EvanZ wrote:ElGee wrote:
Also, the numbers I'm citing are numbers, they aren't *me.* I'm just looking at what happens over the course of the games, in the same way one would tally FG% and points. What does a totally non-causal measurement, like APM, say about this?
Some of your stats are subjective. Specifically, defensive errors. All humans have subconscious (and often conscious) biases. A proper scientific experiment would be to blind yourself from the player, so you wouldn't know who you were tracking. Obviously, that is not feasible. Still, when you tally "blow-bys" and "missed rotations", that's subject to error. I can't imagine that if we got 5 guys in a room and watched a game that there would be universal agreement without having some discussion first. If you are claiming otherwise, I would be skeptical. If a Laker fan were to tally these stats, we might see different results than if a Miami or San Antonio fan did.
This is not to say I believe you are intentionally attributing more or less errors to any particular player. But it is very easy for humans to let even the most subtle biases influence the outcomes of their measurements.
I happen to be very acutely aware of these issues, as someone who is a scientist in academia doing research and supervising students. I make it very clear to them the issues with subjective vs. objective measurements and how we must fortify ourselves against these kinds of biases. It's not an insignificant problem.
Oh, as I've discussed before, not all of these things are black and white. There's human error too. But what I'm doing is no different than any scorekeeper around an NBA arena. Assists are subjective - there is a total gray area. I've seen games in which FGA's weren't even properly tallied. Heck, sometimes steals are called blocks and vice-versa. It's all part of the process, no doubt.
Simply put, my goal is to make every defensive error explainable to a neutral observer. I don't think there's a defensive error I don't watch multiple times to accomplish this goal. Does that mean it's perfect? Goodness no. Every once in a while there's a damn play I can't figure out myself (no error would be tallied), but in terms of anti-player biases in my scoring, I'm less concerned about that since most of the time the process is "how did a player get so open? who was guarding him? what's that guy's damn number? Oh it's 24."
EvanZ wrote:ElGee wrote:I can't tell if you're being condescending.
Not at all. Sorry if I came off that way.
NP.