On this pageT24 and Y powerThe score in one pictureThe four stepsRoundsWhen rules disagreeEarned and projectedWhat it leaves out
An objective way to rank who deserves a playoff spot
The Yardstick produces two things: the T24 ranking, and the Y power rating underneath it. Both are transparent and traceable. Every number can be followed back to the games.
T24
The T24 is an objective measure of which teams deserve to be in the playoff. It takes the logic of the playoff committee and turns it into written rules that are transparent and can be defended. It counts two things: how strong a team is, like other power ratings, and what the team has earned on the field.
- What it is
- A ranking, like the AP, Coaches or CFP rankings
- Question it answers
- Who has earned the higher spot?
- What it uses
- Y power, plus wins and losses, strength of schedule, and head-to-head results
- Cares who won?
- Yes. Every game adds to or takes from a team's score
- The name
- It is called the T24 because it is built to get the top 24 right.
Y power
Y power is the power rating underneath the T24. It measures how strong a team is, and it sets the strength side of the ranking. At picking winners, it is nearly as accurate as the Vegas line.
- What it is
- A raw power rating
- Question it answers
- How good is this team?
- What it uses
- Yards per play and points, for and against, adjusted for opponents
- Cares who won?
- No
- How accurate
- It picked 69.6% of winners over 1,814 games. The Vegas closing line picked 70.5%. See the test
The score in one picture
Two of the four steps add points. Each one asks its question twice: first about the team on its own, then about its schedule.
Example: #1 Georgia (9-1). Every number is in yards per play. The small # is the team's rank among all 138 teams on that part alone. Every team page shows this same grid.
Step 1 measures how well the team played. Step 3 adds what it earned. The two together are the score. Steps 2 and 4 add no points. Step 2 decides which teams can be compared. Step 4 settles games between teams that played each other.
The four steps
Each step below starts with the short version. Open a step to see every rule behind it.
Every team goes through the same four steps, in the same order, every week.
Asks: how good is this team?
Yards per play and points, for and against, adjusted for the opponents it faced.
Step 2 · Record and scheduleAsks: who is it fair to compare it with?
A team cannot pass a team with two or more fewer losses, or one with a much tougher schedule.
Step 3 · Wins and lossesAsks: what has it done?
Points for every win, more for beating a better team. A cost for every loss: small for losing to a top team, large for losing to a weak one.
New in Version 2 An extra charge for an upset loss, and a schedule adjustment: how the team did against the difficulty of its schedule.
Step 4 · Head-to-headAsks: did they play each other?
When two teams ranked close together have played, the winner goes above the loser.
Step 1 · Power rating: how good a team is
How good is this team, play by play?
We look at two things: yards per play (gained minus allowed) and points (scored minus allowed). Both are adjusted for who the team played, so the same margin against a better opponent counts for more. Wins and losses are not used in this step.
Example: Georgia (9-1), #1 this week.
| Per gameits own margin, whoever it played | Schedulehow good its opponents are | Power | |||
|---|---|---|---|---|---|
| Yards per playgained 7.01, allowed 4.57 | +2.44 | + | +0.86 | = | +3.30 |
| Pointsscored 39.4, allowed 16.0: +23.4 a game | +2.23+23.4 points ÷ 10.5 | + | +1.34 | = | +3.57 |
| Power rating | +4.67game performance | + | +2.20opponent strength | = | 6.87#1 of 138 |
Points are put on the same scale as yards, with one fixed rate: 10.5 points a game = 1 yard per play.
Show the details of Step 1
The Yardstick’s power rating is called Y power. At picking winners, it is nearly as accurate as the Vegas line: 69.6% to 70.5% over 1,814 games. See the test
Y power mixes performance and outcomes, each adjusted for the opponents a team played:
- Performance: yards per play gained minus yards per play allowed.
- Outcomes: points scored minus points allowed.
- Both are adjusted for schedule the same way. A team's rating is its own average margin plus the average rating of the teams it played.
- Those opponents are rated the same way, from their own games, so the math is repeated until no rating changes.
- A hard schedule gets full credit, and a soft one cannot hide.
- Each rating is worked out twice, once with full margins and once with blowouts trimmed, and the two are averaged.
- Then yards and points are put on one scale and added with equal weight.
Why start here. A power rating is the best single measure of how good a team is. But on its own it ignores whether a team actually won. Steps 2 to 4 add that back.
The Power ratings page, under Rankings, shows the numbers for every team. Each team's page shows them game by game.
Step 2 · Record and schedule: who gets compared
Who is it fair to compare this team with?
Teams are only compared with teams that have a similar record and a similar schedule. A team cannot pass a team with two or more fewer losses, unless its own schedule was much tougher.
Example. Georgia (9-1) can be compared with the teams ranked #1 to #11. Inside that range, the score decides the order. Outside it, the record or the schedule is too different.
Who is an 8-1 team compared with?
- 7-2Yes. One more loss is close enough.The team with the higher score goes ahead.
- 6-3No. Two more losses is too far apart.The 8-1 team stays ahead, even if the 6-3 team has the higher score.
- 6-3Unless the 6-3 team played a much tougher schedule.Much tougher means three or more schedule groups. Then the two are compared, and the higher score goes ahead.
Who is a 9-0 team compared with?
- 8-1Yes. One loss apart, as always.The team with the higher score goes ahead.
- 7-2Yes, if the 9-0 team played a schedule outside the top 20.An unbeaten record against a top-20 schedule is protected from 2-loss teams.
- 6-3No. Three more losses is too far apart.Only a much tougher schedule changes that.
Show the details of Step 2
A power rating alone is not enough. By power rating, a 6-3 team can rank ahead of an unbeaten one. Wins and losses have to count.
So before teams are compared, they are grouped with teams that have a similar number of losses. Then the groups are checked against how tough each schedule was, because 8-1 against a hard schedule is not the same as 8-1 against a soft one. Only teams that are close on both trade places.
The four rules behind those answers
- The loss rule. A team cannot pass a team with two or more fewer losses. One loss apart is close enough to compare. Example A 7-2 team can pass an 8-1 team. A 6-3 team cannot. It loosens lower down. From 4 losses on, teams within two losses are compared, because that deep in the rankings a record depends more on who a team played. One exception: a team that beat the other team on the field, and played a schedule 40 or more spots tougher, is not held below it. Example A 3-6 team can pass a 5-4 team.
- The unbeaten exception. Most teams are compared with teams one loss better and one loss worse. An unbeaten team has nobody above it to be compared with. So unless its schedule is in the toughest group, it is also compared with 2-loss teams. Example A 7-2 team can pass a 9-0 team whose schedule ranks 49th. It cannot pass a 9-0 team whose schedule ranks 12th.
- The schedule rule. Teams are sorted by how strong their opponents are, in groups of 20. The 20 toughest schedules are group 1, the next 20 are group 2, and so on. A team whose schedule is three or more groups weaker cannot pass a team with the tougher schedule. Example A team with the 70th-toughest schedule (group 4) cannot pass a team with the 15th-toughest (group 1), even with the same record.
- The record rule. If a team beat another team and has fewer losses, it always stays ahead of it. Example An 8-1 team that beat a 7-2 team stays ahead of it, whatever their scores.
If the loss rule and the schedule rule point in opposite directions, they cancel, and the two teams are compared on the field, by score. Example A 6-3 team with a group 1 schedule meets an 8-1 team with a group 4 schedule. Losses favor one and schedule favors the other, so the higher score goes ahead.
What is the score? It starts as the power rating from Step 1. Step 3 adds points for wins and takes points away for losses. “Can pass” always means: goes ahead if its score is higher.
The chart on every team page
Every team page shows these rules as one chart. Rows are the number of losses. Columns are the schedule groups. The green boxes hold the teams this team is compared with. Teams in the grey boxes stay ahead or behind, whatever the score.
Example: Georgia (9-1), #1 this week.
Compared with #1–#11 (14 teams in all). Inside that range, the score decides the order.
- Above: no limit. It can go as high as #1.
- Below: it stops at #12 Kentucky (7-3): Georgia has 2 fewer losses.
Georgia: 1 loss, schedule group 2 (schedule rank (this round) #22)
Each box is a set of teams with the same number of losses and the same schedule group. The number is how many teams are in it. green: Georgia is compared with these teams, and the score decides the order. light green: with some of them. grey: not compared. Teams there stay ahead or behind whatever the score, because their record or schedule is too different. The outlined box is Georgia's own.
The two rules that overlap: the record rule (rows) compares teams within one loss of each other, or within two from 4 losses on. The schedule rule (columns) stops a team from passing one whose schedule is 3 or more groups tougher. An unbeaten team outside group 1 can also be passed by 2-loss teams.
Why. Record is the most basic thing a team earns, so it is the hardest line. But a record means more against a hard schedule, so schedule can offset it.
The schedule is the full regular season. Every opponent counts, played or still to come, each rated by how good it is today. No game is predicted. This keeps a team from dropping just because it played its weakest opponent this week.
Step 3 · Wins and losses: what a team has done
What has this team earned?
Every win adds points. Beating a better team adds more. Every loss costs points. Losing to a weaker team costs more. New in Version 2 Then one more check: how did the team do against its schedule? We compare its record with what a #10 team would do in the same games.
Example: Georgia (9-1), #1 this week.
Show the details of Step 3
Step 3 turns results into points and adds them to the power rating. It has two sides.
- The win-loss side: what the team earned. Every win adds points and every loss takes some away, from one scorecard. A big upset loss costs extra. New in Version 2
- The schedule side: how it did against the difficulty of its schedule. Its record next to what another team would do in the same games. New in Version 2
Example: Georgia (9-1), #1 this week.
| Starts from the power rating (step 1) | 6.87 | |
| Win-loss adjustment What it earned: a win earns more the better the team beaten; a loss costs more the weaker the team that won. | wins +2.13losses −0.14should-win loss penalty −0.64 game by game ↓ | +1.35 |
| Schedule adjustment How it did against the difficulty of its schedule: its record next to what a #10 team would do in the same games. | game by game: earned in wins, given up in losses +0.32whole record: how often a #10 team does this well +0.54 how it is built ↓ | +0.86 |
| Step 3 moves the score from 6.87 to 9.07 (+2.20). By score alone that is #1; by power rating alone it was #1. | 9.07 | |
The win-loss side
The scorecard. Every game is scored from one card. Look up the rank of the opponent, and read across. In Version 2 the card's amounts count ×1.35 in the score, which puts them in yards per play like the power rating.
| Opponent's rank now | How often a top-10 team wins this game | A win earns | A loss costs |
|---|---|---|---|
| #1 to #15 | about half the time (25-27) | +0.30 to +0.25 | −0.05 to −0.09 |
| #16 to #25 | about 6 in 10 (19-13) | +0.24 to +0.20 | −0.10 to −0.12 |
| #26 to #45 | about 7 in 10 (33-14) | +0.20 to +0.11 | −0.13 to −0.24 |
| #46 to #60 | about 8 in 10 (25-6) | +0.11 to +0.05 | −0.24 to −0.35 |
| #61 and lower | almost always (59-3) | +0.05 | −0.35 to −0.60 |
| FCS team | almost always | nothing | −0.70 |
The middle column is the real record of top-10 teams against opponents at each rank, 2024 to 2026, from week 5 on. Inside a row, the amount slides evenly from one end to the other.
Three things adjust each game:
- Where it was played. A road win earns 15% more and a home win 15% less, on the part above the 0.05 every win gets. A road loss costs 15% less and a home loss 15% more.
- How close a loss was. A loss by 3 or fewer costs 25% less. A loss by 28 or more costs 25% more.
- Conference title games. New in Version 2 A lost title game costs points like any other loss. Its one protection is in Step 2: it does not change who the team is compared with, because the loss rule does not count it. Between the two teams that played the game, the real records are used.
The big-loss charge: the should-win loss penalty. New in Version 2 The card charges a loss by the rank of the team that won. Version 2 adds one more charge, for an upset. When a team loses a game it had a 65% chance or more to win, it pays extra, up to 0.60 on the card's scale (0.81 in the score). The chance comes from the two teams' own power ratings and where the game was played. The bigger the upset, the bigger the charge.
Example: Georgia lost at home to Auburn, a game it had a 93% chance to win. In the score, the card charges 0.20 and the penalty adds 0.64.
The schedule side New in Version 2
The card scores each result by itself. The schedule adjustment adds context. Picture a #10 team playing the same games, in the same stadiums. We ask two questions about it and score each from −2 to +2.
- 1 · Game by game: would a #10 team have lost more than this? A win earns the #10 team's chance of losing that game, but only when it was a real test: full credit when the #10 team wins 65% of the time or less, none at 80% or more, sliding in between. A loss gives up the #10 team's chance of winning that game.
- 2 · The whole record: could a #10 team have won this much? How often would a #10 team finish with this record or better? Half the time scores 0. About 1 time in 6 or rarer scores +2. About 5 times in 6 or more often scores −2.
The factor. Half of each score counts, so the two questions weigh the same. The sum is then multiplied by 0.54. The most any team can gain or lose is 1.08. Like the power rating, that is in yards per play.
| Georgia (8-1) | What happened | Score (−2 to +2) | Half | × 0.54 |
|---|---|---|---|---|
| 1 · Game by game | Earned 1.65 in hard wins (at Alabama, at Ole Miss); gave up 0.79 losing to Auburn | +0.85 | +0.43 | +0.23 |
| 2 · The whole record | A #10 team goes 8-1 or better on this schedule 22% of the time | +1.56 | +0.78 | +0.42 |
| Schedule adjustment | +0.65 | |||
Why 0.54. It is 0.4 × 1.35. The 1.35 turns the result into yards per play (1.35 yards per play is the typical gap between two teams). The 0.4 was set once, from the last five seasons, so that by the end of the regular season the differences between top-25 teams' scores come about half from the power rating, a quarter from wins and losses, and a quarter from the schedule adjustment. It does not change from week to week.
Why the card is shaped this way. It follows real results. The harder a game is to win, the more a win earns. The more surprising a loss, the more it costs. Beating a top-15 team is a coin flip even for a top-10 team, so it earns the most. Losing to a team outside the top 60 almost never happens to a top-10 team, so it costs the most. Every win over an FBS team earns at least 0.05, because winning is never automatic.
For scale, the #1 and #10 teams are usually less than one point apart. So one major win moves a team a few spots, not a leap.
Step 4 · Head-to-head: the game between them
Did they play each other?
When two teams ranked close together have played, the winner goes ahead of the loser.
Example. Two teams have the same record and are ranked #6 and #8. The #8 team beat the #6 team in October. The #8 team moves ahead.
Show the details of Step 4
If a team beat another team ranked just above it, the winner goes above the loser. How far that reaches depends on the records:
- Same number of losses, or fewer: the winner moves ahead if the two are within 10 spots.
- One more loss: the winner moves ahead only if the two are within 4 spots and the winner also played the tougher schedule.
- Two or more extra losses: never, with one exception. If the winner’s schedule is 40 or more spots tougher, the loss rule does not hold it below the team it beat.
Where the spots are counted.
- The first time through, the gap between the two teams is counted in the order that comes out of Step 3, so every win and loss is already in.
- In later rounds it is counted from where the teams finished the round before.
- That means a move that was made holds.
- It also means a team can climb again if it beat another team close to its new spot.
Either the winner moves up or the loser drops down, whichever jumps fewer teams that should not be jumped. If results form a loop (A beat B, B beat C, C beat A), no move is made.
A chain of results lands at the group's average score. New in Version 2
Sometimes three or more teams beat each other in a chain: Texas beat Tennessee, and Tennessee beat Alabama. Under the rule above, the team with the best score would pull the whole chain up to its spot. In Version 2 the group stays together, in head-to-head order, but it is placed at the average of the group's scores. A team that scores above that average moves ahead of the group.
One exception (added Nov. 17, 2026). A team at the top of the group whose own score is higher than the team moving up stays above it. The group order never breaks. This week Notre Dame (7.13) scores above the Texas-Tennessee-Alabama average (7.12), so it moves ahead of the group. But Texas's own score is 7.58, higher than Notre Dame's, so Texas stays above Notre Dame. Tennessee's own score (5.91) is lower, so Notre Dame goes above Tennessee, and Alabama stays behind Tennessee, which beat it.
Example: three teams in a chain score 4.8, 3.9 and 5.7. Their average is 4.8. A fourth team that scores 5.0 is placed ahead of all three.
The playoff has the last word. New in Version 2
Once playoff games are played, a team that went out in an earlier round stays below every team that went further. The champion is #1 and the runner-up is #2. A playoff loss also does not change who a team is compared with, the same as a lost conference title game.
Why. A game between two teams is direct proof, so it outranks general measures like stats and schedule. But one game should not erase two extra losses, so it does not break the loss rule, unless the winner also played a far tougher schedule. And a team with a worse record gets less reach than a team with an equal or better one.
How the score adds up New in Version 2
After the four steps, a team's score has four parts. Read down each column: first what the team did by itself, then what its schedule adds. In step 1 the schedule part is how good the opponents were. In step 3 it is how the team did against them.
Example: Georgia (9-1), #1 this week.
| Step 1 summary How well it played the power rating: every play and point, win or lose | Step 3 summary What it earned its results, game by game | |
|---|---|---|
| On its own what the team did, whoever it played |
Game performance #2 of 138 +4.67 its own margins, whoever it played: yards per play +2.44 · points +2.23 (on the yards scale) |
Wins and losses #5 of 138 +1.35 win-loss adjustment: points for each win, a cost for each loss, extra for an upset loss |
| What the schedule adds a different team plays the same schedule |
Opponent strength #18 of 138 +2.20 how good the teams it played are: the same margin against better teams is worth more |
How it did against that schedule #1 of 138 +0.86 schedule adjustment: its record next to what a #10 team would do in the same games |
| Step total | Power rating 6.87 #1 = +4.67 game performance +2.20 opponent strength | Step 3 adds +2.20 = +1.35 wins and losses +0.86 against the schedule |
| Score step 1 + step 3 | 6.87 + 2.20 = 9.07 #1 by score | |
Read down a column: first the team alone, then what its schedule adds. In step 1 the schedule part is how good were your opponents. In step 3 it is how did you do against them. Every number here is in yards per play. Step 2 (record and schedule) and step 4 (head-to-head) add no points. Step 2 decides who can be compared. Step 4 settles games between teams, and can place a head-to-head group at its average score.
The fine print: rounds, and what happens when rules disagree
Rounds: letting it settle
A win is only as good as the team you beat turns out to be. So Steps 3 and 4 repeat eight times, in three phases:
- Rounds 1–3, explore. Everything is re-measured each round.
- Rounds 4–6, narrow. Schedule groups are set once, from the average of the first three rounds, and locked. A head-to-head call that comes out the same way twice is locked.
- Rounds 7–8, confirm. Nothing new is decided. The rounds only check that the order holds.
Each game is judged against the opponent's average spot over the rounds so far, so every round changes less than the one before.
Example: Georgia (9-1), #1 this week.
Higher on the chart is a better rank. Power is where the power rating alone puts Georgia (#1). Step 2 applies the record and schedule rules. Each round then re-scores every game against the newest ranking, re-applies the rules and head-to-head, and repeats until the order stops moving. Settled: extra rounds do not move it.
When rules disagree
The strongest rule is first.
- The loss rule is the strongest.
- The record rule: beat a team and have fewer losses, and you stay ahead of it.
- Head-to-head. It can overrule the schedule rule and the scores, never the loss rule.
- The schedule rule.
- The score (Y power plus wins and losses) decides everything else.
Two views of the ranking: earned, and "if the favorites win"
Earned, and "if the favorites win"
The T24 has two views.
Earned is what teams have done so far: only games already played. One part looks ahead: the schedule rule in Step 2 uses the full regular-season schedule, with opponents still to come rated as they stand today. No game is predicted.
Projected shows where the season is heading.
If the favorites win gives every game still to play to the favorite (the team with the higher Y power, plus about 4 points for playing at home) and runs the T24 once on that finished season. Each team also gets a chance of that record: how likely it is to finish with the projected record, by any route. Results closer than 60/40 are marked as toss-ups.
How to read it. It is one scenario, not a prediction. Tested on 2024 and 2025, it was no closer to the final ranking than the earned ranking was. Its use is to show what each team still has in front of it. When a favorite loses, that is the reason teams move. Conference title games are projected too. Each conference’s own rules pick its two teams from the projected standings, and the favorite wins. See the ten projected title games
What it leaves out on purpose
- No preseason rankings and no polls. Every team starts from what it has done this season.
- No conference bonus and no brand names. A conference only matters through the teams its members actually played.
- No extra credit for a conference title beyond the win itself.
- No eye test. Close calls are shown openly as "judgment calls" instead.
To see these steps for any one team, pick it on the Team tab.