Doctor MJ wrote:I find it hard to completely grasp where you're coming from though. Unless you though Durant had no business being considered a superstar before, him going through a considerable improvement puts him at such a high level it's weird to me you'd feel really any confidence about his limitations. I mean obviously the guys not going to lead the league in blocked shots any time soon, but the truly transcendant players are on a level where it's really hard to list out every way in which they have impact.
Yeah. Let's start over, because you're actually getting this a bit backwards with how I view Durant
Durant has been a bonafide MVP candidate since 2010, and the league's 2nd best player since '10 ('11 was actually a "down" year by his standards, but I'm not going season by season here). He also hasn't led the league in +/- in those seasons. I viewed him as a transcendent talent with or without him dominating the league in +/-, because I keep in mind all of the significant quirks that come with the important but NOT infallible +/- family. He's playing his best all-around season in '13 and some of this is certainly reflected by his high +/-, but if I wasn't holding seasons with lower +/- numbers against Durant I wasn't going to use them to boost his candidacy. I still factor them in, of course. I'm just also keeping other metrics and what I watch on the gametape in mind.
Re: Harden "steal" on/off Thunder. The whole reason why I'm using raw +/- data here is to render this issue moot. Harden on the Thunder had great raw +/-, he left, now Durant's +/- is far, far better than it was before. A particular causality should not be assumed, but by definition Harden's mere presence wouldn't make Durant's raw +/- seem unimpressive.
Right, which is why I edited my post to reflect this.
Re: More on +/-. I use the spectrum of +/- stats. The issues you talk about are virtually all dealt with when you add adjusted +/- into the mix. The one issue that remains...if it actually is a statistical issue...is that a player is going to have his +/- ratings affected by the quality of the fit. It's never been clear to me though that that is a bug. I see it as a feature.
Not "virtually". You can do a good job of filtering out things, but multicollinearity can muck up any +/- metric including RAPM; and at the end of the day the context of team dynamics, roster, and coaching is STILL important to factor in analysis. I'm not saying that this is a "bug", but simply that the stats aren't impervious to differences in context nor do they account for all of it.
Quite frankly his lack of reliance on on-ball dominance makes his game more portable than someone like LeBron.
On-ball/off-ball all depends on your roster. I want nothing to do with an off-ball player if my team is already stocked with shooters, but they aren't adept at putting the ball on the floor or setting up others.