dhsilv2 wrote:scrabbarista wrote:dhsilv2 wrote:
You're creating what sounds like your own "winshare" but your test of it's accuracy is if it looks reasonable.
BUt it says show why you were so much higher than Hayes then most, because you place a huge value on the counting stats.
Fwiw, my old formula, which had less value on counting stats, was even higher on Hayes. His championship in '78 and two other Finals appearances were bigger factors in that formula. And my current formula gives him a big hit for being ranked so low on rgm, which my previous formula didn't. But that proves my point that a great player should be one that measures up to multiple standards. Hayes does. The only arguments against Hayes are "context" and efficiency, but I tend to ignore context (not because it doesn't matter, but because I believe that over the long run the numbers sort it out), and I prefer effectiveness over efficiency. Three Finals appearances, one championship, and mammoth career numbers constitute all of the effectiveness I need to overrule the many arguments I've seen against him.
I have no idea what you're trying to say in the first sentence you wrote. Every creator of advanced metrics that I'm aware of creates in exactly that way: start with a formula, then refine it toward coherence with past performances so that it attains an objective sheen and/or predictive value.
Hayes is also getting a huge boost from your metric because he played in a high pace era. If you're ok with that, then so be it. I think using counting stats is going to over rate that era.
If you used it for predictive power, do you think it would hold up?
He is missing, by a conservative estimate, about 1,640 blocks and steals from the first five years of his career due to those stats not being counted. That's just in the regular season. Then, I also have the era-adjustment, where he loses 4% of his total score (not because I believe that era was only 4% worse, but because with the all of the other disadvantages built into the formula, I think 4% is enough). Then, for the first nine years of his career, there were only three potential playoff rounds instead of four. Then there is the fact that I weigh championships won and Finals appearances by the number of teams in the league at the time, so that if there were 17 teams, for example, then he'd only get 57% of the credit that a player in 2018 will get. And don't forget that his rgm ranking is holding him back significantly in relation to all of the players around him on my list. It's not as though I'm merely counting up pts, etc. The bottom line on Hayes is that I really do believe he is one of the most underrated players in history. Numbers and stories are unlikely to change my mind on this, as I've seen so many already and still feel as I do. The two things that might have more impact would be either:
A) a sea change in my belief that guys who put up the most numbers (including WS, PER, etc.) on championship teams are far and away the most valuable guys in history, or
B) extensive, in-depth video study, so that my eyes would necessitate a major change in my formula.
I do actually agree with you that my formula may be a little biased toward high-pace guys with lots of minutes. That's why I intend to return to something similar to my old formula, which was a little more weighted toward efficiency stats, and less toward totals. [No players were punished for low efficiency, but there were more rewards for high efficiency. Thus, Hayes was actually higher on that list, but so were players like Manu Ginobili, Chris Paul, Larry Bird, and Yao Ming. My new formula won't be a copy of my old one, but it will, in theory, meet the criticism you've been making of my current formula.]
The predictive power I am going for here is the ability to tell us, when a player has already retired, where his career ranks in relationship to other retired players. If by "predictive power" you meant the ability to predict any player's future performance ever, then I absolutely do not believe in the predictive power of the formulas I've developed. But that hasn't been my goal. Rather, I just want a formula that will tell me, with a degree of accuracy that approximates my own independent observation - and, to a lesser extent, the observations of others - whether, for example, Jeff Hornacek had one of the 125 best NBA careers, or whether, for example, James Harden has already had one of the 50 best.