Working on a similar model myself (albeit without Synergy data for ease of updating) and one thing I will note is that my models loooooove Cam Boozer as #1 overall, so I’m very happy yours does too lol
Thank you for reading, and glad that you enjoyed it! Yeah, it's hard to find anything that Cam Boozer doesn't do well, and this model tends to favor that kind of player pretty highly at this stage in the process.
Really enjoyed this piece, especially the transparency around process. One thought I kept coming back to is that the model feels very strong at identifying college production, but it might reward a lot of high-usage, non-draftable profiles if scaled to the full college landscape. That tension actually made the Kornet discussion especially interesting, and I’m curious to see how the model evolves.
Thanks so much, glad you enjoyed the piece! Yeah, trying to make sure that the model doesn't overly reward stat sheet stuffers playing against lower-level competition is something that I'll be thinking about a lot with the model going forward.
That's also another reason why, in my mind, it's super-important to make sure to watch the film as well if you're trying to make a good model--you can parse out those high-usage, non-draftable prospects much more easily and effectively if you can see on the film that they're a "made for college ball" kind of player who probably won't succeed at the next level.
Very very enjoyable read!
Working on a similar model myself (albeit without Synergy data for ease of updating) and one thing I will note is that my models loooooove Cam Boozer as #1 overall, so I’m very happy yours does too lol
Thank you for reading, and glad that you enjoyed it! Yeah, it's hard to find anything that Cam Boozer doesn't do well, and this model tends to favor that kind of player pretty highly at this stage in the process.
What a write-up, awesome work! Glad to see that, as in all things, Mamu was the solution all along
Thanks so much, glad you enjoyed it! Also, yes, absolutely; I am very grateful for Mamu this morning.
Really enjoyed this piece, especially the transparency around process. One thought I kept coming back to is that the model feels very strong at identifying college production, but it might reward a lot of high-usage, non-draftable profiles if scaled to the full college landscape. That tension actually made the Kornet discussion especially interesting, and I’m curious to see how the model evolves.
Thanks so much, glad you enjoyed the piece! Yeah, trying to make sure that the model doesn't overly reward stat sheet stuffers playing against lower-level competition is something that I'll be thinking about a lot with the model going forward.
That's also another reason why, in my mind, it's super-important to make sure to watch the film as well if you're trying to make a good model--you can parse out those high-usage, non-draftable prospects much more easily and effectively if you can see on the film that they're a "made for college ball" kind of player who probably won't succeed at the next level.