NPI Rankings (Week 2)

With the move to the NCAA Power Index (NPI), the relevance of the rankings, whether it be from D3VbWest or the AVCA, has really gone from unimportant to unneeded. Still, with the decision by the NCAA not to release their numbers until mid to late October, I guess the rankings, at least, gives us something to do. Throwing a curve into that thought is that D3VbWest has two entities that are simulating the NPI and allowing the this website to display the data. I have created a Page on the left side menu (or pulldown) that will have the data. It is titled 2024 NPI Rankings.

What I’m doing is getting the data from “Ned” and from my Latvians and comparing for discrepancies. Both require pristine schedule/results data for the numbers to be correct and each is getting their data from different sources. I will do the comparison and let each know where/if there is an issue. After I’m done with that effort, I will update the website with the data noting any areas that could be incorrect. (For this week, Ned is working on some front-end collection so the Latvians’ data will be used.)

I can’t stress enough how important it will be for the NCAA to ensure that the results are correct. Every result can ripple through many teams and as we saw with our simulations of past year’s data, a small change can mean the difference of a team getting an at-large bid. I have already received reports of errors in the results so we tread lightly. Again, it’s one thing for the NCAA to miss a game or two during the year when it doesn’t impact anything. It’s entirely different when it means everything. Another problem we discussed today in our back-channels was who is eligible for NPI. We know that UT-Dallas is not, but there are other teams transitioning to and out of DIII. It would be good to get clarity on every team.

Looking at the data, things have really come around through the first two weeks. Remember that football, with its limited schedule, is also using NPI this year. This is around the number of games (matches) where things get serious for them. I’m on record saying that this system will not work for football but this, at least, gives me some hope.

Remember that with volleyball, each team must count up to 10 wins and all losses count as long as they don’t improve your NPI. Once teams get to 10 wins, I expect to see these numbers jump a bit as bad wins get removed (only from the team that won). Last year, we had 6 teams above 70 NPI and the at-large cutoff was around 60 NPI. We only have 13 teams above that at-large cutoff from last year so our numbers will rise.

I had a question this past weekend that asked how the numbers are used throughout the year. Are they basically static and built upon as the season progresses or do they reset for every match played and get recalculated? The answer is that they always get recalculated whenever a new match is entered. The reason is that a single match will change the NPI for those two teams and for every team that played those teams and so on and so on. If Team A beats a 4-0 team (Team B) and gets a nice NPI boost and a Quality Win Bonus (QWB) then just know that if Team B loses its next 26 matches, that NPI given for that win will be decreased significantly and the QWB taken away. A QWB is given to the winner when beating a team with an NPI of 55.5 or higher. The bonus increases the higher the NPI of the losing team as compared with 55.5.

Let’s take a look at the numbers.

Leading us off is Juniata and by a good margin this early in the season. Juniata’s wins are against Chris Newport (52.39), William Paterson (52.77), Johns Hopkins (59.74), Ohio Northern (56.04) and Swarthmore (51.64). Remember that each team has at least one loss because they played Juniata. If I apply my D3VbWest Rule-Of-Thumb of adding 10 NPI to each team’s NPI that you beat, I come away with a Juniata NPI of 64.52, which is pretty close to the actual number.

I’m not going to step through every team, but Gettysburg sitting second has to be a bit surprising. They had a great season last year and have started off great again this year going 7-0. Like Juniata, Gettysburg’s wins are all pretty good with the exception of York. They have two wins over teams that would give them a QWB.

Hope sitting 3^rd is great to see. MIT sits fourth and they do have two wins over teams that would give them a QWB. It will be interesting to see how NPI likes the New England teams this year. The next highest NE team after MIT is Wellesley sitting at 21. Marietta has to be another surprise sitting at 5^th but they do have a nice win over Calvin and another against Carnegie Mellon.

Right now, bad wins are really hurting teams because they have to count, which is why we see some very good teams not up high on the list. Emory sits 81^st because they have wins over Greensboro (45.17), Huntingdon (33.29) and Piedmont (36.14). These will all go away in time but for now they drag their NPI down significantly. It will be interesting to see if teams shy away from these types of matches in the future but they really don’t cause any harm as long as you have 10 good wins to bank.

I’ve been asked directly what I think about NPI and my response is that I don’t like it because it’s fundamentally flawed. As long as teams with very weak Strength of Schedules (SOS) can do well in this system, it flawed and shouldn’t be used. There are simple minimum levels that could be implemented to fix most of this but I assume that would have to be approved across all sports and that likely doesn’t happen. Let’s take a look at Colorado College who has an NPI of 48.82. That’s not good and this would be a team (as they stand right now) that if you played and beat, wouldn’t help you get to the at-large cutoff. Their problem is that they have played UW-Oshkosh, Gust. Adolphus, Transylvania, Emory, Tufts and Cal Lutheran and are 2-5 in those matches. In the past system, that schedule would be lauded and praised by the Heralds. We really have no way of knowing who would win a straight up match, but do you bet against CC playing teams like Penn-State-Abington, Grove City, Franklin or University of New England? NPI says those teams are better than CC. I say the system is flawed.

The data is posted for you to check on your favorite team. If you have a question or think you see an error, please comment back to this article. I’ll try to figure it out.

Oh!!!! My Latvians sent me an interesting note:

“After two weeks, there is one pod of two (Crown and North Central (MN)).  All the other teams are in one pod meaning that every team is an opponent of an opponent of every other team.  That means every game affects every team’s NPI.”

Maybe this is normal but it kind of blew my mind a bit. After only 2-weeks, every team except for two are impacted each time someone else plays.