Re: Basketball Win Shares?
- --- In APBR_analysis@yahoogroups.com, igor eduardo küpfer
> ----- Original Message -----
> From: Dean Oliver
> To: APBR_analysis@yahoogroups.com
> Sent: Thursday, August 28, 2003 11:51 AM
> Subject: [APBR_analysis] Re: Basketball Win Shares?
> Finding a formula for win% as a function of point differential isn't
> hard. What MikeT did is fine. I think it's equivalent to a couple
> of other things. The Pythagorean does about as well, but it ain't as
> convenient. Same with the Bell Curve method. Those both say that
> points don't have strict linear value in creating win%. That's more
> important in leagues outside the NBA, I think.
> I wasn't really aiming that high. I had an idea, that if an equation
>part without even looking at a player's numbers. My first pass got me
> Offensive Efficiency + Defensive Efficiency = Minutes Played
> yielded meaningful results, we could get a handle on the defense
nowhere, but using MikeT's suggestions I'm going to keep trying. I
know that there's a million other things that explain minutes --
cap-induced roster moves, injuries, doghouses -- so my standard for
"meaningful" results is pretty low. For now.
Having ditched work a little early, gotten some sleep, I think I can
speak. This idea is pretty interesting. MikeT mentioned Minkoff's
stuff, which I found really interesting when I saw it 10 years ago,
but I couldn't really figure out why. I couldn't figure out where to
go with it or how to get whatever was interesting out of it. But I
think you got it or at least started making me think about it.
Statistics can explain some amount of playing time. Other
explanations for playing time are non-statistical -- such as it's a
bad team that just needs someone to play (which may be controlled for)
and, what you're getting at, defense. It's got a chance. I'll be