A lot has changed in the landscape of the women’s NCAA ice hockey tournament over the past 18 or so months and it seemed prudent to try to have an explanation of everything all in one place.
In December 2021, it was announced that the field would expand from eight teams to 11 for the 2022 tournament. The additional teams created a “regional semifinal” round of games. The 4/5 quarterfinal was not affected, but teams 6-11 were matched up in a play-in game of sorts, with the winners playing the 1-3 seeds. Those games were played on Thursday, with the quarterfinal game on Saturday. (See last year’s bracket here).
As part of those changes, the selection criteria that focused on minimizing flights as the primary goal in quarterfinal matchups was removed and a requirement to avoid first-round intra-conference matchups was added. This changed the way the Selection Committee approaches the field using information other than data.
But for this season, the NCAA Ice Hockey Rules Committee also made changes to the way it calculates the data of a team’s schedule, opponents and outcomes.
They approved the move away from RPI (Ratings Percentage Index) to NPI (NCAA Percentage Index) as one of the calculations that goes into tournament selection. The NPI was recommended as a simpler and cleaner way to show a team’s schedule strength.
RPI was calculated using a team’s winning percentage, opponents’ winning percentage and opponents’ opponents’ winning percentage. NPI is calculated based on winning percentage and the opponent’s NPI rating itself, which is intended to provide a more accurate gauge of strength of schedule and much simpler and cleaner math.
The NCAA Competition Oversight Committee (COC) officially approved the change from RPI to NPI for the 2023 championship season at their summer meetings in August, 2022. The relevant part (also here, on page 3):
“The COC approved the sport committee’s request to use the NCAA Percentage Index (NPI) as a replacement for the Rating Percentage Index (RPI) in the selection criteria, effective with the 2023 championship. The sport committee had asked the COC to consider this in 2020 and was advised to use the NPI in conjunction with – instead of a replacement for – the RPI for two years to assess its impact on selections and return with a recommendation to the COC. In that time the sport committee has affirmed its belief that the NPI provides a better calculation that is based on winning percentage and the opponent’s rating itself (rather than the combination of opponents’ winning percentage and opponents’ opponents’ winning percentage) as the measure of schedule strength.”
To be considered for an at-large selection bid, a team must have an NPI of 50.00 or above.
When this move was proposed and then endorsed by the Championship Committee in 2020, the plan was to weigh components in proportion to what the RPI had used, so the NPI would be 30% winning percentage and 70% strength of schedule. (The RPI had been 30% winning percentage/24% opponents’ winning percentage/46% opponents’ opponents’ winning percentage).
Within this calculation “opponent’s NPI” is interchangeable with “strength of schedule.”
When the 2023 Championship Manual was released, that proportion had been changed. NPI weighs winning percentage at 25% of the calculation and strength of schedule at 75%.
The change to NPI has brought with it a change in how the quality win bonus (QWB) is calculated and awarded.
The QWB used to be set out as being “awarded for wins against the top 12 championship eligible teams.” Now, the QWB is awarded for all non-regulation-time losses against teams with an NPI greater than 51.50. Or, put another way, a team gets extra points toward its NPI for a victory or overtime loss against a team with an NPI of 51.5 or better.
To simplify it somewhat, think of each game a team plays as having its own NPI. A team’s season NPI is the average of all of those NPIs together. As with the RPI, “bad wins” (wins that would hurt a team’s ranking) are removed and QWBs from “bad wins” that have been removed are not added. Wins that would lower a team’s NPI are removed and the average is calculated using the remaining number of games as the denominator.
Your final NPI number is the calculation of 25% of the winning percentage and 75% of your opponents NPI plus your own QWB for each game. And that number is what’s divided by the number of counted games to get your NPI.
It should be noted, I think, that none of this is a perfect system. The Pairwise was created for the men’s game, which has 60 teams. There is less interconference play on the women’s side, so strength of schedule is much more impacted by the strength of a team’s conference.
But since it’s the system the NCAA uses to select the tournament field, it’s the framework we have.
