How it works
Every team starts at 1500 points. When two teams play, the winner takes points from the loser. How many depends on how surprising the result was and how big the margin was. Try it, then read on for every rule.
Try it
Where
Who won
Team A
50% to win going in
+71.2
1500 → 1571
Team B
50% to win going in
−71.2
1500 → 1429
120 points at stake × 1.19 for a 7-point margin × 0.50 surprise = 71.2 points. Team A had a 50% chance, so 50% of the win was a surprise.
Pick two teams (this season's ratings, or your own), where they played, who won and by how much. The math is the same the rankings use, and the rules below explain each part of it.
How the scoring works
Rule 1
Everyone starts even
Every FBS team opens every season at 1500. No preseason poll, no recruiting rankings, no memory of last year. A team's points come only from this season's games.
Last year's champion, week 1.
same as everyone
Last year's winless team, week 1.
same as everyone
Rule 2
Before kickoff, every game has a win chance
The bigger the gap in points, the more likely the better team wins. Home field adds 55 points to the home team's side of the gap, for this calculation only.
| Points gap | Better team wins |
|---|---|
| Even | 50% |
| 55 (same as home field) | 58% |
| 100 | 64% |
| 150 | 70% |
| 200 | 76% |
| 300 | 85% |
| 400 | 91% |
The formula is the standard Elo curve: chance = 1 ÷ (1 + 10−gap ÷ 400). A 400-point gap is 10-to-1.
Rule 3
Surprises move the most points
80 points are at stake (more in a team's first games, rule 7), times the margin factor (rule 4). The winner gets the part of that it “wasn't expected” to win: 80 × (1 − its chance).
A 1650 team beats a 1500 team by 10. It had a 70% chance.
expected, so a small reward
Same game, but the 1500 team wins by 10. It had a 30% chance.
the upset is worth more than twice as much
Two even 1500 teams on a neutral field, one wins by 10.
a coin flip: half the stake
Same even teams, 10-point win, at home vs. on the road.
- Won at home
- +44.3
- Won on the road
- +60.8
winning on the road is the bigger surprise
Rule 4
Margin counts, up to 17
A loss is always a loss on the record. Margin only changes how many points move: from 0.5× for a 1-point game to 1.5× at 17 or more. Nothing extra past 17, so there's no reward for running up the score.
| Won by | Margin factor | Points (even teams) |
|---|---|---|
| 1 pt | 0.50× | ±20.0 |
| 3 pts | 0.89× | ±35.5 |
| 7 pts | 1.19× | ±47.5 |
| 10 pts | 1.31× | ±52.5 |
| 14 pts | 1.43× | ±57.3 |
| 17 pts | 1.50× | ±60.0 |
| 30 pts | 1.50× | ±60.0 |
| 50 pts | 1.50× | ±60.0 |
So a 1-point loss costs 20.0 and an equal blowout costs 60.0: a third as much. The curve rises fast early (a field goal to a touchdown matters a lot) and flattens out.
Rule 5
All FCS teams are one team
Teams outside FBS play too few FBS games to rate, so they all count as one team called “FCS” rated 1150. That rating never changes. Games between two FCS teams are ignored.
An average FBS team beats an FCS team by 30 at home.
barely anything
Same game, but the FCS team wins by 3.
a disaster
A 1750 team beats an FCS team by 40.
almost exactly nothing: scheduling an FCS team can't pad a rating
Rule 6
Points are traded, not created
Whatever the winner gains, the loser loses, exactly. So a team's points are always 1500 plus what it won and lost in its own games, and nothing else.
The one leak is FCS games: the FCS side never gives or takes points, and FBS teams win most of them, so the FBS average ends a few points above 1500. That's left alone. Correcting it would mean nudging every team, and then a team's points would no longer add up from its own games.
Rule 7
Early games count more
In week 1 every team is at 1500, so its rating says nothing yet. For a team's first 3 games, 120 points are at stake instead of 80, so the ratings sort themselves out fast. It's the same idea chess uses for new players.
Two even teams, both in their first games. One wins by 10.
at stake: 120
Same game later in the season.
at stake: 80
One team is in its first games, the other isn't.
at stake: 100, the average, so what one gains the other still loses
80 and 120 weren't guessed. Every game from 2014 to 2022 was replayed with dozens of settings, scoring how well each one's pre-game win chances matched what happened, and these predicted best. They also predicted best on 2023 to 2025, which the test never saw.
Rule 8
Each week is frozen
Points are added one game at a time, in the order the games were played. A later result never reaches back and changes an earlier week, so the history shows what the rankings said at the time.
New rankings come out Sunday at 8 AM Mountain, with a recheck Tuesday for late scores and stat corrections.
Questions people ask
- Why is an undefeated team below a team with a loss?
- Wins aren't equal. Beating a 1350 team by 3 at home earns +16.7; beating a 1650 team by 14 on the road earns +87.6. A team that keeps squeaking past weak opponents can trail a team that lost once but beat good teams soundly.
- My team had a bye. Why did it move?
- They didn't. Points only change when a team plays. Its rank can still move, because the teams around it played.
- Why did a 50-point win barely move my team?
- Margin stops counting at 17 (rule 4), and if your team was a big favourite it had little to gain anyway (rule 3).
- Can I check my team's number myself?
- Yes. Start at 1500 and add up the Points column of the game log on the team's page. That's its rating, exactly. The week-to-week change is the same: just that week's games.
- Does it know about injuries, quarterbacks or weather?
- No. It only knows who played, where, who won and by how much. That's the point: it can't be talked into anything.
- Why does everyone start at 1500 instead of where they finished last year?
- Rosters turn over every year. Starting even means a ranking is earned this season, with no head start from reputation. The cost is a noisy September, which is what the bigger early stakes (rule 7) are for.