The top third of the MORE is in lockstep with the latest NPI. Though lots of high leverage matches are still to be played! The at-large projection for each only differ in who out of the MAC will get the automatic bid and who will get the at-large bid. Though there is still a chance Misericordia could weigh in on that subject.
A phenomenon I have been looking at recently is where all these models see the 30 Midwest teams. (CCIW, NACC, MCVL plus Wisconsin Stevens Point & Westminster who are Independent.) I have known for years that Massey and IH see the Midwest a little more favorably than the T100 does. Recently it has been noticed the T100 sees the Midwest far more favorably than the NPI. However, the greatest disparity starts to build right about the rank of 25 and continues to grow linearly. Approximately a rate of 1 ranked spot weaker for every 6 positions moved on average. The graphic and table below summarize these findings:
Imagine I was to introduce 20 Senior International teams who had played a round robin of 19 matches against each other into this NPI data set. We all know they’d be the best 20 teams. But because there’d be no match played between any one of them and one of the D3 teams its measuring, the algorithm would have no choice but to straddle these 19 teams with any who went either 10-9 or 9-10 near a rank of 60 on its list. i.e. the middle. I’d expect those with winning records to be distributed above the median and those with losing records in some similar fashion distributed across the lower half.
Now consider the subset of 30 Midwest teams which do typically play about 10% of their matches against others not in the region. It is enough for the NPI algorithm to tilt the median Midwest team closer to a rank in the mid 50’s. (Not 60 as above.) However, the 3 experts who have been measuring the strength of these teams for years all agree the Midwest median is somewhere in the 40’s. What ends up happening is the teams not of the Midwest near the center of the whole landscape will typically win well more than half their matches because they are above the median of their subgroup. The teams from the Midwest near the center of the whole landscape will typically win less than half theirs because they are below the median of their subgroup which constitutes almost 90% of their matches played. These win rate differentials in the middle of the landscape in these respective regions are significant enough to exacerbate the kind of algorithmic bias seen above, particularly more and more after teams ranked among the best 20% in the landscape. The net effect of this is wins against above average teams from this region are systematically devalued. A ripple effect of this phenomenon together with the other points I’ve made regarding unbalanced incentives related to risk and reward might not have derailed who the NPI chooses in a few weeks, but it certainly may in future years.
I thought I’d offer the win matrix for the 290 matches played thus far between teams ranked in the Top 50 by the T100 through 3/16/25. In this subset of the whole population, 86% of the better ranked teams have retrospectively won their matches. I don’t think it a coincidence that about half of all the wins seen below the diagonal (upsets) were performed by the less than one-third of the Midwest teams who made this Top 50 list. And about half of those were not against other teams from their region, either. Which is to suggest, maybe they weren’t upsets, after all? It lends some anecdotal evidence to some conjectures offered above.
TRIVIA ANSWER:
[1] Misericordia is 9 ranked spots higher than its Median according to Inside Hitter.
[2] Dominican is at least 6 ranked spots higher than its Median according to the NPI.
[3] Lancaster Bible is 6 ranked spots better than its Median according to the NPI.