Still in Arizona & missed the below. I want to comment on a few points & have edited nate's post & numbered the points to make it easy to follow what I write. I don't think I've altered your meaning anywhere, nate -- but let me know if so.
nate33 wrote:1. I disagree with PIF's belief that Porter is actually better than Beal, and I categorically reject the notion that he was better than Beal in Beal's career year two years ago.
2. PIF's evaluation system overrates low-usage, high-efficiency players and does not fairly recognize the degree at which high-usage, moderate-efficiency guys make life easy for low-usage, high-efficiency players.
3. Box score numbers do not capture everything. They don't capture the way a high-usage offensive threat draws defensive attention.
On the first point: note the phrase I've put in italics. I don't believe in the idea of "actually better." I.e. a "better" that is independent of numbers. To put it another way (as I've written before): basketball is two things -- entertainment & competition. I'm writing about the competition part, which is judged by wins (obviously), & it's only numbers that affect wins. I don't believe that anything has any judgmental relevance except numbers. A player is "good" to the degree that he contributes to wins. He contributes to wins via his numbers. Period.
On the 2d point: what you wrote, nate, pertains only to scoring. I'm guessing this was unconscious, but it highlights a problem. Other things affect wins. For example, in both years Porter's rebounding was 50% above average for position, while in Beal's career year his rebounding was well below his career average & way way below average for a wing -- in fact, Beal has a career year only b/c of his shooting. It was a career year for him -- it wasn't a killer year for a SG (except in shooting!). Which is why it really wasn't such a high hurdle to be better than Brad in 2016-17.
Moreover, in 2017-18 Otto Porter was not a "low usage" player. His usage was above average at his position. He was a slightly below average usage in 2016-17.
One more comment on your 2d point: feel free to quantify for me "the degree at which high-usage, moderate-efficiency guys make life easy for low-usage, high-efficiency players." &, if it's not possible to quantify it, then maybe you are doing the "over-rating"!
. Plus, using category terms like "high-usage" & "low-usage" obscures the fact that these are numbers on a continuum not categories.
On the 3d point: box score statistics determine wins/losses 100%, as cannot be denied. Which means that on the team level they do capture everything. So your meaning must be slightly different -- i.e. that every box score number goes to an individual, but you question whether that fact really accounts for the interdependent activities that lead to one of those numbers going to an individual. A gets a rebound, it goes on his box score numbers, but maybe he got it b/c B boxed somebody out, which doesn't get captured by the box score
That's a fair point. 2 responses:
a. Running regressions using SAS or other statistical software does a lot to minimize the problem you point out. Numerous other kinds of investigation also help. For example, when a high-usage player shifts teams -- do low-usage high efficiency players on that new team start putting up statistically better numbers? People who have studied this can't find much to support the idea.
(In fact, people make these kinds of interdependency claims all the time -- & many of them are altogether contradictory! E.g. one person says "rookie A is only playing well because he's on a good team where he's getting excellent coaching & his teammates, being good, draw all the defensive attention." Then another person says about another rookie: "rookie B is only playing well because he's on a bad team where he has the playing time to show what he can do & to get better in real competition.")
b. You are right anyway: no analytic system is a replica of reality. You can't compare e.g. "PIF's evaluation system" to reality. You can only compare it to other such systems! & the only way to make that comparison meaningful is to look for correlation with wins/losses. Higher correlation = more useful. There is no more than that. Use the one that's most useful -- duh!
Edit: that last point I just made... that's why I stay away from the metaphysical "better than" judgments. You can't just take measurements -- how high do players jump -- then add them up & fully understand the world.
(Not to hijack our exchange to the issues discussed in the politics thread, but this is why IQ is a deficient measure & why immigration has to be understood in a much more complex way.)
