Diaper Dandy wrote:-= original quote snipped =-
That's all well and good, but if we're only looking at one stat, one stat alone, to tell you how good a basketball player is compared to someone else, it's easily the winner.
What? You're gonna go off ppg by itself? True shooting by itself? And you think that's gonna work better than PER? I mean, you said rebound rate, and looking at that alone tells you that Jayson Williams is a better player than Charles Barkley.
My interpretation of the question is: If you can only use/look at/whatever one stat to assess a player, which would it be? PER is one a whole different level there than rebound rate. Roland rating is decent as well.
I'm not going to use
any single stat to assess players. I think the idea is ridiculous. And I don't think it's what the thread is asking.
coolcono wrote:
MVS... Most valuable statistic
I was wondering what everyone thought the best statistic for a player to have is. My vote goes to the blocked shot. The obvious answer (of points) is derailed with the intimidation and presence of the blocked shot. Others' thoughts?
Does that really sound like it was intending to ask about a stat that judges a player's overall ability? By brining up blocked shots? It sure doesn't to me. The best stat for me, again, is rebound rate. It doesn't tell you a huge amount...but what it says, it says accurately. It adjusts for court time, era differences...pretty much everything.
With respect to PER, Roland Rating, etc. ... look, I could come up with an analytic tool right now. I just did! It looks like this:
(15*ppm)+(5*rr%)+(35*apm)+spg+bpg-tpg)*(mpg/100)*(% of team games played)*(ts%)
That took about 90 seconds to think of and write down. I'm sure it's a long way from perfect. But, on the other hand, I think if you applied to all the players in the NBA, you'd probably get a good idea of the top 20 players. It would probably give you answers similar to PER. See, that's why analytic tools fall down. They tell us what we already know. With PER...well, here are the top 10 players in PER right now.
LeBron
Paul
Amare
Bosh
Garnett
Ginobili
Bryant
Howard
Duncan
Dirk
Is there anybody that even casually follows basketball that
wouldn't know that those are great players? The fact that you can state or show what's obvious doesn't mean
anything. It's kind of like saying, "Well, because I knew the sun would rise today, I can tell you where Mars and Earth are in relation to each other in their orbits." Being able to "prove" something obvious doesn't mean you have the ability to "prove" something more difficult and complicated to determine. And that's where statistical analysis falls flat. PER is zeroed to 15, so the "average" player rates a 15. Which means that Danny Granger is kinda average and Al Horford has been lousy this year. Rashard Lewis has been pretty run of the mill too; McDyess has been a below average player, and Nocioni is lousy. Don't play McCants; he's below average. Raymond Felton and Kirk Hinrich are worse than that--you'd be better off with Keyon Dooling. When you've got a statistical analysis that can only tell you what you already know, and screws up pretty often on more difficult questions, I call it "not useful."
Rebound rate does only tell you one thing--but it always gets it right. It says Jayson Williams is a better rebounder than Charles Barkley because, when they were on the court, Jayson Williams got a higher percentage of rebounds. Of course, Barkley was a better player...but rebound rate is only intended to do one thing. It tells you what percentage of rebounds you can expect a player to get, regardless of pace, regardless of era. I'd much rather have a smaller, more effective analytic tool than a big flashy one that largely gives me crap.