RC/25 Introduction (6/18)
Nelson Lu
position. Here are a bit of definitions:
RC/25:
A stat created by Bill James to measure how many runs a lineup of 9
of the same individual would score in a game. (The "real" version is RC/27,
since there are 27 outs in a game; however, since certain stats are not readily
available to us during the season, James has an abbreviated version known as
RC/25 to adjust for the absence of those stats. For the formula to RC/27,
please e-mail me. It is also available in Total Baseball and many other
reference manuals.)
A = H + BB - CS
B = TB + .64 * SB + .24 * BB - .03 * K
C = AB + BB
RC = ((A + (2.4 * C)) * (B + (3 * C)) / (9 * C)) - (.9 * C)
O = AB - H + CS
RC/25 = RC / O * 25
Bill James has since revised the formula for RC. The formula that he currently
uses has somewhat different weights for BB, SB, and K. (The full formula
also has changes for the weights of HB, SH, and SF.) I have decided to adopt
these changes, as well as the change to measure marginal lineup effect; see
above for the modification. There are two other changes, however:
1. He adjusts for BA with runners on base. Originally, the adjustment was
not published; but it has been, in the most recent STATS Major League
Handbook and other sources. I am, however, not incorporating this
change since I have found this change to be arbitrary and also
increasing team-dependence.
2. He also adds additional points for a non-bases empty HR. Again, this
is highly arbitrary -- why not add points for a non-bases loaded
1B, 2B, or 3B? Further, this increases team-dependence as well, which
reduces the usefulness of RC.
PRC/25:
Park-adjusted RC/25. Park adjustment is a topic that I don't know how
to explain well in a paragraph, so I will not attempt to try. However, many
works that explain it are available on the net, and any search engine should
yield a number of useful articles. For these numbers, only this year's park
factors are used.
Value:
An estimate of how many runs a player is above the ever so elusive
"replacement level" (that is, a level of performance at which a player can
easily be found at little or no cost to a team). For these posts, replacement
level is estimated at 50 points of OBP and 50 points of SLG below the average
for each position.
Notes:
1. For these posts, the players who were traded during the middle of the
season have their park factors prorated by AB with the teams. (This is
not mathematically correct, but is the best I can do for now.)
2. Please send to me corrections if I misplaced someone's position.
3. As some have suggested, I am considering incorporating last year's
factors somehow into the mix. They are not included with *this* batch
of reports, however, as I am still pondering over whether and how to do
this.
"SABREmetrics isn't about statistics, it is about the search for new
evidence." - Bill James (thanks, Bill Reich)
===============================================================================
GO ANAHEIM ANGELS!
===============================================================================
Nelson Lu (n...@cs.stanford.edu)
John DiFool
> The following posts list major league teams' regular position players' RC/25 by
> position. Here are a bit of definitions:
>
> RC/25:
>
> A stat created by Bill James to measure how many runs a lineup of 9
> of the same individual would score in a game. (The "real" version is RC/27,
> since there are 27 outs in a game; however, since certain stats are not readily
> available to us during the season, James has an abbreviated version known as
> RC/25 to adjust for the absence of those stats. For the formula to RC/27,
> please e-mail me. It is also available in Total Baseball and many other
> reference manuals.)
>
Since several sites DO have most of the secondary stats (The Official
Site of ML Baseball has everything but IW, which has a miniscule effect
on the results if omitted), why haven't you gone to the RC/27 formula?
John DiFool
Nelson Lu
John DiFool <jdi...@earthlink.net> wrote:
>
> Since several sites DO have most of the secondary stats (The Official
>Site of ML Baseball has everything but IW, which has a miniscule effect
>on the results if omitted), why haven't you gone to the RC/27 formula?
Well, I am not totally sure if the effect will be miniscule; without a
confirmation that would in fact be the case, I felt that the additional data
may introduce unexpected results into it, and that's why I haven't done it. I
am willing to reconsider the issue, however, and I am in fact doing so.
In any case, though, the data also has to be in a useful form -- for example,
if the additional secondary stats are on individual player stat/splits pages,
then they are way too time consuming for me to track down.
Samson
(Nelson Lu) wrote:
> In article <3D0FCA07...@earthlink.net>,
> John DiFool <jdi...@earthlink.net> wrote:
> >
> >Since several sites DO have most of the secondary stats (The
> >Official Site of ML Baseball has everything but IW, which
> >has a miniscule effect on the results if omitted), why haven't
> >you gone to the RC/27 formula?
>
> Well, I am not totally sure if the effect will be miniscule;
> without a confirmation that would in fact be the case, I felt that
> the additional data may introduce unexpected results into it, and
> that's why I haven't done it. I am willing to reconsider the
> issue, however, and I am in fact doing so.
Even if the effect of IBB on team runs isn't miniscule, one
could argue that it should be held constant when comparing
individuals. I'm not sure if omitting it is the same as
holding it constant, though. But why omit HBP? That's
certainly available.
> In any case, though, the data also has to be in a useful form --
> for example, if the additional secondary stats are on individual
> player stat/splits pages, then they are way too time consuming for
> me to track down.
I'm pretty sure cnnsi has GDP in the form you're looking
for, which could bump outs/game up to ~26. The number isn't
all that important unless you use James's new, improved
player-outs/games/league-outs/something-or-other value in
place of 27. Are you currently using an adapted
version for the denominator of RC/25, or do you just
use 25?
Nelson Lu
Samson <sams...@hotmail.com> wrote:
>In article <aeoic2$3lo$1...@usenet.Stanford.EDU>, n...@Xenon.Stanford.EDU
>(Nelson Lu) wrote:
>> In any case, though, the data also has to be in a useful form --
>> for example, if the additional secondary stats are on individual
>> player stat/splits pages, then they are way too time consuming for
>> me to track down.
>
>I'm pretty sure cnnsi has GDP in the form you're looking
>for, which could bump outs/game up to ~26. The number isn't
I'll take a look, but again, adopting a secondary stat without adopting the
others may produce unexpected results. I'll consider it, though.
>all that important unless you use James's new, improved
>player-outs/games/league-outs/something-or-other value in
>place of 27. Are you currently using an adapted
>version for the denominator of RC/25, or do you just
>use 25?
I'm using straight up 25 right now.