Very intersting article, thanks for the deep-dive write-up of your model! Are your final scores in your model tiered by statistical significant differences (e.g. the drop off from Tyson/Tate/Lemon to Concepcion)?
Not necessarily, a player's "score" is just their percentile rank among every historical prospect the model's seen (e.g., an 80 is better than ~80% of WR prospects in the last decade or two). Tiers derived from that are used somewhat loosely, but since draft capital is such a huge part of our model, you can roughly map model Score to, say, buckets like a "mid-first" or an "early-second" grade.
Model's more forgiving of Tyson and Cooper; Ted Hurst also just a hair under the NGS >= 77 criterion from that other piece, but if we counted him he'd have cleared the bar. Both like Skyler Bell, model low on Chris Bell while last piece's metrics liked him.
Very intersting article, thanks for the deep-dive write-up of your model! Are your final scores in your model tiered by statistical significant differences (e.g. the drop off from Tyson/Tate/Lemon to Concepcion)?
Not necessarily, a player's "score" is just their percentile rank among every historical prospect the model's seen (e.g., an 80 is better than ~80% of WR prospects in the last decade or two). Tiers derived from that are used somewhat loosely, but since draft capital is such a huge part of our model, you can roughly map model Score to, say, buckets like a "mid-first" or an "early-second" grade.
This is such a great insightful article, thank you for sharing.
Thank you for reading!
What’s the overlap of likely hits in this model and the other you posted a couple weeks back?
Model's more forgiving of Tyson and Cooper; Ted Hurst also just a hair under the NGS >= 77 criterion from that other piece, but if we counted him he'd have cleared the bar. Both like Skyler Bell, model low on Chris Bell while last piece's metrics liked him.