What is this?
Each season, USA Ultimate publishes power rankings for each team using an algorithm it developed. If the algorithm thinks highly of a team, its region is rewarded with an extra bid to the national championships. Teams that want to go to Nationals care deeply about their ranking. But because math behind the algorithm is rather complicated, it's difficult for teams to understand why they're ranked where they are.
This site's goal is to give players and fans some insight into the workings of the algorithm. By going to a team's page, you can see the effect of each individual game and tournament on the team's rating.
The "effect" column on each page answers the question "what would my rating be if we had never played this game/tournament?" As an example, say a team has a rating of 1000 and an particular game has an effect of +50. That means if that game had never been played, the team would instead have a rating of only 950.
How does the algorithm work?
In short, it looks at the ratings of the teams you played and your point diff in each game, and comes up with an average. Losing a close game against a top-rated team can actually raise your rating, and winning a close game against a weak team can lower it. For the gritty mathematical details, see USA Ultimate's description.
Why are some games listed as "ineligible?"
This means the game is not eligible to be included in the rankings. This could be because one of the teams is in the wrong division -- alumni, high school, or international, for example. It could be because one of the teams played with an unrostered or academically ineligible player. It could be that the game was not actually played and was misreported on the USA Ultimate website. In many cases, I don't know the exact reason and merely inferred it from official win-loss records.
Why are some games listed as "ignored?"
In the past, if a highly-ranked team played a much weaker team, their rating would be always go down even if they won the game 15-0. This led to teams simply forfeiting games to preserve their rating. So in 2014 the algorithm was modified to completely ignore the results of blowout victories under certain conditions.
Ignored games, unlike ineligible games, were considered by the algorithm and would have been included if the score had been closer.
How did you get the data?
I used a script to scrape data from the USA Ultimate website. I fixed some data manually in an attempt to match my results to the official ones.
How accurate is your data?
In most cases, it's identical to what USA Ultimate used in their official rankings, but there are some errors. If you see a mistake, please let me know.
Why don't the effects add up to the team's rating?
The short answer is that's just not the way the math works out. Removing two games together is not necessarily the same as removing each game individually. If a team only goes to one tournament, however, the effects will add up to zero.
Who are you?
Hi, I'm Nate! I played club for 8 years in San Francisco and Sydney with CTR, Umlaut, grasSFace and Groove. Now I'm semi-retired and relegated to masters, rec leagues, observing, and dissecting the ranking algorithm.
If you'd like to contact me, please send me an email at nate.paymer@gmail.com.