Your team
The rankings
Select a team to see every result it has posted, with each opponent’s power rank.
| # | Team | Move | W–L–T | Pts | Power | Rating vs. average | SOS |
|---|
Where the model disagrees with the table
Teams whose power rank is furthest from where the standings put them. This is the strength-of-schedule effect made visible.
Next up
Projected margins from the ratings, posted on the half goal. No home-field adjustment — see the method note.
How this is calculated Method & caveats
Two ratings, blended
Every team gets two independent ratings, then they are combined into the Power score on a 0–100 scale where 50 is an average team in that division.
Colley looks only at wins and losses. Every team starts at .500, and a team’s rating is pulled toward the average rating of the opponents it has faced. Beat a good team and your rating rises more than it would for beating a bad one.
Massey looks at goal margin, so a 4–0 win counts for more than a 1–0 win over the same opponent. It is solved as a single system across the whole division, which means a margin is judged against the opponent who conceded it.
Blowouts are capped, because the league caps them
This league records a maximum of four goals per team per game, and caps each game’s differential at four. Those caps are applied here too, so beating a weak team 9–0 is worth exactly the same as beating them 4–0. Running up the score earns nothing.
Strength of schedule
SOS is the average power score of the opponents a team has actually played. A team with SOS rank 1 has faced the hardest slate in its division so far. Early in a season this varies a lot, which is most of the reason the power order and the table order differ.
What the numbers mean
Rating is on a goals scale: it is the margin a team would be expected to win by against an average team in its division. The gap between two ratings is a projected margin, which is where the “Next up” calls come from.
Those calls are posted on the half goal, the way a betting line is: a projection of 1.28 goes up as 1.5. Tenths would claim precision the ratings do not have on a sample this size. A game that rounds to zero is shown as too close to call.
The crests
Every team gets its own crest, so a division of ten arrives as ten identities rather than ten grey rows. Each one is a kit — two or three colors in a pattern, with the team’s letters across the band — drawn as SVG in the browser. No badge artwork is used anywhere on the page.
Read it with the sample size in mind
There is no home-field adjustment. These are club fields with short travel, and a sample this small could not separate a real home effect from noise. The model also has no knowledge of who was missing, who is playing up, or what the weather did.