Okay, why are we burying the lede here?
eminence wrote:OhayoKD wrote:That doesn't seem too shabby for a season most wouldn't have at the very top of MJs years.
The 97 Bulls were Jordan's
second best regular-season team, relevant when you are doing "on+on". Jordan being worse also does not necessitate his
situational value is lower when he is playing in a watered down-league. Most people would not have 97 as Jordan's peak, but
by the metric you're using, it should probably score
as one of his highest. All considered, him scoring
8th is not really positive evidence towards parity with Lebron. Nor is him ranking
11th in 98 after the Bulls posted their
5th best regular season. You are using what is partially a team-stat with Jordan's 2nd best and 5th best regular season teams(relative to the league). There is no reason to be treating these as off-years.
A) On-Off shouldn't really be compared between players on different teams (except as a measure of importance to their own team, but that's a largely meaningless measure on its own), it measures teamwide hierarchy (even more accurately positional hierarchy).
It measures the difference between a team's mov when a player is on and a player is off. It is noisy, but it is also all-inclusive. If you want to curve for team-quality you can, if you want to curve for positional backups, you can. If you want to weigh it less than other pieces of data, you can. But "it should be used here and not here" is not well-founded, and whatever merits on+on holds, it is not a substitute.
As is, neither metric helps Jordan. If you go with on+on/off, Jordan is scoring lower than nearly half of Lebron's career when the first "on" should be at it's highest. If you go with on/off, 2 presumed down-years at the back-end of his prime score right at the bottom. And that's not even getting into...
B) Large sample APM stats aren't particularly vulnerable to collinearity (it could theoretically happen, but lineups aren't generally strict enough that I'm ever worried about it).
They absolutely are, as Lamar Odom and the 09 Lakers can attest. Having your minutes tied with your best-teammates can spike it(96-98), having your minutes staggered with your best teammates can suppress it(12-14). And even, if, none of that is at play, historic outliers tend to see their value misattributed to supporting pieces. Adjustments and all, the premise of APM is still "winning on the court is good, as is seeing your team become worse without you on the court". One of the advantages of something like WOWY/Indirect is that you
can see what truly happens when a player is removed from a team. And when we use that for Jordan, with the largest possible samples(it doesn't get bigger 82), and the most favorable possible assumptions(oakley does not exist), Jordan does not grade out as a rival for Lebron. He doesn't even gain separation over someone like Hakeem(concentrated samples or extended). Mind you, that's just the
regular season. As it so happens, the two guys whose teams elevate the most post-merger are...Lebron and Jordan's draft-mate.
Actual APM also doesn't help mike but...partial samples so whatever. As of right now, the only real arguments for Jordan(rs-box excepted) is the assumption that there is, somewhere, better data that helps him, or reasons to disregard the evidence that actually is available and
doesn't favor him.
This is not a matter of "absence of evidence". This is a matter of some evidence favoring one guy, and virtually no evidence favoring the other. That is not considered "equality" almost anywhere else, but I guess basketball built different.
