twabby
Registered User
- Mar 9, 2010
- 14,165
- 15,701
I appreciate your approach, and while I am an old school eye test guy when it comes to hockey, I get the value in analytics and believe there is a place for them.
I do have a question to add, specifically the % chance of becoming a Star metric. Do you have the placards, or can you get the percentage of being a star values this algorithm produces for all the Caps we consider stars just before they were drafted? Can the vales be backward calculated?
I would especially like the see how that metric assessed Wilson, Oshie, Carlson, Orpik, etc.
I my field, verification of model prognostication is really important.
Oshie and Orpik fall outside of the window. It only goes back to 2007.
Carlson was roughly 70% to make the NHL, but 0% to be a star. I suspect that this was some sort of error importing data into Tableau because I’ve never seen anything approaching a 0%/70% star/NHL split.
Wilson was 1% star/9% NHLer.
Full data here, up until 2021:
This is for the TopDownHockey model.