The portal is officially open for business on Tuesday. I've been in search of a publicly available model that attempts to predict how successful a player would be in a different conference. I
found this from Kalidrafts (a data scientist named Nick Kalinowski). Kalidrafts focuses on box plus minus (BPM) and the models the likelihood a player would provide poor, below average, average, above average, or excellent BPM in a Power 6 or Mid-Major conference. The model also shares the teams it considers a "best fit" for the player, though I believe these are naive to coaching changes.
As of today, the site is still showing last year's players. I've very hopeful it will update on Tuesday, but it was interesting to review how the model viewed the most significant Gopher transfers from this season:
- Jaylen Crocker-Johnson - The model believed Crocker-Johnson would have an excellent BPM for a Power 6 team, Minnesota was not listed as a fit.
- Bobby Durkin - The model believed Durkin would have an excellent BPM for a Power 6 team, Minnesota was not listed as a fit.
- Cade Tyson - The model believed Tyson would have an average BPM for a Power 6 team, Minnesota was not listed as a fit.
- Langston Reynolds - The model believed Reynolds would have an above average BPM for a Power 6 team and Minnesota was listed as one his best fits.
The limitations of a data-only model are visible here. Given Tyson's lack of playing time at UNC, it was difficult to predict he could transfer to another high-major and thrive. However, I would have been even more enthusiastic about Reynolds if I'd looked at this model when he committed. Reynolds was probably considered the 6th or 7th most important transfer by most on this board, but perhaps we should have expected more before the season began.
Hopefully Willis and Vaihola are back for another year. Kalidrafts predicted Willis would be have an average BPM for a Power 6 team and Vaihola would have an above average BPM.
I'd love to know if anyone is aware of other publicly available prediction models. I'm sure there will be commits and targets that we don't know much about, it'd be great to have a few neutral, data-based resources to reference.
