2024 MLS Power Rankings averages, week 12

14 sites published for week 12;, Toronto fall, NYC climb.

by Kreg AsayMay 16, 2024

14 sites published power rankings for week 12, 3rd Degree has missed 6 updates, they will be dropped next week if they don’t update. Goal.com has been retroactively added, they have apparently started bi-weekly rankings as of Wk10.

Week 12 rankings DO partially include midweek games played May 15th, many of the analytic sites (including betting odds) updated on Thursday with those results so their new rankings have been used.

A reminder of what Power Rankings should be (not that every site does, of course):

"If two teams played tomorrow, given their current injury concerns, form, and other considerations, who is likely to win on a neutral field? All things being equal, who is the best team going forward, based on what we know now?"

— Richard Farley, NBC Sports ProSoccerTalk. March 11, 2014.

You may notice the inclusion of 'analytic' sites; they serve to function as a 'control group' to others which are based on opinions. There are also sites included that some will argue are 'prediction rankings' - but that also is a form of Power Ranking (see the quote at the top) since it uses algorithms to calculate a team's likely final position.

Bias for or against certain teams are sometimes an unfortunate feature of power rankings. It can be based on their analytic formula, or the raters preference. In some cases the bias can become extreme - If you feel a site got the ratings wrong for your team I invite you to contact them regarding it. Bias ratings for every site are listed below.

I will typically only remove sites for two reasons.

Duplicating / copying ratings from another source on the list.
Failure to update for 6 consecutive weeks. Previous ratings will not be removed, but new ones may or may not be added. I will still periodically check them to see if they have resumed.

All the averages use the Olympic method (I.E. Truncated Mean value) of dropping the highest and lowest scores, and then averaging the rest to reduce bias.

The full spreadsheet is here.