Quote:
Originally Posted by jcard
Just curious: given that the projections are usually of actual statistical production and therefore include the influence of park factors, did you pursue and/ or find any methods or cases where adjusting for this yielded better and fairer derived ratings?
|
Great question.
The roster ratings are a blend of both Steamer and ZIPs projections. How ballparks come in to play….. I try to be cognizant of the extremes. For instance, Great American Ballpark is a launching pad for home runs, and a Flyball pitcher like Hunter Greene is projected to give up 1.5 HR/9 per ZIPs. I take his ballpark in to account and “boost up” his movement rating to try and compensate so he is not “double penalized”. His OOTP ratings will probably show something like 1.1 (or so) in HR/9, and he will probably ultimately in a season give up 1.3-1.5 HR/9 based on his ballpark. Hopefully that makes sense?
So again I import projections for thousands of players, and then try to calibrate ballpark extremes so the player is not either double rewarded or penalized.
It’s not perfect, but I’ve been very pleased with the statistical outcomes both macro and micro for the 2024 season.