Owly wrote:Perhaps this wasn't clear so I'll restate. The question wasn't about whether your -RP (Replacement Player) metric does have a replacement level value. It is about whether replacement level is the "best" or most appropriate baseline (you seem to be addressing this later, but initially about the metric so hopefully this clarifies - if indeed there was confusion).
(Is it, for instance a plausible estimate for what it's intended to be as a replacement level [is that the level of a back-up center that comes in for such a guy]? At what numerical point does should a player start adding value with any minutes that moves them toward this conversation?)......
....And having thought on that, more broadly and for the other tools in that family is "replacement level" the right level?.....
I'm not saying it isn't, and certainly not confident in any answers, just giving the questions that came to my head.
I guess we could deliberate on a more appropriate baseline (if indeed you think there is a more appropriate one). As I've written it, it kind of runs on the assumption that when a starter sits, he is going to be "replaced" by a "replacement level" player. Obviously, that is not true in all instances; but where the differences [in who's replacing him] lie is in luck of the draw (in the quality of your supporting cast). And I didn't necessarily feel "above average" was appropriate, because many [most] teams are not lucky enough to replace each and every starter with a league-average player. I mean, it's somewhat the balance of starter-level players to replacement level players that creates the "average" NBA player, no?
That's why I went with an estimation of a replacement level guy. At any rate, I certainly feel a level playing field is appropriate.
As to whether a typical replacement level center is indeed a -0.75 (or something nearby that figure) DRAPM.....again, I'm not sure. My hunch is this figure [the DRAPM of an average replacement level player] varies by position; I just haven't investigated it to know for sure, or to know what a more accurate figure for each of the five positions would be. So for now I stuck with a uniform figure for all five positions (and I feel the -1.5 total RAPM assumption is probably fairly close to the typical replacement level player).
That said, if I shifted the baseline up or down by 0.5 or 0.8 or similar, it wouldn't much change the hierarchy or rank of the players listed in the above figures (as long as we're using that same baseline for each center looked at): the exact numbers would change, but the players' positions in relation to each other would [mostly, or possibly entirely] remain the same.
Owly wrote:Nor is this about your broader case for Kareem the array of stats you use. But the idea was mooted that Eaton as a consistent "+4" defender wouldn't top Jabbar with Jabbar posting [hypothetical] numbers mostly above average but likely below positional starter average from 83-89.
To be clear, I didn't necessarily mean that statement to be the "door definitively shutting on Mark Eaton" clincher. Just stating the situation as it appears to be.
We can quibble about the hypothetical figures I'd suggested for Kareem's career, but I don't feel I've been liberal or overly generous with the figures I suggested (if you disagree, feel free to point out where).
But here's the thing (just as a "for instance"):
We could lower each and every one of Kareem's single-year DRAPM estimates I'd suggested by 0.25---->this will leave him with THREE seasons where he's a very small negative as a defender, two other seasons where basically a neutral (+0.05), FOUR other seasons in which he's a positive defender but < +1, a career peak DRAPM of "only" +2.75, and total of only two seasons where he's > 2.5.........maybe I'm mistaken, but I suspect such figures would understate Kareem's career defensive impact. But for the sake of this example, we'll still use them.
We could then allow Mark Eaton a career minute-weigthed DRAPM average of +4.25 (likely OVERstating things, considering this is full career [including his rookie season [where he's averaging 6.1 PF/36 minutes] and his post-prime]; but again we'll run with it).
Kareem would STILL come out a little ahead of Eaton in cumulative DVOR (using the -0.75 baseline).
Again, not that it's all about the cumulative (I didn't mean to imply that). I was merely making the statement of "fact" (or as close to "fact" as we can come with reasonable estimates) that there isn't anyone left on the table who can top Kareem in terms of cumulative DVOR (with the baseline currently being used). Of the players I have run the numbers for who ARE still available candidates, the next closest is Mourning, and he's >50,000 points behind. We can make some pretty substantial reductions to Kareem's estimates, and he'd still stay WELL ahead of Zo.
I brought up Mark Eaton because he seemed like the only center who's numbers I HADN'T yet run who might have a small chance of coming close to Kareem's career total. But as illustrated above, it's very unlikely he could match or top Kareem's cumulative career.
Yes, Kareem's is spread out over significantly more years (never said otherwise). Was merely stating very directly: he's tops in career cumulative among all candidates who are left (for whatever that is or is not worth to you).
Owly wrote:Interesting output on the hypothetical longevity monster - disregarding for the moment that you if there were any "bug i n the system, you've got it covered by the various other tools - what does that guy bordering on Bradley make you think (should they be in that vicinity....
"Should" doesn't really figure into it. That simply IS. That IS where a league-avg (+/- 0) defender who played 90,000 career minutes would be in cumulative career DVOR (using our current baseline).
As I have hopefully impressed by now, though, that ranking in cumulative DVOR does NOT fully dictate where a player "should" [to me] rank defensively for the purposes of this project. It's merely one figure or indicator among many.
Note, for example, that this same hypothetical player would have [by FAR] the lowest avg DVOR per season or game (for both career or best 5-years) of ANYONE listed.
Owly wrote: and what would be the outcome at +1 - a guy, if I interpreted right, you still wouldn't consider)?
A guy who was consistently a +1 DRAPM for 30 years/90k minutes would have a cumulative career DVOR of 157,500 (higher than everyone except Hakeem and Russell) at the current baseline used. However, he'd have an average per season (career) of 5,250 (more toward the bottom, only ahead of Chandler, Ratliff, M.Gasol, and Camby, among those I've run); his avg in his best 5 years (per season or per game) would be below [by a handy margin] EVERYONE run so far.
Such a hypothetical character I suppose I'd give consideration somewhere in the 11-15 range or so??? idk, it's such an unrealistic hypothetical it's hard to say. Probably not a top 10 figure for me, though. But being a small positive for such an inhumanly long period of time (3,000 minutes per season for three decades) would certainly have me looking at him shortly after the top 10, I would guess.
