What is an upset?
Also, a way-too-early look at RPI (my nemesis).
A couple of weeks ago, I talked about the VBelo model at a high level and mentioned there was an important use case for creating a ranking system. That use? Figuring out if a win is really an upset or not. So, what is an upset?
I think an upset is when the winning team was not expected to win. The real question becomes how you determine who is likely to win. (Spoiler: humans have a lot of ways to do this.)
When an upset isn’t an upset
Let’s start with some of the ways I have seen “upset” been used. To be clear, a match can have several of these all at the same time. Sometimes these are upsets and other times, it is debatable.
Ranking Mismatch
This is the one that gets under my skin a little bit. Sure, sometimes when #10 defeats #6, it is an upset. But sometimes, it is just because the rankings are poorly calibrated.
Program Firsts
When a program gets their first victory over a team, it often feels like an upset. Sometimes this is true. Other times, it is just the fact that the winning team is better than they were in the past. This is often an emotional upset, especially if a team hasn’t lost to a specific opponent in a long time.
Home Court Spoilers
Home court advantage is a real thing, but just because the team is a visitor does not mean their victory is an upset. It definitely impacts the odds of winning, but there is a lot more to it.
A “Real” Upset
Part of the the joy of an elo model is that there is a quantifiable way to look at matchups. When it comes to upsets, it is more a matter of where to draw the line. Sure, the factors above can play into the narrative of the match, I like to think of a real upset as involving a >40% difference between the probabilities for winning (i.e. the winning team has less than 30% chance to win).
For example, yesterday UCLA had a 43% chance of beating UC Irvine. There were a lot of “other” factors in the narrative for the match. UC Irvine was ranked ahead of UCLA and UCI was at home. UC Irvine was also undefeated coming into the match. Was this an upset? According to VBelo and my 40% rule, no. It was a great match that could have realistically gone either way.
Biggest Upsets of the Year (So Far)
First, any match that includes a team not in D-I or D-II will not be included. The model has a static elo rating for these teams so the probabilities are not finely calibrated. I’m also not going to include new teams, because I think the model still needs to learn more before an “upset” can be determined. So, with those matches omitted, let’s take a look at the biggest upsets of 2025 (so far).
UCSB (6.6%) def. Grand Canyon (93.4%) - 1/24/25
Ohio (12.4%) def. UCLA (87.6%) - 1/18/25
Stanford (15.0%) def. Hawaii (85%) - 2/7/25
Quincy (15.5%) def. Queens (84.5%) - 2/4/25
St. Thomas Aquinas (16.7%) def. Daemen (83.3%) - 2/15/25
Some of these were called upsets at the time, others, less so. What do you think? Are these all upsets? Do you think there have been bigger upsets this season?
Way-Too-Early RPI
I have decided to throw caution to the wind and publish an up-to-date RPI rating. The NCAA will not release RPI until much later in the season and I understand that. This rating will change dramatically in the next 2+ months of play. That being said, I think there is value in getting this out early. The more information we have about these teams, the better.
So, in an effort to create some better transparency, here is where RPI stands after matches on 2/19/25. Just remember, this will change. (Don’t say I didn’t warn you.)
Rankings Update
VBelo Rankings
With their solid win over Grand Canyon, CSUN has leaped into the top 10 this week. Last week they sat at 14th and 16th the week before. This is the type of movement that I like to see because it rewards the solid performances we have seen out of the Matadors. Will the coaches agree and rank CSUN in the top 10 in their poll next week? Only time will tell, but I really think they should.
Coaches’ Poll
I will admit there is a utility for the coaches’ poll, but I am disappointed by the continued apparent lack of nuance. This is nothing new. I understand there are a lot of factors that go into voting here, but I’m glad we have other ways to look at teams.
Media Poll
I’m a big fan of the work that Off The Block has done and is doing.1 That being said, I think it is time to expand beyond 10. I think it does a disservice to the sport as a whole. I understand that it is easier to think about the top programs, but there is so much happening in the world of men’s volleyball that I think it is time for a media top 20 (at least).
In’s & Out’s
A section for all of the random things that don’t fit anywhere but belong somewhere.
I can’t fully explain how ridiculous this is. 28 aces, in only 3 sets. To make this more ridiculous, Hu had only 1 ace in the first set (yes, I looked through the play-by-play). In Set 3, there was a run of 10 straight aces. I don’t care who the opponent was, this is incredible.
AVCA Player of the Week: Nolan Flexen (Opposite, UC Irvine)
Triple-Double Watch 2025: nada (but I’m not giving up)
Full disclosure: I have contributed to Off The Block before and I am currently a voter for the media poll.