A Boneyard Special Pre-Committee Reveal Poll: The Super 16

[Vowelguy]

Does the NET algorithm over-value good losses. Look at UConn and Utah in particular

Both are good teams that have had close losses against very good schedules. And yet, yo have to ask: "where are the good wins?" I note that both the committee and our voters here on the BY Poll weren't buying into the lofty NET rankings.

Calling @Plebe

My theory has always been the RPI or the NET are useful for the fat part of the distribution where the teams anre bunched very tightly and you are comparing very disparate teams about who should get in and where they are seeded.
The tails of the distribution are always harder to measure empirically and they are needed less, because we have ample head to head and comparison data.

 

[Vowelguy]

My theory has always been the RPI or the NET are useful for the fat part of the distribution where the teams anre bunched very tightly and you are comparing very disparate teams about who should get in and where they are seeded.
The tails of the distribution are always harder to measure empirically and they are needed less, because we have ample head to head and comparison data.

And I will add that I think the committee finds greater value in things like “record top 50” etc. On the men’s side you hear a lot more discussion about quad 1 wins than you do about actual NET ranks.
The NET just serves as an initial way to sort the teams, which then allows the committee to look at who did well vs good teams, what’s a bad loss, etc