Fun With Numbers: 2025-2026 NPI, KRACH, GRaNT, and whatever else is out there

[TonyTheTiger20]

I have started the process of updating my calculators for the new season and adding in the newest member of D1, Delaware. As many of you know the Pairwise is goneeeeee!!!!!!! HOORAY!

The bad news is it's just using the NPI which is 99% of what the Pairwise was anyway. Common Opponents and Head To Head points are now gone. This is a little sad to me personally because it means all the hard stuff that I built over the years now gets deleted But on the plus side, an NPI calculator is going to be a billion times easier to maintain.

Here are the notes from the summer meetings talking about the changes for the women's game:

https://ncaaorg.s3.amazonaws.com/championships/sports/icehockey/nc/women/Aug2025NCWIH_Report.pdf

I'm happy to see the women are sticking with a 67/33 split for overtime (the men are moving to 60/40 which is just so stupid), and the women are not going to implement home/road weights (thank God).

One interesting tidbit was this paragraph:

The committee members acknowledged that they must include a new dial (minimum wins) as part of the transition to solely using NPI. When asked for a recommendation, Mr. Danehy advised 12 as the minimum number of wins, as that would not have impacted the last four years of the championship (all teams had kept at least 12 wins). This dial will be added to the NPI if it is adopted. Additionally, the committee confirmed that the first six ranked teams in the NPI that do not receive automatic qualification into the championship should be selected into the field, no matter their win/loss record.

I could not for the life of me figure out wtf this was talking about (it made it sound like you wouldn't get in as an autobid unless you had at least 12 wins or something), but CHN explained it in their men's article:

NCAA Committee Officially Approves Move to NPI

The NCAA Men's Ice Committee formally approved the move to NPI (NCAA Percentage Index) for use in selecting and seeding the NCAA Tournament. It replaces the Pairwise system, which, in conjunction ...

www.collegehockeynews.com

There is a final factor, which is "minimum wins," which is 12. This can be confusing, because it doesn't mean a team needs 12 wins to qualify. It is related to the concept of throwing out "bad wins" from the NPI. In the past, any win that actually had the affect of lowering that team's RPI (because their opponent was very poor and thus the poor strength of schedule overshadowed the positive on a team's winning percentage), were tossed out of the RPI calculation. Now, a team must count at least 12 wins, regardless of whether those wins lowered their NPI. Once they've counted at least 12, a team can throw out its remaining "bad wins."

This is mildly interesting; it used to be if you were undefeated your NPI (or RPI in the past) would be solely based on your one best win with all other games thrown out of the calculation, since every other win would necessarily reduce your NPI. If you have only one or two losses, you would end up getting most of your results thrown out as "bad wins". This change instead means that the calculation will ensure that at least 12 wins are included in the calculation, even if 11 of those are "bad wins" that reduce your NPI.

In practice this probably won't really change anything... but it's an interesting thing for them to have thrown in there.

I'll have the NPI ready in a couple weeks once the Ivies have a couple of games in the books and will release the new GRaNT and KRACH for the upcoming season too.

 

[ARM]

This is mildly interesting; it used to be if you were undefeated your NPI (or RPI in the past) would be solely based on your one best win with all other games thrown out of the calculation, since every other win would necessarily reduce your NPI. If you have only one or two losses, you would end up getting most of your results thrown out as "bad wins". This change instead means that the calculation will ensure that at least 12 wins are included in the calculation, even if 11 of those are "bad wins" that reduce your NPI.

Why don't they just use a system that doesn't have a ridiculous concept like a "bad win"? It's not like such a system doesn't exist. Instead, every so often they roll out another bandage to try to hide that blemish.

I'll have the NPI ready in a couple weeks once the Ivies have a couple of games in the books and will release the new GRaNT and KRACH for the upcoming season too.

Thanks for all of the work that you're doing, both figuring out what needs to be done, and then doing the faithful task of updating throughout the season.

 

[TonyTheTiger20]

Why don't they just use a system that doesn't have a ridiculous concept like a "bad win"? It's not like such a system doesn't exist. Instead, every so often they roll out another bandage to try to hide that blemish.

It's honestly so ridiculous. I don't understand why they went to this pseudo-KRACH that is absolutely not KRACH but can be explained to the coaches in the exact same was as KRACH.

The more I programmed this the more ridiculous it all looked. You can very easily draw up a scenario where a team gets a bad win that only gets partially thrown out! For example, if a team has 11.5 bad wins (ties count as half a win, and the win portion of a tie gets thrown out if it hurts your NPI), and then they get another "bad win" to give them 12.5, that last bad win will only count as half a win to get them up to exactly the 12.0 required, and then the rest gets thrown out. It's so incredibly goofy.

Don't even get me started on home/road weighing which fortunately only matters on the men's side. A home OT win in men's hockey this year counts as 0.48 wins and 0.48 losses. What are we even doing?

Thanks for all of the work that you're doing, both figuring out what needs to be done, and then doing the faithful task of updating throughout the season.

As usual I'm the only person on earth who takes such great enjoyment out of doing this stuff, so the pleasure is all mine lol

 

[ARM]

As usual I'm the only person on earth who takes such great enjoyment out of doing this stuff, so the pleasure is all mine lol

Hardly, although some with the mathematical abilities needed to do so aren't social enough to find their way to a social media site to talk about it.

 

[robertearle]

It seemed to me that this year Wisconsin built their non-conference schedule so as to have a "common opponent" in each of the other conferences. If there's no more Pairwise, there's no "common opponent" element to pay attention to. I wonder if going forward, this will change how teams look at NC scheduling.

In women's volleyball, the NC is a series of top 20 vs top 20 matches; everybody looking to boost their RPI (a major factor in tournament seeding), hoping for "good wins" but OK with taking a "good loss". Do we see more of that now in women's hockey?

 

[ARM]

It seemed to me that this year Wisconsin built their non-conference schedule so as to have a "common opponent" in each of the other conferences. If there's no more Pairwise, there's no "common opponent" element to pay attention to.

I'd be surprised if UW weighed the PairWise that heavily in NC scheduling. Over the years, Johnson looks to have transitioned from trying to include an OOC NCAA tourney team to including weeks in the schedule where the focus can be largely on the Badgers and playing the way that they want to play. Particularly after winning the title via a lower seed, a team like Wisconsin doesn't have to obsess over the rankings. If they can find a way in, then even an extra game against a team from the bottom half of the field wouldn't seem to matter much. Given the uncertainty of the Olympic year with absences, even the top seed might not be that valuable; the biggest threat to a UW could easily come out of the #4 vs #5 game.