Do they publish evaluation metrics for their models? It seems like they vastly over-estimate both nhler and star probabilities
Me make big points, me nhl probablyDo they publish evaluation metrics for their models? It seems like they vastly over-estimate both nhler and star probabilities
Yup, it's age, league, and points. So a guy projecting well on the model just means he's scoring points well in his league relative to other guys his age, NHLe adjusted.isn't bader's model just super simple and based on historical PPG standardized to NHLe?
Both Makar and Fox are listed as 2 inches taller which believe or not does make a big difference. Cagnoni might as well be Pavelski when compared to Hughes and Makar’s skating.But since 2019...Fox, Hughes, and Makar have unquestionably elite. It may be a sign of changing times...
Those are also the only 3 examples anyone can name and frankly Fox and Hughes are overrated losers. The only other sub-6ft defenseman in the league are third pairing offensemen like Girard, Hutson and Gostisbehere.Both Makar and Fox are listed as 2 inches taller which believe or not does make a big difference. Cagnoni might as well be Pavelski when compared to Hughes and Makar’s skating.
Cagnoni is nothing like any of those 3.
It seems like the model would be pretty easy to evaluate. E.g., fit the parameters on 2012-2018, report accuracy on 2019 players (repeat for all years). But I am not sure they do that, it seems mostly vibe-basedYup, it's age, league, and points. So a guy projecting well on the model just means he's scoring points well in his league relative to other guys his age, NHLe adjusted.
Short hand for points good. Which overall, isn't totally incorrect.
There is value in a model like that when it identifies late round candidates that need to be scouted more. You just can't take it at face value.
third pairing offensemen like Girard, Hutson and Gostisbehere.