clock menu more-arrow no yes

Filed under:

NCAA Changes Pairwise Criteria For Women's Hockey Tournament Starting This Season

Changes could affect which teams make the NCAA tournament

BC Athletics

The women's hockey Pairwise rankings appear to have undergone a major change.

A deep dive into the NCAA manual for the 2016 women's hockey tournament shows that two changes to the selection criteria were made over the most recent offseason.

The 2015 women's hockey manual read as follows:

In addition to Bylaw 31.3.3, the Women’s Ice Hockey Committee has received approval from the Division I Championships/ Sports Management Cabinet to consider the following criteria in the selection of at-large teams for the Women’s Ice Hockey Championship (not necessarily in priority order):

  • *Rating Percentage Index (RPI) [won-lost record (30 percent), opponents’ winning percentage (24 percent) and opponents’ opponents’ winning percentage (46 percent)];
  • Head-to-head competition;
  • Results versus common opponents; and
  • Results versus teams under consideration (defined as those teams with an RPI of 50.00 or better).

*If points awarded for any win lower a team’s average RPI, those points will not count toward the RPI.

This has been the criteria for several years now—essentially the classic Pairwise Ranking formula that was also used by men's hockey (with different RPI weightings) before the men made similar changes to its tournament for the 2013-2014 season.

The 2016 women's hockey manual was updated to read as follows:

In addition to Bylaw 31.3.3, the Women’s Ice Hockey Committee has received approval from the Division I Competition Oversight Committee to consider the following criteria in the selection of at-large teams for the women’s ice hockey championship (not necessarily in priority order):

  • *Rating Percentage Index (RPI) [won-lost record (30 percent), opponents’ winning percentage (24 percent) and opponents’ opponents’ winning percentage (46 percent)]; including quality win bonus**
  • Head-to-head competition; and,
  • Results versus common opponents.

*If points awarded for any win lower a team’s average RPI, those points will not count toward the RPI.
**Quality win bonus awarded for wins over the top 12 championship eligible teams in the final selection RPI.

The two major changes, you may notice, are that the "results versus teams under consideration" criteria was removed, and the "quality win bonus" was added. This is essentially the same change that the men's hockey committee made in 2013-2014, without the extra home/away weighting.

These changes could affect which teams make the NCAA tournament, particularly with such a logjam for the last few available at-large spots.

This also means that the only place you could find the women's hockey Pairwise rankings, USCHO, is now outdated. But never fear! Here at BC Interruption, we recreated the Pairwise rankings using the new criteria as spelled out by the NCAA manual, removing TUC from the calculation and adding the Quality Win Bonus.

It will be interesting to see if the new criteria affects the seeding in the tournament field, or if a team that would have made the tournament under the old system is left home under the new system.

UPDATE: USCHO has taken down its Pairwise Rankings with the following message:

USCHO has learned of changes to the selection criteria for the Division I women's NCAA tournament for the 2015-16 season, which made our existing PairWise Rankings and Ratings Percentage Index not representative of the current process. We are working on a solution and hope to have the updated PairWise and RPI available soon.

So, there you go.

In addition, we've obtained some clarification from the selection committee regarding the quality win bonus as well as the number of comparisons:

The QWB scale for the women starts at a .060 RPI bonus for a win against #1 and descends .005 per team, ending in a .005 bonus for a win against #12.

...

Like the men, all eligible teams are now compared [as opposed to only teams with an RPI of > 0.500].

Note that the data is being pulled from a pretty large spreadsheet, so it will take a minute to load.