For those who believe the 3 extra bids and the dismantling of the Pool B concept this year made a kinder-gentler at-large bid procedure, consider this. If Loras were to defeat Carthage in a CCIW Conference Final and Messiah were to defeat Stevens in the MAC final, given the MORE model’s order of strength seen below, then Springfield at #8 would not be in the NCAA’s this year. These conditions would make it so the at-large bids would come from those ranked #3, #4, #5, #6, & #7. What if the AVCA Poll out today ruled the decision instead, under the same set of conditions? A reprieve for Springfield at the price of the defending champion Cal-Lu. OUCH! Right about now there are 4 humans wiping their brow, almost relieved that decision won’t be up to them, if it comes down to that. Then there are about 40 others wiping theirs hoping the NPI can be so lucky! Not to be an alarmist, though, because the chances are still 7 to 1 against it happening – On par with flipping a fair coin to heads 3 times in a row. I just picked up a coin on my desk and flipped Tails 3 times in a row! Whew! That was close… WAIT!
ONE OF THESE THINGS IS NOT LIKE THE OTHERS: The AVCA as it pertains to Hobart, the Massey as it relates to Cal-Lutheran, Inside Hitter’s Juniata position, and the T100’s Messiah rank. I am willing to take a stab at each of these to see where it takes me.
AVCA’s Hobart at the price of Loras and Aurora? Aurora has lost just 3 times in 34 matches the last 13 months, to Carthage & Cal-Lu last year and Southern Virginia earlier this season. Aurora defeated Cal-Lu and the team that defeated Cal-Lu this year, too. Cal Lu’s only losses so far. Hobart’s best wins are against the AVCA’s own 16th and 17th ranked squads this season, and this brings them to #11. How many of the 30 voters aren’t Midwest coaches? LOL
Massey’s Cal-Lutheran better than Stevens & Carthage? To be fair, according to its own system, Cal-Lu does have as many wins against Top 30 teams as the other two combined. I can forgive that, particularly because there is plausibility for it being correct while the rest of us are wrong. They are the defending National Champ, after all.
Inside Hitter’s Juniata beyond what Springfield, Stevens, and Carthage have done so far? All I can say is Carthage, Juniata, and Stevens went 1-1 in their round robin at the FJ Invite and the other losses by those three are two to NYU, the #1 team in the land, and the defending National Champ Cal-Lu. Now, I happen to think pretty highly of Messiah as this next paragraph will indicate but compared to those other losses by these 3 teams, Juniata losing at home to Messiah just doesn’t register as much in their favor. I would recommend not counting quality wins because their case becomes weaker.
T100’s Messiah rivaling Springfield or Loras? Sure! Why not? Messiah and Springfield both lost to NYU and the Falcons with 3 times the number of good wins against Top 30’s doesn’t discourage it above the Pride. Massey and Inside Hitter believe this is wrong because they see Loras as better than Messiah. The T100 has a history of ranking the Midwest teams a little below these two models. They could be right! Time will tell.
The MORE isn’t the only composite way to rank D3MVB teams because ordinals (ranks) aren’t the only characteristics the experts publish. Other than the AVCA, each offers a numerical metric for every team, the very thing that ranks them in the first place. The metric for each team can then be standardized to determine any team’s median. For example, St. John Fisher has metrics from Massey (7.19), T100 (16.65), and IH (21.56) that are as different as apples to oranges to grapefruit. However, when standardizing them the following z-scores are determined for each: Massey (1.50), T100 (1.46), and IH (1.17). These are the number of standard deviations from the expert’s mean to their team’s metric. The median of St. John Fisher’s standardized scores is therefore 1.46 because it is the middle of them. (I could use its mean of 1.38, but then there’s a risk for just one model exerting too much influence over it.) Listed below are the 5 best Median z-scores in the landscape. Maybe they look pretty familiar?
| NYU | 2.01 |
| Southern Va. | 1.90 |
| Vassar | 1.78 |
| Stevens | 1.72 |
| Carthage | 1.67 |
However, my purpose for determining these is not to provide another ranking because the MORE is already likely closer to the truth than any one of them. My interest is in measuring rank order noise each expert’s model shows about the signal as a way to demonstrate where they are on the continuum of converging to the truth, here less than half-way through the 2025 season. i.e. The expectation is for less noise than what is seen below in March, and even less than that in April.
This makes the graph below less about the teams and more about the color-coded points they produce about its signal for team’s strength in playing volleyball.
Rank order noise along a continuous independent axis of standardized scores is seen by the scatter above, and the relationship of that noise to its expert model from where it comes tells a story of the differences in ranking models, presumably based on what their algorithms value more or less than the others.
More noise is higher variability which may produce less confidence in a model. If I were to superimpose the NPI ranks on the graph above and the noise were noticeably higher for it, that could signal a lack of overall confidence in its order. The amplitude of noise seen between them whispers the degree to which differences exist, not necessarily for which is closer to the truth. However, should any of its noise be systematically driven in a way that appears not to be randomly produced, that could be an indicator of systematic differences in the models. i.e. algorithmic bias.
If I were to standardize the NPI metrics to see that they were very close to the other three models, for example, that would provide evidence for it being less different than the others. Less differences than others, even with systematic differences in noise patterns, together, shouldn’t raise concerns nearly as much as if the differences were both glaring and systematic.
Later this week I intend to demonstrate ways to search for differences in categorical variables related to the expert models. How do they rate newer programs in the landscape? Are there any differences in how they rate teams from the best conferences? What is each one’s view of the new conferences formed in 2025? How does it perceive of the strength of teams based on their regional affiliations and/or conferences as a whole? Are there differences in the way each ranks teams whose mascots are some form of wild cat? (LOL, maybe not that last one…)