Originally Posted by
USAFChief
I've never understood why people accept that defensive metrics in small doses are "basically useless," but if you add up enough useless info, it then becomes useful and accurate.
For hitting data, people understand that small sample sizes are often unsustainable. But nobody says "it's a small sample size, Plouffe didn't really hit seven HRs in a week.". They say "he did hit those HRs, but I doubt he'll keep ii up."
But with defensive metrics, that's exactly what people say. Even the "inventor" of UZR says exactly that: "I know UZR says Delmon Young was a plus defender for the first half, but he really wasn't. What we need to do is take a whole bunch of this unreliable data, so it becomes reliable."
If the math is wrong in small sizes, adding up a bunch of wrong samples doesn't somehow make it right.