NPI Potpourri

With the release of the NCAA Power Index (NPI) parameters, I had a coach suggest I go back and look at the scheduling article I did a few days ago. Yes, it does changes things slightly. I also wanted to look at a couple of teams that were on the bubble last year and compare/contrast. Thirdly, there have been some general questions coming in that I should probably cover for the masses. Finally, I wanted to know if this 10-match wins minimum would change much with our original simulation that assumed no minimum. Whew. I guess what we have here is a little NPI Potpourri and, like any potpourri you come across, it’s going to smell okay but really make no sense of purpose when you come right down to it.

When I wrote my scheduling article, I was kind of all over the place but one point I tried to make is that scheduling bad teams really doesn’t hurt you anymore. One of the more interesting byproducts of NPI was that an undefeated Juniata team from last year really had their NPI constructed from one match (their best win according to NPI). Every other match (win) would have lowered their NPI so it was thrown out. With the 10-match winning minimum that is not the case. What this minimum does, in effect, is slightly change NPIs based on how hard a team’s best 10 wins were. So, looking at my fourth point above, having this minimum really doesn’t change much. In our simulations, it did change some ordering with UW-Oshkosh moving ahead of Juniata being a funny one. The reason is that UWO had 10 better overall wins (per NPI) than Juniata so this minimum does induce a scheduling component that is good to see. Unfortunately, it really didn’t do enough to eliminate the bad Strength of Schedule (SOS) teams from now getting into the tournament as an at-large.

Back to scheduling, and what these parameters do is sort of eliminate some crazy paths to an at-large bid that shouldn’t be allowed even if they were possible. You can’t just schedule a few quality teams and then pad the schedule with 25 poor teams. You’ll need to get 10 quality wins without piling up the losses that will drag down your NPI. Did you catch that? When it comes to scheduling, a coach’s job is now to get the 10 best wins possible without losing (or losing much). The teams that scheduled tough in the past (looking at you UChicago) shouldn’t do this unless you can win almost all of them. There is a real risk/reward with scheduling top teams now but the bottom line is that it makes no sense to schedule a bunch of them.

One thing to mention is that our simulated NPI uses percentages while the NCAA looks to be between 0 and 100. So, the NCAA Quality Win Bonus (QWB) is set at 55.5 and 0.555 in our formula, if that makes sense. (I just point that out in case I start accidentally swapping them around.) Looking at our simulated results for 2023, there were around 100 teams that had an NPI of 55.5 or higher. That’s around 25% of the total teams and this is important because beating one of these teams gets you a bonus. If you are trying to find 10 quality opponents, this is where I’d start. You want to beat these QWB teams. With that said, our numbers indicate that you want to get around 60 NPI for an at-large chance. You can get to 60 by beating average teams and not playing any QWB teams as long as you win. There were about 250 average teams last year (in our last simulation) meaning they had an NPI of 50 or more. What happens when you beat a team with an NPI of 50? With our 20/80 parameter, you get 0.2 for the win added to (0.8 * 0.500 NPI). That equals 0.600 (or 60 NPI) for the match! Lose that match and it nets you 40 NPI and because it’s a loss, it will count towards your total NPI for the team. (I should note here that we believe all losses that don’t increase your NPI will count. We have asked that question to the NCAA but have not received a response.)

One question that will come up is should a coach target more than 10 quality opponents? Sure, if you can beat them. The tricky aspect to this is that you have to identify these quality teams months in advance of the season before rosters are released. Of course, if you then lose some of these matches, you’ll want some backups.

I also talked about how there are 250-average to better teams last year that can help you (this is a change from the article when it was at 215 without the 10-match winning minimum). What I mean by this (just to be clear) is that they will help you to get to a 60 NPI and at-large consideration. What might be better, when it comes to scheduling is to target a safer number like 62. If you do this then you are back targeting the Top 25% (Top 100) plus around the next 80 or so as you want teams that have a 52 NPI number or better.

I believe I also wrote about targeting teams that are the class of a weak conference when scheduling. The theory here is that they will build up their NPI on the backs of these weaker opponents. With the old selection criteria, there was a problem doing this because opponent’s opponent’s SOS would trip you up. This is no longer considered at all in the NPI formula. The only question is how much can those teams build up their NPI? Here is a spot check of some 2023 teams that won their conference with their simulated NPI numbers:

• Penn State Altoona 51.4

• Marymount 55.5

• Hunter 54.5

• Wentworth 58.3

• St John Fisher 54.3

• Simmons 55.1

• UMass-Darthmouth 57.7

• Manhattanville 54.5

All of these teams are average or better with two giving you a bonus (what I call QWB teams) if you beat them.

I guess the final thought I have on scheduling involves the elite teams. What should they do? I’d say, nothing because they are elite. They will beat most of the teams across the net from them. NPI is going to impact the bubble teams, but I will say a team that schedules tough and doesn’t get a lot of wins will not get an at-large. We’ve seen this in the past with Ithaca where they played like 20 quality teams but only came away with a few wins. They still got an at-large because their SOS was outstanding but that will not happen with NPI. You have to win.

I’ve kind of mixed in some questions I’ve received over the last few paragraphs but another one I got was about how conferences impact NPI. In the past, playing in a conference with a couple of elite teams or a deep (quality-wise) conference would help your selection chances. Playing in a weak conference meant you had work to do. I think now it’s kind of up in the air. It’s true that with NPI having weak opponents in your conference or even a weak conference doesn’t hurt you like it did before. The trick in these cases is making sure you have enough quality opponents to go for whatever NPI number you are shooting towards. If you play a 16-match conference schedule and everyone is bad then you’d better fill out your out-of-conference with 10+ quality teams. As an example, we can look at Mary Hardin-Baylor (UMHB) out of the ASC. There were typically two quality teams in the ASC before it started to disintegrate and even now with just 5-teams total, they still have UMHB and ETBU. I’ve criticized UMHB in the past for their poor scheduling but they actually have a good formula (no pun intended) for NPI. (We’ll take a close look at this below.) For teams in tougher conferences, they will need to take a hard look at whether they need to schedule a tough out-of-conference schedule. In the West, I think this change to NPI can really help out the NWC. This conference is essentially isolated outside of the first two weeks and most of the schools can only travel for one of those weeks. In 2023, none of the NWC schools got close to an at-large when considering NPI. The top programs basically all came in around 53.5 NPI although conference champion Pacific Lutheran finished at 56.9. How do they get to that 60 mark? They need to ensure their out-of-conference teams are QWB teams (the Top 25%) or at least in the Top 250. If they can find tournaments with these teams or attract these teams to them and win, they will have a great start. Another thought is to expand their practice of playing conference teams in out-of-conference matches. Pacific Lutheran and Puget Sound do this already. The downside to this is that there were only 5 teams in the NWC in the Top 250 (PLU, Whitworth, George Fox, Puget Sound and Willamette) last year and these are the teams you need to be playing.

Another thing I keep seeing from coaches are references to SOS. This really needs to be forgotten in the way they think of it today. You need to focus now on targeting the Top 250 schools, for the most part, and as many beatable QWB (Top 25%) teams as possible. Once you’ve identified 10 or so wins, you can back off because the risk of a loss becomes too great. You should also never worry about scheduling a poor team you know you can beat as long as you have those 10 wins banked. One thing that isn’t talked about is how does this impact the elite programs when it comes to scheduling. Right now, most of these top programs can request and get a match against anyone else if the travel can be worked out. At some point, the teams just out of the elite range are going to realize that playing (and losing) to these elite schools is, most often than not, a losing NPI game. We may see a future where the elite programs will have difficulty scheduling in the Top 250. I hope we don’t get there. What will really be fascinating is if an elite program “suffers” a 20-10 season playing a really strong schedule and doesn’t get an at-large. Then we may have elite teams not wanting to play a lot of other elite teams while not being able to play teams in the Top 250.

To give you an example, Christopher Newport lost to Juniata in 2023 and received 58.2 NPI for that match. This is less than their team’s NPI but it has to be kept and averaged into their team NPI because it’s a loss. They also played and beat 8-20 Shenandoah in three sets 25-12, 25-12 and 25-18. They received more NPI from this match (58.6) but it wasn’t kept because it would have lowered their NPI. What they should have done instead was schedule another match with 16-17 Southern Virginia or a third match with 14-16 William Peace as those wins would have raised their team’s NPI.

At this point, I think I’ve covered everything I wanted to write about except for some bubble team comparisons. I asked my favorite Latvian (sorry Izzy) to give me the NPI breakdown for three teams that were on the bubble last season. They were Christopher Newport, Tufts and Williams. All received Pool C (at-large) bids. I also asked for a breakdown of Mary Hardin-Baylor who was never considered for an at-large due to their very poor SOS. Here are their Record, SOS and NPI (simulated) for each team from 2023:

• Chris Newport: 0.636 Record / 0.634 SOS / 60.2 NPI

• Tufts: 0.583 / 0.608 / 58.3

• Williams: 0.640 / 0.611 / 59.1

• Mary Hardin-Baylor: 0.926 / 0.460 / 61

Our simulated NPI has Chris Newport as the 21^st and last at-large team from 2023 and this is where we get our “strive for 60 NPI to have at-large consideration” from.

As mentioned, if NPI had been used for selection last year, Chris Newport would have been the last team in. Both Tufts and Williams would be out and neither would have been within 5 places of the last bid available. (First team out was Augustana at 60.1 NPI.) Mary Hardin-Baylor would have an easy at-large selection (17^th Pool C bid). I think we all can look at that SOS number for UMHB and scratch our heads but let’s look a little closer at how these team’s NPI got created.

UMHB’s NPI consists of 10 matches plus their two losses to ETBU. The rest of their matches are thrown out and not considered. The 10 wins were against Hardin-Simmons (3 times), Autin College, UT-Dallas (twice), Concordia-M’head, Schreiner, Hendrix and ETBU (they beat them once). They received a Quality Win Bonus just once (against ETBU as they went 1-2 against them). That’s really not an impressive list of teams but this will get you into the NCAA Tournament moving forward. The trick? You have to win, which UMHB did losing only the two matches to ETBU. To summarize, UMHB went 1-2 against the NPI Top 25%, 13-0 against the next 25% and played 11 matches against teams in the bottom half of our division.

Chris Newport had 18 of their wins count towards their NPI and had 9 wins against QWB teams. That is far and away better than UMHB but they are hurt by their 12 losses. The worst loss for Chris Newport according to NPI was to Centre (59.3 NPI). The only team UMHB played with a higher NPI was ETBU. Their next closest opponent was Hendrix who sported a 53.9 NPI. Chris Newport’s best win was against Carnegie Mellon who had a 62.9 NPI. Let all of that sink in. Chris Newport played a much tougher schedule, beat a ton more quality teams, had the better overall win and only lost to teams much better than who UMHB played. NPI says UMHB is the better choice.

Looking at Tufts, and they are interesting because no matches would have been removed. They went 14-10. Their worst loss was to Hamilton (54.1) but their best win was Springfield (63.0). They beat six QWB teams (which also included Trinity who had a 62.8 NPI). Same story when comparing to UMHB. The reason they were below Chris Newport are the wins. Tufts just had too many losses to get overcome by those 14 quality wins.

Williams went 16-9 with only one of their wins being removed in the NPI calculation. Their best win was Babson (61.3 NPI) and their worst loss was against Connecticut College (51.9 NPI). They also had six wins against QWB teams. So, again, a better overall schedule with better wins than UMHB but not good enough for NPI.

The reason I wanted to highlight some of these bubble teams is that it shows how NPI is a seismic change to how we evaluate the quality of DIII volleyball teams. The 2024 schedules are, for the most part, in the books (if not posted to their websites), but do teams take the UMHB path in 2025 or the CNU/Tufts/Williams path? Yes, you can say that UMHB won but they never challenged themselves. They never really came close to challenging themselves. Even with the 10-match winning minimum, this is still a path to an at-large bid that NPI favors.

To finish off this potpourri of goodness, the NCAA has decided the path forward for at-large selection is NPI. I believe it has serious flaws and with no human oversight we will be scratching our heads at a number of at-large teams this fall. In two-years there will be changes to some of the parameters to correct the flaws inherent in the formula. After four years, we will see how many coaches and administrators have seen enough or if the benefit of not wrangling with the selection criteria is enough to continue forward. We will also see how this impacts the elite teams and whether they will want to absorb the risk of continuing to play quality schedules. It will also be interesting to see how the non-elite teams feel about being NPI fodder when the risk of a match far exceeds the slight chance of a reward. As for me, I love this. I have a logical brain so give me the formula and I’ll schedule to it to give my program the best chance. I mean, the NCAA Tournament isn’t about bringing in the best teams, it’s about which team’s schools have the best statistics programs.