Y,” they say “Team X deserves a playoff spot before Team Y does.”
In summary, the best outcome for a team is to win games (of course), and then hope your opponents from the season also go on to win games.
In early weeks the algorithm has minimal results to compare teams on, but by the end of the season there are 12 team results and approximately 130 opponent results to gauge how strong a team really is.
FAQ
Why do the rankings look so "wrong" for the first few weeks?
The rankings benefit from more games, as it only learns from game results. During week 1, it may only know that Maryland is better than Eastern Michigan, but by week 2, 3, 4, when Maryland has beaten/lost to other opponents, it continues to factor in the success of Eastern Michigan. As Eastern Michigan wins games, Maryland receives bumps to their ratings while their network of opponents grows. By week 12, Maryland has 10 opponents each with their own 10 wins and losses.
If we know Alabama is better than Akron, why do they startwith no rank advantage?
We as fans may think we know that Alabama will be better than Akron, but it must be proven on-field. The rankings are based on only the data from the season in question, so there are no pre-conceived biases.
College football team success is volatile, Michigan State went 12-2 with a playoff appearance in 2015, followed up with a 3-9 record in 2016. Not to mention the ever-changing landscape of transfers & coaching.
For reasons like this, the Power Digits gives no credit or punishment to teams based on previous seasons. The start of every year is a blank slate. Do we really know Alabama is better than Akron, or are we just assuming?
How are there conference strengths when conferences have changed so much?
Conference strengths are built by considering the wins and losses in out-of-conference FBS vs FBS games over the last 3 years, of the teams currently in the conference. Texas’ results count for the SEC, Cincinnati’s results count for the Big 12, etc.
Do teams who go undefeated against easy opponents get ranked higher because of win percentage?
No, because the Power Digits Rating holds Strength of Schedule in high regard, and an even higher weight to the strength of the teams beaten. Teams will benefit from being undefeated, yes, but 45% of their ranking will be hampered by them having weak opponents.
Do Group of 5 teams benefit from the Power Digits?
No, they don't. They actually have it harder, as their conference games hurt their win percentages. A win against another Group of 5 is worth less than 1 win, and a loss is worth more than 1 loss. If they are undefeated, they can float around the CFP Top 4, but after a loss they very rarely crack the top 10. Because their strength of schedule is usually very low, they need to have an almost flawless season otherwise to crack the Top 10.
How accurate are the rankings to the real life rankings?
This is an independent ranking and should be seen as something different than the AP Poll and the CFP Committee's rankings. There is no 100% accurate way to rank these teams, unless they all played each other in a season. The Power Digits are a bias-free alternate way to rank teams.
Is scheduling the worst Power 5 teams is the best way to defeat the algorithm?
To a point, yes. A win on the road against Vanderbilt credits you the same as a win on the road at Alabama. For your team win/loss rating, yes, it helps you. However, the difference between Vanderbilt and Alabama wins are made up for in the Strength of Wins, and Strength of Schedule categories, where Alabama’s success will provide more of a boost, and Vanderbilt’s struggles will bring your rating down.
Why do undefeated teams get such a boost?
They’re undefeated. If nobody that they played were able to beat them, then they should be rewarded. Until a team loses, we can only guess as to how good a team truly is.
How are FBS Independent teams classified?
They are classified as the conference they are most closely represented by, or their previous conference. Notre Dame is classified as the ACC, UMass to the MAC, and UConn to the American. Their Wins/Losses do not count toward the conference strength.
Two teams are ranked next to each other who played each other. Why is the loser ranked higher?
Across the sum-of-parts of the rankings, these two teams happened to fall next to each other. We are ranking resumes, and not who-beat-who. Theoretically, we can end with a never-ending loop of teams needing to swap to have accurate results, so this “swapping” is never considered.