Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site decwrl.UUCP
Path: utzoo!linus!philabs!prls!amdimage!amdcad!decwrl!schneider@vlnvax.DEC
From: schneider@vlnvax.DEC
Newsgroups: net.sport.baseball
Subject: Re: Statistics and Carter vs. Pena
Message-ID: <3655@decwrl.UUCP>
Date: Thu, 15-Aug-85 10:52:49 EDT
Article-I.D.: decwrl.3655
Posted: Thu Aug 15 10:52:49 1985
Date-Received: Mon, 19-Aug-85 20:55:33 EDT
Sender: daemon@decwrl.UUCP
Organization: Digital Equipment Corp., Palo Alto, CA
Lines: 112

<> ---> I don't know who wrote this
>  ---> David Rubin	
   ---> personal pearls of wisdom

<> In regards to the David Rubin mega-comments:
<> 
<> One thing I don't see (in Pena-Carter Arguments 1985) is that to be
<> a legitimate statistic, it cannot reflect ANY teammate interaction.
 
>All statistics reflect, to some extent, teammate action.  Some,
>however (such as R's and RBI's) are positively DOMINATED by it.
>The best we can do is to rid ourselves of statistics which are
>directly influenced by teammate actions; we cannot rid ourselves of
>indirect infuences.  So I just did the best we could...

We can do much better than to rid ourselves of such statistics.  Yes
there are those who have developed further statistics which better
capture a players offensive production, but that does not render the
lesser stats meaningless, deserving of expulsion.  And besides, the
Globe does not provide daily statistics of who's leading the league
in runs created.

Also there exists methods by which runs and ribbies can be isolated
somewhat by teammates offense, or those that would yield important
results when seen as a per-game number and compared to others.  For
instance, how many runs per game does the lead-off hitter for the
Yankees score when its Rickey Henderson?  How many when its not?

(One of the things I've longed imagined is to have all baseball statistics
on-line.  I mean ALL, meaning basically a score sheet of each game.
Wouldn't it be nice then if they were in a huge database, ready to be
compiled into new statistics with a few simple instructions.  The
Sabrematicians would certainly get a run for their money.  Bill James
uses a PC, but seems to regard it a little as the enemy and a little
as a simpleton.  I imagine he (and others) probably, with the aid
of whatever staff they have, tabulate by going through the score sheets
by hand, etc.)

My point is that it would be nice to live with more relevant statistics,
but while we can only read them roughly once a year, they are difficult
to accept and understand.  The established ones have a huge lead that
can only be limited when new stats are readily available.

Has anyone here seen the Bill James Newsletter?  What's it like?
 
<> I don't buy that.  Mr. Rubin states that for example, RBI's, Runs, 
<> and Batting average are somewhat meaningless but on base percentage
<> and slugging percentage aren't.
 
>Runs and rbi's are almost meaningless (that I said) in judging
>individual performance; batting average I merely regard as poorly
>conceived.  It was established as a hitting statistic when power was
>unimportant and walks exceedingly rare, i.e. under circumstances that
>no longer hold.
 
Rather than poorly conceived, the batting average works very well to measure
something which is not as important as the general conception.  But it
does offer its own one-dimensional view of offense with no error whatsoever
(well, not counting the official scorer or umpire).  This is a useful
tool when isolated, and useful in comparison.  I'm somewhat skeptical
of the BA's history as offered here.  I think its just a conveniently
compiled statistic which works to a great degree.

...

>Intentional walks should be subtracted off, but there are not so
>frequent as to really screw up OBA.  OBA isn't perfect, just superior.

BA and OBA measure different aspects of the game.  Inherently they are
correlated, but neither can do justice in explaining a player's
offensive contribution.  The problem with OBA is that a walk is a
single is a catcher's interference.  But obviously a single is BETTER
than a walk.  OBA isn't perfect, nor superior.  As statistics go, they
both do their (different) jobs.


>                          ... however, OBA and SA both far outperform
>any of the other standard stats in predicting run production (should
>I post a summary on Pete Palmer's study of the issue?), and together,
>they do VERY well, indeed.

In tandem, I surmise, OBA and SA can well approximate a players offensive
production.  That doesn't obliterate the usefulness of other stats, namely
the ones derided here.  I would be very interested in knowing what Pete 
Palmer has to say.  I would note that Bill James has developed his own
Runs Created stat.  In the past he has given some evidence of its
superiority to Palmer's stat, but to show his objectivity this year he
published someone else's stat which outperforms his to a large extent.
I don't subscribe to everything James has to say, but as far as I know,
he's the best at what he does.
 
<> I personally would rather have a Carter than a Pena, but even with
<> all the stats produced, I don't think anyone could give a DEFINITIVE
<> answer to who is better.
 
>Au contraire -- the evidence of an offensive difference is
>overwhelming.  If you are not persuaded that this is so, there is
>probably nothing that could persuade you.  We'd be left without a
>decent way to carry on an argument beyond the level of SEZ WHO; now
>what fun would that be?
 
Ahhh, the crux of the issue.  I do believe that Carter is offensively
quite superior to Pena.  Defensively, I can't say because I almost
never see Pena play (I know there are stats;  I don't know their tallies
though, and I'm not sure if they can begin to measure what a catcher does).
From most accounts both are among the premeire defensive catchers in the
game.  But Pena has quite a few years on Carter, giving him the edge
in current value in my book.  It depends on the team, but in general I'll
take the younger Pena.

		Daniel Schneider
		{decvax}!decwrl!dec-rhea!dec-vlnvax!schneider