Sunday, January 18, 2015
Inside Edge Index
?Jeff compares all the players based on the Inside Edge "zones", where the zone is not physical, but rather, what the scorer thinks the fielder should have done. I'm ok with what IE is trying to do. Like everything, it's an estimate. And of course, it'll have its bias.
I don't think what Jeff did was a good choice, by making it an index, rather than making it plus/minus, like UZR. (Mostly, it's because the index metric itself will be biased. Making 7% of the plays if the average is 5% is the same ratio as making a play 70% of the time in the "50% zone".) And if you did differential, then you wouldn't need to throw any data out.
Also, he used the same mean across the positions, rather than position-by-position means.
Anyway, the interesting thing is that he shows a correlation of r=.50 when the number of these plays was around 70, which looks like it's about two-thirds or three-fourths of a season. Which is pretty close to what UZR's r=.50 point is.