from Amazon
Monday, March 01, 2010
Chone v2.0 - adding objective scouting data
Niiiiiiice. This is exactly the kind of place where you need to look to improve forecasts. What John Mayne has done here is to show that the observed past performance means more if you are a hard-tosser than a soft-tosser.
This can be an issue of selection bias: if you are a soft-tosser with a bad performance, you don’t get much chances. If you are a hard-tosser with a bad performance, you keep getting chances. So, when you look at the results, all the bad performances you have left above a minimum IP threshhold will likely be filled with hard-tossers.
So, what I would prefer is to look only at pitchers with an ERA under 4.50 in year X. This at least controls for quality in some respect.
Otherwise, fantastic work.