Jump to content

mewfert

Members
  • Content count

    3
  • Joined

  • Last visited

  1. Selection Committee Games of Interest

    That one is interesting, actually: what's happening there is that if Maryland loses to Rutgers, its projected RPI falls to 78, which means that Nebraska's home win over Maryland falls from Q2 to Q3, which drops their "resume" rank. The Rutgers win would indeed raise Nebraska's raw RPI, but my model obviously thinks the loss of a Q2 win would be more important. Subjectively, I agree with that, especially in Nebraska's situation where the lack of committee-defined "quality wins" is the gaping hole in their resumé. But overall I definitely agree none of this is to be taken seriously.
  2. Selection Committee Games of Interest

    Definitely a goofy result, mostly driven by the fundamental goofiness of RPI: KSU losing would drop Nebraska one spot in the projected RPI and projected resume rank (two of the five inputs to the model), which is enough to have that effect. RPI creates a lot of essentially random, chaotic interactions. Just from eyeballing this, I'd say anything under 3% leverage, maybe even 5%, is probably not "real"—but maybe still fun to have a rooting interest in some Atlantic Sun games.
  3. One thing about the "Teamcast" tool is that the projected efficiency rating (T-Rank) is not dynamic, so you should change it if you're gaming out scenarios that differ a lot from the expectations. E.g., Nebraska is currently 61st, and that weighs them down both because it effects the projections for unplayed games and because efficiency rating is part of the algorithm for predicting the tourney. The other thing is that it's not a crystal ball—it's very good at showing relative movement, but the absolute predictions obviously have a margin of error. This is especially true for a team like Nebraska that will have an unusual resume (if it keeps winning).
×