After going through your site some more, I must say how impressed I am. A couple of further comments. ... From: roland_beechOct 9, 2003 1 of 5View SourceAfter going through your site some more, I must say how impressed I am. A
couple of further comments.
----- Original Message -----
From: "roland_beech" <roland@...>
Sent: Thursday, October 09, 2003 8:27 PM
Subject: [APBR_analysis] Re: NBA Stats site
> The plus/minus stuff is just one facet of what we are doing. I like
> the +/- in conjunction with player W-L numbers (each game is scored
> as a win or loss on the player's plus minus data), which helps show
> consistency of performance.
Perhaps the Pythag would be informative here. Using Tim Duncan as an
OffPts DefPts Min Win% W-L
OnCourt 97.0 88.9 3151 75.6% 50-16
OffCourt 91.7 97.9 775 29.9% 5-11
50-16 is a number people can easily understand, while points scored and
allowed appeals only to us statheads :-)
BTW DeanO: how do your individual win numbers match up to these?
BTW2: Roland, nba.com shows Duncan as playing 3181 minutes. Typo, or is
there a rounding error? Not that a 1% error is a big deal.
> The 5-man units are often undermined by limited data (a unit that
> plays 2% of the total minutes for a team may only be playing 78
> minutes total together), but we can also break out the numbers by
> two-man pairs, etc to understand which players play well together.
The 5-man units provide enough information for now -- I could spend hours
looking at them! I would suggest showing the floor time as absolute minutes,
as well as the percentage of total team minutes.
> As far as Ed's comment on clutch shooting stats are concerned, we
> certainly could publish the actual numbers instead of the per 48
> minute basis summaries. I don't like the format on several of the
> pages that much, but trying to get 2,500 pages of content (5 per
> player, 6 per team, with articles) published in a timely fashion
> called for some compromises in the short term.
I'll bet. It's obvious how much work that must have gone into the project.
> It's likely the layout will change on certain fronts for the coming
> 03-04 season.
I have a few other suggestions. I'll send them off list.
... In http://www.rawbw.com/~deano/articles/iwldef.html I post 3 versions of individual win-loss records. The version that Roland posts -- how many games thatOct 10, 2003 1 of 5View Source--- In APBR_analysis@yahoogroups.com, igor eduardo küpfer
> Perhaps the Pythag would be informative here. Using Tim Duncan as an
> OffPts DefPts Min Win% W-L
> OnCourt 97.0 88.9 3151 75.6% 50-16
> OffCourt 91.7 97.9 775 29.9% 5-11
> 50-16 is a number people can easily understand, while points scored and
> allowed appeals only to us statheads :-)
> BTW DeanO: how do your individual win numbers match up to these?
I post 3 versions of individual win-loss records. The version that
Roland posts -- how many games that person's team outscores the
opponent when the player is on the court -- is most like the first
listed there, I believe because players on good teams generally look
pretty good. Roland's is better than what my first one does. Method
2 shown on my page is one of my favorite tools because it indicates
how well players do their roles. For some players, I think it is
similar to Roland's, especially the better players (Duncan 58-23 by
Roland and 62-19 by me). Method 3 of mine is applying Pythagorean
methods to individual offensive and defensive ratings, which is more
answering the question you raise. It looks, however, to estimate how
many games a player contributes to, so how many wins can they add.
Duncan comes out at 15.9-1.0 (.941), so it's not comparable in
I think it's key to note that all these interpretations of an
individual's win-loss record are fair. The one Ed coins is
interesting. The one that Roland has should be similar to Ed's
(theoretically, they should be very comparable), but I haven't checked.
>That data set could lead to a flood of ideas on what to do with it.
> I have a few other suggestions. I'll send them off list.
How many years of pbp's do you have?