OPS vs. Total Average
Jim_...@transarc.com
has a nice chapter in there on total average as the best of all stats.
With it, he makes many of the same points as folks make here using
OPS: batting average is overrated; there are many players with
reasonable-looking BAs that aren't really hitting their weight; Pete
Rose's last year was even worse than it looked on the surface; and so
forth.
My question is: why is OPS better than total average? (I assume that
some people made this determination, which is why we rarely see total
average discussed here.)
******************************************************************
Jim Mann jm...@transarc.com
Transarc Corporation
The Gulf Tower, 707 Grant Street
Pittsburgh, PA 15219
(412) 338-4442
Sherri Nichols
>My question is: why is OPS better than total average? (I assume that
>some people made this determination, which is why we rarely see total
>average discussed here.)
I can't tell you from first principles why OPS is better than TA. From
first principles, you'd never add OBA and SLG together to begin with; the
units make no sense. But we measure the accuracy of an offensive stat by
how well it predicts actual team performance; that is, how well it predicts
the number of runs a team will actually score. There are a number of
offensive stats that correlate to team scoring better than TA: OPS,
OBA*SLG, Bill James' Runs Created, and Pete Palmer's Linear Weights are
among the better known. OPS gets used a lot around here because it's
pretty good, and simple to calculate. TA isn't a bad stat; it's just not
any simpler to calculate as some of the other stats which are more accurate
(it is simpler than Linear Weights).
Sherri Nichols
snic...@adobe.com
John C. Davenport
>I just picked up Tom Boswell's How Life Imitates the World Series. He
>has a nice chapter in there on total average as the best of all stats.
>With it, he makes many of the same points as folks make here using
>OPS: batting average is overrated; there are many players with
>reasonable-looking BAs that aren't really hitting their weight; Pete
>Rose's last year was even worse than it looked on the surface; and so
>forth.
>
>My question is: why is OPS better than total average? (I assume that
>some people made this determination, which is why we rarely see total
>average discussed here.)
>
You'll also notice that Boswell makes the claim without providing any
numbers to show how well it does.
It simply doesn't correlate with runs as well as OPS, or LW, or RC, or
EqA, probably because the weights are wrong: it overvalues home runs
and walks.
Lets see: 1-2 with a HR vs 1-2 with a single (since TA uses outs in a
denominator, I can't do 1-1) and 0-1 with a walk.
HR Single Walk HR/1B HR/BB 1b/BB
TA 4.000 1.000 1.000 4.000 4.000 1.000
OPS 2.500 1.000 0.500 2.500 5.000 2.000
RC 2.000 2.000 0.125 2.000 8.000 8.000
LW 1.220 0.220 0.080 5.545 15.250 2.750
EqR 1.413 0.426 0.163 3.317 8.669 2.613
A stat which says a walk is as good as a single is NOT going to be
highly accurate.
--
Clay D.
jc...@virginia.edu
A manager who'd bat Flynn and Blake ahead of Casey is an idiot.
Jim_...@transarc.com
OPS and total average for a few specific players.
Boswell does show some total averages for a few players, and, at least
for those he shows, it agrees very well with the consensus around here
of who was very good, who was overrated, and so forth.
From what people have posted, I tend to agree: OPS is better and as
easy to compute. Total average is, however, better than BA, RBI, or
many of the stats that show up in the newspaper.
I personally don't like RC, EqA, etc. all that much for one simple
reason: as accurate as they are, they are NOT very easy to see,
understand, estimate in your head, calculate on the back of an
envelope, etc. And I tend to like stats that I can look at and say
"OK, his OBP is .420. He gets on base 42% of the time, which is quite
good." (Yeah, this is probably a silly and unscientific way to decide
which stats I like.)
Sherri Nichols
>From what people have posted, I tend to agree: OPS is better and as
>easy to compute. Total average is, however, better than BA, RBI, or
>many of the stats that show up in the newspaper.
Yep.
>I personally don't like RC, EqA, etc. all that much for one simple
>reason: as accurate as they are, they are NOT very easy to see,
>understand, estimate in your head, calculate on the back of an
>envelope, etc.
If Runs Created fails this test, then so does Total Average. The simplest
form of Runs Created is
(Hits + walks)(Total Bases)/(At Bats + walks)
There is a somewhat more complicated version that takes into account stolen
bases and caught stealing.
Total Average is
(Total Bases + Steals + Walks + HBP)/(At Bats - Hits + Caught Stealing +
GIDP)
> And I tend to like stats that I can look at and say
>"OK, his OBP is .420. He gets on base 42% of the time, which is quite
>good." (Yeah, this is probably a silly and unscientific way to decide
>which stats I like.)
Nothing silly about it at all. Simplicity is perferctly reasonable as a
metric for evaluating stats. That's why you see OPS so much on this net.
Sherri Nichols
snic...@adobe.com
Roger Lustig
>OPS and total average for a few specific players.
>Boswell does show some total averages for a few players, and, at least
>for those he shows, it agrees very well with the consensus around here
>of who was very good, who was overrated, and so forth.
>From what people have posted, I tend to agree: OPS is better and as
>easy to compute. Total average is, however, better than BA, RBI, or
>many of the stats that show up in the newspaper.
Quite right. Another way of doing it is Bill James' Secondary Average,
which is (EB + SB + W)/AB. I know, the denominator is a problem,
but the point of this one is to produce a "batting average" for
all the things that don't show up in batting average.
The neat thing about 2A is that the league 2A is about the same as
the league BA. And the typical player at a given level of BA will
have a similar 2A. This is a great, quick-and-dirty way of sniffing
out players who are overlooked in the BA-based rankings, and of
deflating the Ralph Garrs and Bill Buckners. You see who's got
a monster .268 (Mike Schmidt) or a hollow .310 (Buckner).
If you want to see more TA numbers than you can shake a stick at,
try Total Baseball I and its 1990 Update. They have it as a
seasonal and career stat for all position players. TB II and III
carry it too, but only for top-5 per season and all-time leaders.
They also have RC, OPS (they call it Production), and the Linear
Weights stats.
Roger
John C. Davenport
>
>I personally don't like RC, EqA, etc. all that much for one simple
>reason: as accurate as they are, they are NOT very easy to see,
>understand, estimate in your head, calculate on the back of an
>envelope, etc. And I tend to like stats that I can look at and say
>"OK, his OBP is .420. He gets on base 42% of the time, which is quite
>good." (Yeah, this is probably a silly and unscientific way to decide
>which stats I like.)
>
Actually, EqA isn't too tough to estimate in your head. Are you
familiar with James' secondary average? Extra bases, plus walks, plus
steals, divided by atbats? Well, EqA in its earliest incarnation was
2/3 BA + 1/3 ScA = CA
I called it "Combined average" at first.
Values which James himself suggested ("batting average is about twice
as important as secondary average") Just add up the various components
and compare the number to hits. If its a lot higher, like it is for
Bonds or Thomas, their EqA is a lot higher than their BA. If its
similar to hits, their BA and EqA are about the same. If its a lot
lower than their hit total, like Polonia, EqA is less.
It still gets you in the right ballpark, although you'll be too high
(take away a sixth and you'll be better.)