Latest USAU Ranking Approximations: Easterns & Centex Shake Things Up

Get an early look at the bid scenarios.

March 22, 2016 by Charlie Eisenhood and Cody Mills in Analysis with 37 comments

Ultiworld’s Cody Mills has put together a statistical tool to approximate the USA Ultimate rankings by using publicly available score data and applying the algorithm that USA Ultimate uses to run their numbers. While the results below are unofficial and could include some scores which are liable to be altered or removed from the official rankings, they are a good representation of the what the current rankings would look like.

Throughout the remainder of the college season, we will be running this USAU ranking approximation and, later, applying different algorithms and/or ranking methods to test possible alternatives to the existing algorithm.

For more information about the USA Ultimate algorithm, ranking system, and some of its perceived flaws, see this FAQ from last year.

Men’s Division

The rankings below include only teams with at least five sanctioned games.

Implied Bid Allocation

Atlantic Coast: 2 bids
Great Lakes: 1 bid
Metro East: 1 bid
New England: 2 bids
North Central: 3 bids
Northwest: 4 bids
Ohio Valley: 1 bid
South Central: 2 bids (-2)
Southeast: 2 bids (+1)
Southwest: 2 bids (+1)

Discussion

The big moves this week: TCU and Missouri both fell from the top 18, significantly darkening the South Central’s hopes for a bid bonanza. Georgia’s strong play at Easterns vaulted them into the promised land, adding strong hope for a multibid Southeast, and Cal Poly SLO is boosted just above the cutoff thanks to teams above them falling out, adding a second bid for the Southwest (at least for now).

Much rides on the two coming weekends, with Northwest Challenge, Huck Finn, and Centex likely to generate many more changes. Teams like Victoria, Cal Poly, TCU, and Texas A&M will have a lot on the line heading into the final push of the college regular season.

Note that UMass’ dominance at Easterns has sent them to #1 overall, which will give them a good shot at the overall #1 seed at Nationals.

An interesting side note: BYU is the first team outside the cutoff at #19. They played a very good game against last week’s #1 Wisconsin at Trouble in Vegas this weekend, losing 12-10. BYU will be competing at the Northwest Challenge. As discussed here, there are some very interesting conversations ahead if BYU manages to earn a bid for the Northwest.

Women’s Division

The rankings below include only teams with at least five sanctioned games.

Implied Bid Allocation

Atlantic Coast: 1 bid
Great Lakes: 1 bid
Metro East: 1 bid
New England: 1 bid
North Central: 1 bid
Northwest: 5 bids
Ohio Valley: 2 bids (+1)
South Central: 3 bids
Southeast: 1 bid
Southwest: 4 bids (-1)

Discussion

The biggest move is the only move: Ohio State reclaiming a bid for the Ohio Valley, to the detriment of the bid-rich Southwest. Fever put together a good enough showing at Centex, particularly with a notable win over UCF, who is the 4th best team to beat. Meanwhile, UC San Diego opted to go to College Southerns instead of their usual trip to Centex, and that decision could contribute to Southwest’s bid total. Despite the Psychos cleaning up at College Southerns, the meager competition dropped their total points; USC’s Centex struggles also hurt the Southwest’s chances and second order scoring.

There’s a chance these numbers are pretty close to the final rankings. Western Washington, currently at #18, is the only team near the cutoff that will be playing at Northwest Challenge, and barring key losses to the tournament’s lower ranked teams — Victoria and Carleton — it will likely be hard to WWU to move much. That means we are that much closer to a shocking four bid Southwest and to another year of west coast dominance, with nearly half of the teams at Nationals coming from the states of California, Oregon, and Washington.

Some Notes About This Approximation

We, of course, looked at the results of the initial USA Ultimate rankings and how they compared to our approximations. There were few discrepancies, but there were some. Here is some discussion from Cody Mills on the differences:

1) Two weeks ago was the first test for the projection algorithm in the college season (as opposed to club), where sanctioning is trickier (alumni teams, etc.) and team names are more troublesome.

– For example, many teams are listed on score reporter as Stanford (Bloodthirsty) on their team page, but when listed as an opponent appear as simply Stanford. This seems pretty straightforward until a team with a name like Wheaton (Illinois) (Mastodons) comes along. Consequently, these teams were erroneously counted twice in the earlier projections.

– Alumni teams/unsanctioned teams also give some trouble. While they can be filtered out on one side by using the filters in UI, all games a team has played appear in their opponents list. Thus, at PLU BBQ, while PLU Alumni did not have their team page scraped for scores, their scores still came in by way of all their (legitimate college) opponent’s team pages.

2) The projection algorithm is consistently about 28 points lower in rating for all teams than the official USAU ratings — this occurred in club also. However, despite the translation, the relative distance and order of the teams did not change. It appears this is being caused because USAU is using 1-0 scores when they are running their algorithm (mostly from Big D in Little D). There used to be a regulation that required a game reach seven points (or half) for it to count, but it’s apparently gone now. That said, I’m pretty sure these 1-0 games will eventually be thrown out.

3) Some game results were unpublished or changed, which caused some small downstream effects, especially on the women’s rankings from two weeks ago.