BBBA question for Don Malcom
Steve Cox
article on the site slams him. It's not my opinion that his
work merits this sort of sneering treatment, though I'll grant that
the mistakes you point out in his column are in fact mistakes.
Neyer still seems to be miles above almost every other mainstream
columnist. I don't know what's going on, but it looks as if there's
some BP/BBBA element in your ill feeling towards him. There's also
no doubt some very clever reason why you keep misspelling his
name.
A lot of the work on the BBBA site is very interesting and worthwhile,
but the onesided flamewar is pretty tiresome.
-Steve Cox
Keri Olsen and Arne Olson
Scott Cantor wrote:
>
> In article <19990518102315...@ng-cg1.aol.com>, nawr...@aol.com.not (NawrockiT) suggested:
> >
> >My mistake: I see now that the piece was by Kenn Adams, not Donn Malcolm. My
> >apologies.
>
> Right, Tomm. :-)
>
> Seriously, the one valid point I did think the article made was that there
> probably should be a bit more study done to validate some of the conclusions
> being reached vis-a-vis the PAP totals. It's a good first try, but accepting
> it at face value is probably a little premature at this point.
>
> Finding that point buried in all the snarky invective was an unpleasant
> challenge, however.
The rhetoric always appears to get turned up when there's a competing
product at BBBA. In this case, there appears to be a concerted attempt
to use all possible means to undercut PAP in favor of the three-start
model that Malcolm introduced earlier this year. That article wasn't
pretty either, essentially accusing Rany Jazayerli of being dishonest
and disingenuous.
Arne
NawrockiT
:: What's up with the war on Rob Neyer? Lately almost every new
:: article on the site slams him. It's not my opinion that his
:: work merits this sort of sneering treatment, though I'll grant that
:: the mistakes you point out in his column are in fact mistakes.
:: Neyer still seems to be miles above almost every other mainstream
:: columnist. I don't know what's going on, but it looks as if there's
:: some BP/BBBA element in your ill feeling towards him. There's also
:: no doubt some very clever reason why you keep misspelling his
:: name.
: I just took a look at Don's (or should that be Donn's?) angry piece about
: Neyer's criticism of Bobby Cox for letting Odalis Perez throw 121 pitches in
a
: game last month. It's a lot like his defense of Sandy Koufax here on rsbb,
: where he decided that people were criticizing Koufax for not being manly
enough
: to pitch through the tears or something like that, even though no one had
said
: anything close to that.
My mistake: I see now that the piece was by Kenn Adams, not Donn Malcolm. My
apologies.
Tom Nawrocki
NawrockiT
: article on the site slams him. It's not my opinion that his
: work merits this sort of sneering treatment, though I'll grant that
: the mistakes you point out in his column are in fact mistakes.
: Neyer still seems to be miles above almost every other mainstream
: columnist. I don't know what's going on, but it looks as if there's
: some BP/BBBA element in your ill feeling towards him. There's also
: no doubt some very clever reason why you keep misspelling his
: name.
I just took a look at Don's (or should that be Donn's?) angry piece about
Neyer's criticism of Bobby Cox for letting Odalis Perez throw 121 pitches in a
game last month. It's a lot like his defense of Sandy Koufax here on rsbb,
where he decided that people were criticizing Koufax for not being manly enough
to pitch through the tears or something like that, even though no one had said
anything close to that.
Neyer's sidebar was very fact-laden, and the only pejorative thrown at Cox was
to call him an "abuser." But Don goes on to write "One start over 120 pitches
labels a
manager as a criminal, or worse." Of course, Neyer didn't call him a criminal,
much less anything worse.
Calm down, Don. As long as you do good work, people will recognize that. It's
not necessary for you to attack the other writers in the field. And if you do
attack them, it always helps your case to stick to the facts and not inflate
the other guy's sins.
Tom Nawrocki
basil
I think that was the point. Correct me if I'm wrong, but I took it as
one straw man slaying another straw man. Hyperbolic, if anything.
>
> Calm down, Don. As long as you do good work, people will recognize
that. It's
> not necessary for you to attack the other writers in the field. And if
you do
> attack them, it always helps your case to stick to the facts and not
inflate
> the other guy's sins.
Well, you do seem to have a point here. I'm probably missing the origin
of BBBA's issues with Neyer (like "why the extra b?"), but it's not only
Malcolm and Adams; C.O. Jones rips him a good bit in his weekly Foul
Tips. Maybe they don't respect Neyer's research methods (all the PAP
crap), maybe they think he's playing both sides (stathead and
mainstream), maybe they don't like his name--I don't know, but at times
they sound as sophomoric as Gammons does when he calls Duquette a
rotisserie foof.
>
> Tom Nawrocki
>
>
--
William "Basil" Tsimpris btsi...@fec.gov
--== Sent via Deja.com http://www.deja.com/ ==--
---Share what you know. Learn what you don't.---
dtk
>
> Well, you do seem to have a point here. I'm probably missing the origin
> of BBBA's issues with Neyer (like "why the extra b?"), but it's not only
> Malcolm and Adams; C.O. Jones rips him a good bit in his weekly Foul
> Tips. Maybe they don't respect Neyer's research methods (all the PAP
> crap), maybe they think he's playing both sides (stathead and
> mainstream), maybe they don't like his name--I don't know, but at times
> they sound as sophomoric as Gammons does when he calls Duquette a
> rotisserie foof.
I believe it's called jealousy--"Robb" gets to be on ESPN (sort of) and
they don't
Sean Forman
> I believe it's called jealousy--"Robb" gets to be on ESPN (sort of) and
> they don't
Actually, ESPN is now advertising Neyer "Baseball Scribe" and Sickels
on their 28 and 58 tickers.
I realize this whole topic isn't particularly germane to baseball in
general, but I thought I would waste some cyber-ink on this issue as I
don't have very thick skin. Since I haven't changed the subject line to
"BP Sucks" or "Neyer Bites," I hope that you will indulge me or quickly
hit delete. For doing so, I guarantee I won't argue that Bagwell for
Anderson wasn't as lopsided at the time as it appears now.
First off, Donn is on vacation and won't be back for a week or two, hence
the silence on his end. I won't speak for Don for many reasons, one
because my vocabulary is a good deal smaller and two, I haven't seen as
many art house films. I'll just give my take and my opinions on the
matter.
As for the recent BBBA columns, I don't know the entire story, but there
have been some clubhouse chemistry problems in the past. Now Don, et al,
may be pulling a Bobby Valentine (that appears to be the popular opinion),
but sometimes you get fed up with being nice and start lobbing a few
grenades.
As for jealousy, Am I jealous of Neyer's job? You bet, I'd love to
have his job. Do I feel a streak of jealousy when Rany Jazayerli gets to
write Hot Stove Heaters and Keith Law is in Baseball Weekly? Absolutely.
I'm a competitive person and I like our book a whole lot, perhaps more
than it actually deserves (if Amazon.com is to be believed). Those guys
are fine writers, but everyone wants to get a shot. Does Neyer have any
responsibility to mention BBBA? not at all. Are the rhetorical questions
getting old? probably.
I've read every Neyer column for the last two years (and will continue
to). There is no daily columnist I read more with the exception of Art
Martone of the Providence Journal Bulletin (Art writes strictly about the
Red Sox and quite well I might add).
What I have noticed about Neyer that bothers me is how lax he is on
checking out the numbers. He has data that many of us would kill for
and yet he can't check if any team has ever doubled their walk totals from
one season to the next? I'm sure someone can dig up an example of me
doing this, but I dislike it when baseball writers (sabermetricians in
particular) make minor unsubstantiated claims that would be easy to back
up or major unresearched claims that they choose not to spend the time
backing up.
My other pet peeve is more particular. As many times as Neyer has talked
about closers and relievers in the last two years, he has not once
mentioned Doug Drinen's tireless and groundbreaking work in this area.
Many of the questions Neyer wishes could be answered or believes he knows
the answer to in this area are laid out in Doug's reliever usage matrices
and his WPA reports.
Overall, I figure that between rsbb, SABR-L, Neyer, and Martone I'm going
to learn something new every day, and really that is what I'm hoping for
as a sabermetrician.
later,
seann
Sean Forman Program in Applied Mathematics U. Of Iowa
"Are there any math majors in the crowd??"--Natalie Merchant,5/6/96 Ames,IA
analyze any of over 70,000 starts at theHILL www.backatcha.com/demo/
Ted Frank
Sean Forman <sean-...@uiowa.edu> wrote:
>What I have noticed about Neyer that bothers me is how lax he is on
>checking out the numbers. He has data that many of us would kill for
>and yet he can't check if any team has ever doubled their walk totals from
>one season to the next?
I think this is a little much to ask for. Neyer writes a daily column;
I'm willing to forgive him for overstating a claim like this that he makes
as a throwaway comment when he's so ready to follow up with a
comprehensive correction when he does make a mistake. That he sometimes
screws up about arcane baseball trivia is alright when he gets the modern
stuff right so often.
If you want to get mad at Neyer, get mad at him over parroting the mediot
line re small markets and baseball Chicken-Little economics and other
self-fulfilling prophecies. The combination of the Astros and Orioles
this year should've put the final nail in that coffin, but Neyer still
can't see past Royal incompetency. Don't get mad at him over what he does
better than any other baseball writer -- unless what BBBA is mad at is the
fact that Neyer is plugging BP all the time.
>Many of the questions Neyer wishes could be answered or believes he knows
>the answer to in this area are laid out in Doug's reliever usage matrices
>and his WPA reports.
Something BBBA may want to consider is readability. I'm not a genius, but
I've had five semesters of calculus, three of statistics, two of financial
theory, one of econometrics, one of quantitative modelling and one of
optimization theory, with a 3.8 or so average amongst the ten classes, so
I daresay I'm in the top decile or so of potential BBBA readers, and *I*
find VORP and WPA and reliever usage matrices and most of the analysis of
BBBA impenetrable. I purchased the 1998 BBBA, but not the 1999 BBBA
because the only readable part of the earlier edition was your Iowa
reports (which I found frustrating because of the black-box quantitative
analysis). The acronym-and-jargon-to-prose ratio is otherwise jarring.
Impenetrable is probably too strong. I'm sure if I put my mind to it, I
could figure it out. But I'm past the age when I want to put that kind of
effort into a baseball book.
The thing is, it's not even clear what all that complexity adds to the
analysis. Does a two-dimensional VORP work substantially better than one
that combines hits and walks on one dimension?
Please don't look at this as a flame: I'd love to see BBBA succeed. But I
don't read backatcha any more because what isn't impenetrable consists
almost entirely of sniping at other sabermetricians. Chris Kahrl's BP
Transaction Reports are alone worth the price of admission, were there
one.
Disclosure: I'm friends with Chris, worship Gary, have drank with the BP
editorial staff, and keep getting asked to participate more than I have
with BP (one essay in the 1997 edition). This personal bias isn't what's
driving these opinions.
--
"You're a Libertarian, aren't you? You're the guys with the candidates who
get about 1,800 votes. You don't really want to win, do you? You just want
to complain."
-- Gov. J. Ventura
B. David Harrison
>
> I think this is a little much to ask for. Neyer writes a daily column;
> I'm willing to forgive him for overstating a claim like this that he makes
> as a throwaway comment when he's so ready to follow up with a
> comprehensive correction when he does make a mistake. That he sometimes
> screws up about arcane baseball trivia is alright when he gets the modern
> stuff right so often.
Here's the thing. Neyer does this for a living. 24 hours is a long
time to do 5 minutes worth of fact checking. If you get paid to do a job,
you do it right, and if that means making sure what you get paid to write
is actually correct, you do it. When Neyer doesn't fact check, that, IMO,
is just as bad the the mediots that spit the same cliches without checking
to see if they're true.
>
> If you want to get mad at Neyer, get mad at him over parroting the mediot
> line re small markets and baseball Chicken-Little economics and other
> self-fulfilling prophecies.
Here's where I stopped caring about Rob. Once he started on this
topic, I gave up on him.
That said, there's no reason why baseball writing has to be an
adversarial thing. Just put out good product and put forth your ideas.
Doing that doesn't demand that you trash the work of others.
B. David Harrison Anti-Knuckleheadosity:Last Updated 4/21/1999
Live from Seattle http://staff.washington.edu/~bdharris
to reply, spell "warshinton" like you're not from the sticks
"Not feelin' precise, I got a healin' device. It's called a
walkman, on track, bettin' that I get back" -Siah
"This ain't no time when the usual is suitable" -Mos Def
Ted Frank
B. David Harrison <bdha...@u.warshinton.edu> wrote:
>The huddled masses gained the infinite wisdom of Ted Frank
>> I think this is a little much to ask for. Neyer writes a daily column;
>> I'm willing to forgive him for overstating a claim like this that he makes
>> as a throwaway comment when he's so ready to follow up with a
>> comprehensive correction when he does make a mistake. That he sometimes
>> screws up about arcane baseball trivia is alright when he gets the modern
>> stuff right so often.
>
> Here's the thing. Neyer does this for a living. 24 hours is a long
>time to do 5 minutes worth of fact checking.
That assumes that there's one fact to check. Or that this fact would've
taken five minutes -- this one was more like an hour-long project.
This is an especially arcane piece of baseball trivia. He missed it. He
corrected it, and wrote an interesting column about the correction.
>is actually correct, you do it. When Neyer doesn't fact check, that, IMO,
>is just as bad the the mediots that spit the same cliches without checking
>to see if they're true.
No, because Neyer admits when he's wrong. If he was making careless
slip-ups, and wasn't acknowledging the people who correct them, that's one
thing. This particular mistake was one anybody could've made (I would've
made the same assertion without checking: who knew, given that the most
wild variances have tended to be a fourth that of the early-20th century
one he missed). I'm not even sure it's a relevant one: baseball was, in
many ways, a different sport then, and his underlying point (One can't
expect the Pirates to double their walk rate this year) remains true,
even with the minor inaccuracy in supporting data (no team has ever
doubled their walk rate).
It's not like BBBA is so accurate with their data that they can preach
about this sort of thing.
Keith Woolner
> In article <Pine.A41.3.95.990519...@green.weeg.uiowa.edu>,
> Sean Forman <sean-...@uiowa.edu> wrote:
> I daresay I'm in the top decile or so of potential BBBA readers, and *I*
> find VORP and WPA and reliever usage matrices and most of the analysis of
>
> The thing is, it's not even clear what all that complexity adds to the
> analysis. Does a two-dimensional VORP work substantially better than one
> that combines hits and walks on one dimension?
Point of clarification -- VORP (Value Over Replacement Player) has never
appeared in the BBBA. VORP lives on Stathead.com (http://www.stathead.com),
and I post weekly updates to rec.sport.baseball.analysis throughout the
season.
Though I wrote part of BP99, my creation of VORP substantially predates my
involvement with them. VORP has not been part of my contribution to BP to
date, and probably should be considered an independent system at this point.
I think you're confusing VORP with BBBA's QMAX pitcher rating system.
--
Keith Woolner ke...@woolner.com http://www.woolner.com
Stathead Library and VORP reports, http://www.stathead.com
Owner, Red Sox mailing list, http://www.woolner.com/redsox/list
Newsgroup Moderator, news:rec.sport.baseball.analysis
Author, Baseball Prospectus 1999 http://www.baseballprospectus.com
Doug Drinen
m...@Radix.Net (Ted Frank) wrote:
> In article
<Pine.A41.3.95.990519...@green.weeg.uiowa.edu>,
> Sean Forman <sean-...@uiowa.edu> wrote:
> >Many of the questions Neyer wishes could be answered or believes he
knows
> >the answer to in this area are laid out in Doug's reliever usage
matrices
> >and his WPA reports.
>
> Something BBBA may want to consider is readability. I'm not a genius,
but
> I've had five semesters of calculus,
> three of statistics, two of financial
> theory, one of econometrics, one of quantitative modelling and one of
> optimization theory, with a 3.8 or so average amongst the ten classes,
so
> I daresay I'm in the top decile or so of potential BBBA readers, and
*I*
> find VORP and WPA and reliever usage matrices and most of the analysis
of
> BBBA impenetrable. I purchased the 1998 BBBA, but not the 1999 BBBA
> because the only readable part of the earlier edition was your Iowa
> reports (which I found frustrating because of the black-box
quantitative
> analysis). The acronym-and-jargon-to-prose ratio is otherwise
jarring.
Opening remarks: I hope this post doesn't come off as *too* defensive,
because I don't intend it that way. I was, however, mildly offended by
the above paragraph and, like Sean, I'll admit to having skin that is a
little thinner than I'd like it to be.
Essentially, you register two complaints above: (1) "black-box
quantitative analysis," and (2), impenetrability. I think (1) is a
little ironic in light of the fact that almost every number in BP comes
from a "black box." (Note that I am not arguing that there is anything
wrong with this, only that I find it inconsistent that you could be
frustrated by Forman's black-boxes, but not by BP's.)
The glaring counterexamples are Michael Wolverton's SNWL and now his
Reliever Run Expectation Report. First off, let me say that I think
Wolverton's work is superb. But is WPA really more impenetrable than
SNWL or the RRER?
WPA is the net change in the team's at-the-time win probability while a
certain guy was on the mound. P is what his WPA would've been if he had
been perfect. PPPI is P divided by partial innings pitched.
If you understand that paragraph, then I believe that you can follow at
least 80% of everything I've ever written in BBBA. As to the other 20%,
I'll admit that it can get pretty arcane -- some of it is stuff that *I*
would probably skip over if I hadn't written it. I do not expect anyone
to memorize the definitions of CDF and MRF, but I never refer to these
things without reminding the reader what they are. And again, I don't
see how this is different from Wolverton's stuff. Can you tell me, off
the top of your head, what EIRP, ERBS, and EIRf are?
My introductions to WPA in both BBBA 98 and BBBA 99 consist almost
entirely of prose. I put a great deal of effort into making them
understandable, even to people who weren't particularly interested in
the technical aspects. I certainly realize that I'm not Bill James, but
I honestly thought I had succeeded in conveying the general idea without
getting too technical and jargony. I respectfully submit that if you
find WPA impenetrable, you should find SNWL and RRER equally so (Once
again let me make it clear that I have nothing but respect for Mike
Wolverton and the work he does. I am not attempting to make a
qualitative comparison between his work and mine. I am merely
suggesting that the two are equally difficult to grasp the gist of).
Those two introductions are at my web-page if anyone would like to judge
for him or herself.
http://www.public.asu.edu/~ddrinen/wpnew98.html
http://www.public.asu.edu/~ddrinen/wpnew99.html
For some reason, I feel compelled to apologize for this post, so...
Sorry,
Doug
DougP001
>Something BBBA may want to consider is readability.
YES!
>I daresay I'm in the top decile or so of potential BBBA readers, and *I*
>find VORP and WPA and reliever usage matrices and most of the analysis of
>BBBA impenetrable.
A writer's job is to communicate. When an intelligent reader, comfortable
with the topic and concepts under discussion, still can't understand an
article without multiple readings, that's the author's fault. And when the
problem recurs every year, that author had better study a good guide
to clear writing (for example, Joseph M. Williams' "Style: Toward Clarity
and Grace") before his readers give up in disgust.
If the BBBA staff wonders why the Prospectus writers get more respect
from ESPN, they need only think how the average fan would react on first
exposure to the BBBA's acronyms, circumlocution, and general self-
indulgence. I've bought almost every edition since the first, but if the
value-to-BS ratio keeps falling...
Doug Pappas
B. David Harrison
>
> That assumes that there's one fact to check. Or that this fact would've
> taken five minutes -- this one was more like an hour-long project.
I'm speaking in general terms (having not actually read the article in
question). I just think that as a principle anybody who's getting paid to
write, be it Rob Neyer or Peter Gammons is obligated to do the necessary
research to make sure they're right in what they say.
basil
Doug Drinen <Dri...@asu.edu> wrote:
> Opening remarks: I hope this post doesn't come off as *too*
defensive,
> because I don't intend it that way. I was, however, mildly offended
by
> the above paragraph and, like Sean, I'll admit to having skin that is
a
> little thinner than I'd like it to be.
With all due respect, I think this is a point that is a bit puzzling:
the disconnect between two of BBBA's writers by their own admission not
having very thick necks, yet their publication repetitively and
sophomorically denegrating another writer at the same time. I have no
idea how the inner-workings of the BBBA operates, and Sean Forman did
indicate that he didn't contribute to "Operation Neyer" so perhaps I
should cut you guys some slack. Both of you do good work, and I
generally enjoy your book and website. I just believe that Malcolm,
Adams, and Jones can make an argument in more productive ways, and to be
honest they usually do.
As for readability, I have an undergrad degree with two decidedly
un-statistical majors (religion & communication) but I have no problem
reading your work if I focus enough on what you're saying. Heck, I spend
all day looking at codes, numbers, and acronymns more arcane (I'm an
analyst at the Federal Election Commission) than those you guys put
out:)
Of course, it's not casual reading. That's when I pick up the BJ
Baseball Book 1990 or something low maintenance like that.
--
William "Basil" Tsimpris btsi...@fec.gov
Ted Frank
>Opening remarks: I hope this post doesn't come off as *too* defensive,
>because I don't intend it that way. I was, however, mildly offended by
>the above paragraph and, like Sean, I'll admit to having skin that is a
>little thinner than I'd like it to be.
>
>quantitative analysis," and (2), impenetrability.
There was a third complaint, related to the second, which is plain style.
And a fourth complaint, which is snarkiness. I'm pleased to see the
number of people who agree with me; I don't put much weight in argument
from popularity of position, but in this case, I don't think I'm too off
base and the number of people who've weighed in with substantially
identical positions supports me.
> I think (1) is a
>little ironic in light of the fact that almost every number in BP comes
>from a "black box." (Note that I am not arguing that there is anything
>wrong with this, only that I find it inconsistent that you could be
>frustrated by Forman's black-boxes, but not by BP's.)
Who says I'm not frustrated by BP's black boxes?
>The glaring counterexamples are Michael Wolverton's SNWL and now his
>Reliever Run Expectation Report. First off, let me say that I think
>Wolverton's work is superb. But is WPA really more impenetrable than
>SNWL or the RRER?
>
>WPA is the net change in the team's at-the-time win probability while a
>certain guy was on the mound. P is what his WPA would've been if he had
>been perfect. PPPI is P divided by partial innings pitched.
Have you ever seen the movie "A Day at the Races"? There's a scene where
Chico sells Groucho a system for decoding horse-racing tips. This
paragraph reminds me of nothing but. You have a non-intuitive four-letter
acronym that is derived from a new concept that in turn is derived from a
new concept. This is perfectly alright in a clear self-contained essay,
but the BBBA doesn't understand that its readers aren't going to treat a
six-hundred page volume as self-contained. I may have read Bill James
cover to cover as a kid, but that doesn't happen today: I don't have time.
I'll read a team essay here, look up a player there. I can do that with
the Baseball Prospectus. I can't do that with the BBBA without having to
check the master volume index that Chico provides me. So, "impenetrable"
is too strong, because I suppose BBBA would make sense with effort that
I haven't been willing to spend (and if I'm not, who in your readership
really is?), but "unnecessarily dense and difficult to read" is accurate.
>see how this is different from Wolverton's stuff. Can you tell me, off
>the top of your head, what EIRP, ERBS, and EIRf are?
No, I can't. I can't ever recall seeing it in a team essay or player
essay. There may be counterexamples, but my recollection is my
recollection: I think Wolverton's stuff is in a few-page essay in the
front or back of the book, and sometimes I read it and don't understand
it, other times I read it and think the methodology flawed, and other
times I don't even read it. But it doesn't hurt my enjoyment of the rest
of the book, whereas QMAX permeates BBBA.
My problem with both WPA and SNWL is that they both, through a lot of
work, give the illusion of greater accuracy without necessarily providing
it. Conceptually, both make sense. In practice, by measuring some
situation-dependent things and not others (park effects, platoon
advantages, which batters are being faced), I'm not sure they add much in
the way of understanding. I respect the work that's been done by both you
and Wolverton, which is why you're both in my auto-select file when I read
rsbb. But in neither case is it what I'm talking about when I
differentiate BP from BBBA.
David Marc Nieporent
>Sean Forman <sean-...@uiowa.edu> wrote:
>>What I have noticed about Neyer that bothers me is how lax he is on
>>checking out the numbers. He has data that many of us would kill for
>>and yet he can't check if any team has ever doubled their walk totals from
>>one season to the next?
>I think this is a little much to ask for. Neyer writes a daily column;
>I'm willing to forgive him for overstating a claim like this that he makes
>as a throwaway comment when he's so ready to follow up with a
>comprehensive correction when he does make a mistake. That he sometimes
>screws up about arcane baseball trivia is alright when he gets the modern
>stuff right so often.
>If you want to get mad at Neyer, get mad at him over parroting the mediot
>line re small markets and baseball Chicken-Little economics and other
>self-fulfilling prophecies. The combination of the Astros and Orioles
>this year should've put the final nail in that coffin, but Neyer still
>can't see past Royal incompetency. Don't get mad at him over what he does
>better than any other baseball writer -- unless what BBBA is mad at is the
>fact that Neyer is plugging BP all the time.
That's only part of it, because they also whined that he plugged STATS
stuff. I really couldn't figure out what their complaint was, there. I
understand their jealousy of BP, but BBBA and STATS aren't exactly direct
competitors.
And I really don't get the spelling-Rob's-name-wrong thing.
>>Many of the questions Neyer wishes could be answered or believes he knows
>>the answer to in this area are laid out in Doug's reliever usage matrices
>>and his WPA reports.
>Something BBBA may want to consider is readability. I'm not a genius, but
>I've had five semesters of calculus, three of statistics, two of financial
>theory, one of econometrics, one of quantitative modelling and one of
>optimization theory, with a 3.8 or so average amongst the ten classes, so
>I daresay I'm in the top decile or so of potential BBBA readers, and *I*
>find VORP and WPA and reliever usage matrices and most of the analysis of
>BBBA impenetrable. I purchased the 1998 BBBA, but not the 1999 BBBA
>because the only readable part of the earlier edition was your Iowa
>reports (which I found frustrating because of the black-box quantitative
>analysis). The acronym-and-jargon-to-prose ratio is otherwise jarring.
>Impenetrable is probably too strong. I'm sure if I put my mind to it, I
>could figure it out. But I'm past the age when I want to put that kind of
>effort into a baseball book.
Thank you. I'm glad someone else said this. Like you, I've had enough
upper level math that baseball analysis shouldn't be complicated, and yet
I have a hard time getting through BBBA. Or perhaps it's just a hard time
staying awake through BBBA. Much of the analysis is interesting, but
someone needs a new editor. Apparently the one they have now is of the
opinion that one should never use one word where seventeen will do,
especially if eleven of them are obscure references and four more are
condescendingly letting you know how inferior you are for not getting
those references.
>The thing is, it's not even clear what all that complexity adds to the
>analysis. Does a two-dimensional VORP work substantially better than one
>that combines hits and walks on one dimension?
You mean QMAX. VORP is a very straightforward value over replacement
player, and I don't even think it has anything to do with BBBA.
And I agree. We hear all this talk about how superior QMAX is, but I'm
not sure what it's supposed to be superior _for._ But it must be,
because there are lots of charts.
>Please don't look at this as a flame: I'd love to see BBBA succeed. But I
>don't read backatcha any more because what isn't impenetrable consists
>almost entirely of sniping at other sabermetricians. Chris Kahrl's BP
>Transaction Reports are alone worth the price of admission, were there
>one.
Apparently someone at BBBA reads those, says "Aha, calling someone an
idiot sells, so we ought to do that, too." But there's a big difference
between calling Herk Robinson an idiot and calling a competitor one.
There's a column on backatcha which I read a couple of times before I got
so disgusted I stopped going to the site. (C.O. Jones?) One criticized
some of Neyer's analysis in a rude way, but then another trashed Neyer for
suggesting that Chief Wahoo is an inappropriate logo. That's an opinion
on which honest people can disagree, but this column didn't bother
disagreeing; apparently the author was being paid by the ad hominem.
--
David M. Nieporent "Mr. Simpson, don't you worry. I
niep...@alumni.princeton.edu watched Matlock in a bar last night.
2L - St. John's School of Law The sound wasn't on, but I think I
Roberto Petagine Appreciation Society got the gist of it." -- L. Hutz
NawrockiT
: You mean QMAX. VORP is a very straightforward value
: over replacement
: player, and I don't even think it has anything to do with
: BBBA.
: And I agree. We hear all this talk about how superior
: QMAX is, but I'm
: not sure what it's supposed to be superior _for._ But it
: must be,
: because there are lots of charts.
I got the impression from the column ripping Pitcher Abuse Points for not being
sufficiently peer-reviewed that what the BBBA boys really are upset about is
that nobody peer-reviews (i.e., notices) their own stuff. And their pride and
joy would seem to be QMAX. I'm sure the backatcha-ites think that if the Rob
Neyers of the world would just discover QMAX, it'd quickly become the pitching
stat of choice.
I'm not a mathematician, or even a sabermetrician, but from what I've seen of
QMAX, the old adage of "be careful what you wish for; you just might get it"
would apply. You really want careful scrutiny applied to this thing?
It tells you how many hits *and* how many walks a pitcher gives up in his
starts--gee, guys, couldn't I just read the box score?
Tom Nawrocki
Mosey
m...@Radix.Net (Ted Frank) wrote:
> > Here's the thing. Neyer does this for a living. 24 hours is a
long
> >time to do 5 minutes worth of fact checking.
> That assumes that there's one fact to check. Or that this fact
would've
> taken five minutes -- this one was more like an hour-long project.
Frankly, I don't see this as too much of an excuse. Say he puts in your
average 8 hour day. If he checks that one (admitdly harder than most
facts) fact over an hour, he has 7 other hours to do his thing.
As often as he goes on about how he looked something up in his historic
Sporting News back issues, which most of the time looks like something
that would take several hours, "it would take an hour" seems pretty
flimsy.
I don't know if he has an intern or something doiung research for him,
but as Sean said, he has resources most of us drool over.
> No, because Neyer admits when he's wrong.
To be honest, he seems to do that quite a bit these days.... More than I
remembered him doing last year, but that is just my observation, I could
be wrong.
--
John Mosey FIRE BANDO!
"I drive too fast, talk too loud and I'm an
asshole. Deal with it."
http://www.mosey.com Mosey's Fantasy Baseball
Doug Drinen
I understand your points better now, and I no longer have any major
complaints. Just a few comments...
In article <7i1np9$klj$1...@saltmine.radix.net>,
m...@Radix.Net (Ted Frank) wrote:
> In article <7i1k1i$ap2$1...@nnrp1.deja.com>, Doug Drinen
<Dri...@asu.edu> wrote:
> >The glaring counterexamples are Michael Wolverton's SNWL and now his
> >Reliever Run Expectation Report. First off, let me say that I think
> >Wolverton's work is superb. But is WPA really more impenetrable than
> >SNWL or the RRER?
> >
> >WPA is the net change in the team's at-the-time win probability while
a
> >certain guy was on the mound. P is what his WPA would've been if he
had
> >been perfect. PPPI is P divided by partial innings pitched.
>
> Have you ever seen the movie "A Day at the Races"? There's a scene
> where
> Chico sells Groucho a system for decoding horse-racing tips. This
> paragraph reminds me of nothing but.
> You have a non-intuitive four-letter
> acronym that is derived from a new concept that in turn is derived
> from a new concept.
Coupla things here...
First and least importantly, what acronym *isn't* non-intuitive? BASIC
and ICQ are the only two I can think of, and I'm not sure ICQ is
actually an acronym.
I will grant you that if you don't want to keep the gist of those three
stats (WPA, P, PPPI) in the quick-recall section of your brain, you will
probably have a hard time enjoying a Drinen article. But out of
curiosity, how does my paragraph differ from the following:
OBP is the number of times a player reaches base via a hit or a walk as
a fraction of the number of times he comes to the plate. SLG is a
weighted batting average, with a single counting one, a double two, and
so forth. OPS is the sum of OBP and SLG.
The only difference I can see is that we were all exposed to OPS earlier
and oftener.
Now, you can claim that WPA isn't worthwhile in the first place, and I
won't argue with that, at least not in this thread (I'm certainly not
in any way trying to claim that WPA is on par with OPS in terms of
importance). But let's assume for the sake of argument that it is a
valuable contribution to sabermetrics. The question now becomes: how
is one supposed to integrate a new piece of sabermetrics into the
sabermetric community without using it in places besides a
self-contained article?
Changing gears...
> My problem with both WPA and SNWL is that they both, through a lot of
> work, give the illusion of greater accuracy without necessarily
> providing
> it. Conceptually, both make sense. In practice, by measuring some
> situation-dependent things and not others (park effects, platoon
> advantages, which batters are being faced), I'm not sure they add much
> in the way of understanding.
This is a totally separate argument. I won't get into it here, except
to say that your point has merit. Also, a factual correction is in
order: park effects *are* accounted for in both the SN stats and the WP
stats.
Doug
Sean Forman
> David Nieporent wrote:
> : You mean QMAX. VORP is a very straightforward value
> : over replacement
> : player, and I don't even think it has anything to do with
> : BBBA.
> : And I agree. We hear all this talk about how superior
> : QMAX is, but I'm
> : not sure what it's supposed to be superior _for._ But it
> : must be,
> : because there are lots of charts.
[snip]
> It tells you how many hits *and* how many walks a pitcher gives up in his
> starts--gee, guys, couldn't I just read the box score?
> Tom Nawrocki
I don't mean to be coy, but which is it "dense and unreadable" or too
simple and obvious? David calls VORP very straightforward compared to
QMAX and Tom you are saying that it is just rearranging the box score.
later,
sean
Sean Forman Program in Applied Mathematics U. Of Iowa
"Sanity is not in numbers," from "1984" by George Orwell
**The Big Bad Baseball Annual**, from Master's Press www.backatcha.com
NawrockiT
: I don't mean to be coy, but which is it "dense and
: unreadable" or too
: simple and obvious? David calls VORP very
: straightforward compared to
: QMAX and Tom you are saying that it is just
: rearranging the box score.
Well, I am not David, and David, if you were to ask him, would probably say
that he is not me. But I think we're both right. It takes a simple concept, one
for which the information is readily available, and complicates it.
Near as I can tell, QMAX takes the number of walks and the number of hits (not
total bases) a starter allows and turns them into a plot point. Then it puts
them on a graph. Then you look at them and say, boy, Greg Maddux sure has
gotten lit up, hasn't he? Or at least Don Malcolm says he has and he has a
nifty little graph, so I assume he's right. Plus the box scores and his ERA and
his hits per nine innings all say he's getting lit up, so tell me again why I
needed to look at a QMAX graph?
Perhaps I am not qualified to comment, as I'm not an expert and haven't put a
great deal of study into QMAX. But then again, I'm your target audience, so if
my cursory readings of it haven't impressed me, that might tell you something.
I should also say that it's a bit unfair for us to be having this discussion
while Don Malcolm is apparently on vacation and unable to respond.
Tom Nawrocki
Doug Drinen
basil <spi...@my-dejanews.com> wrote:
> With all due respect, I think this is a point that is a bit puzzling:
> the disconnect between two of BBBA's writers by their own admission
> not
> having very thick necks, yet their publication repetitively and
> sophomorically denegrating another writer at the same time. I have no
> idea how the inner-workings of the BBBA operates, and Sean Forman did
> indicate that he didn't contribute to "Operation Neyer" so perhaps I
> should cut you guys some slack.
By my count, there are more than 20 people who contributed to BBBA 99
(BTW, I have never actually met any of those 20+, and have had more
than brief e-mail contact with only 2). As Don says on page 10, "we do
not fear to disagree, and make those disagreements public knowledge."
So it should not be assumed that all 20+ of those people share the same
views on anything, necessarily. Fortunately, every single word in BBBA
and every single article at the website has the author clearly marked,
so it is easy to attribute any opinion expressed in BBBA to an
individual.
Ira Blum
The problem I have with WPA (and I'll admit that I haven't read the
BBBA)
is that I don't understand the metric for it.
with SNWL, break it down into SNW and SNL and the metric for each is
wins
and losses. Can that be done with WPA? is it? if so, then it makes
sense,
if not, then it doesn't. All of the numbers in BP may be translated,
but
at least they have the same metrics as those of any normal sports fan.
Translated ERA may be different than Raw ERA, but its still in the same
metric.
Ira
Wheat Thin
Sean Forman <sean-...@uiowa.edu> wrote:
>On 20 May 1999, NawrockiT wrote:
>> It tells you how many hits *and* how many walks a pitcher gives up in his
>> starts--gee, guys, couldn't I just read the box score?
>
>> Tom Nawrocki
>
>simple and obvious? David calls VORP very straightforward compared to
>QMAX and Tom you are saying that it is just rearranging the box score.
I think that's the point. Correct me if I'm wrong, but QMAX appears to be
founded on a simple -- perhaps too simple -- assumption, that good pitchers
give up less hits and walks. It's in the manipulation of the data,
particularly in the 1-7 rating boxes and the start-by-start analyses and
charts, that QMAX begins to appear dense.
To me, this appears to be making a mountain out of a molehill. The only
factor being accounted for here, in the end, is opponent OBP. Slugging
is completely neglected by QMAX, as are stolen bases, park effects, and
defense. That doesn't mean it's not interesting or useful, but it's hardly
authoritative, and putting the data through all the charts and acronyms
doesn't do much to help me understand the argument being presented.
--
Tommy J.M.D. Strong
"Beware when the so-called sagely men come limping into sight."
--Zhuangzi
Ted Frank
>> taken five minutes -- this one was more like an hour-long project.
>
>question). I just think that as a principle anybody who's getting paid to
>write, be it Rob Neyer or Peter Gammons is obligated to do the necessary
>research to make sure they're right in what they say.
I'll agree with that principle in general, and state that it is
inapplicable to the complaint about Rob Neyer in particular.
As I wrote to Sean Forman in e-mail:
But Neyer's exact quote was "You have to wonder if any team in the
history of the game has more than doubled its walk rate from one season to
the next." C'mon! How much would it have added to this column to mention
four teams 90 to 110 years ago? Nothing. It would've turned an
interesting column into a pedantic one. It was an interesting topic in
and of itself, because Rob could spend time talking about the 1909
Cardinals in details, but to add four sentences in the middle of the
column about the Pirates would've been distracting and made the column
worse.
So even I was too harsh on Neyer when I said he "corrected" himself; he
never made a mistake to begin with.
Jeremy Buchman
DougP001 <doug...@aol.com> wrote:
>
>A writer's job is to communicate. When an intelligent reader, comfortable
>with the topic and concepts under discussion, still can't understand an
>article without multiple readings, that's the author's fault. And when the
>problem recurs every year, that author had better study a good guide
>to clear writing (for example, Joseph M. Williams' "Style: Toward Clarity
>and Grace") before his readers give up in disgust.
Rather than jump into the BP-BBBA battle (sounds like a Dr. Seuss
title, no?), I'll just concur in the virtues of Joseph Williams'
book. I can't recommend it highly enough. (And no, I'm not being paid
to say this.)
--
Jeremy Buchman
Department of Political Science
Stanford University
jbuc...@leland.stanford.edu
"Don't look back--something might be gaining on you."--Satchel Paige
David Marc Nieporent
> m...@Radix.Net (Ted Frank) wrote:
>> > Here's the thing. Neyer does this for a living. 24 hours is a
>> > long time to do 5 minutes worth of fact checking.
>> That assumes that there's one fact to check. Or that this fact
>> would've taken five minutes -- this one was more like an hour-long
>> project.
>Frankly, I don't see this as too much of an excuse. Say he puts in your
>average 8 hour day. If he checks that one (admitdly harder than most
>facts) fact over an hour, he has 7 other hours to do his thing.
Well, out of curiosity, do we know what "his thing" is? I'm curious. I
know Gammons is a full-time columnist. All he has to do is talk to
sources and write articles. OTOH, I can't imagine that it's a full-time
job for Neyer to write his columns. (If it is, I want it.)
>As often as he goes on about how he looked something up in his historic
>Sporting News back issues, which most of the time looks like something
>that would take several hours, "it would take an hour" seems pretty
>flimsy.
OTOH, there's a difference between taking an hour to look something up,
and taking an hour looking to see whether something exists, IMO. If he
spends an hour looking for tht one fact, sure, that's no big deal. But
there are *lots* of factoids in a column, and if he has to spend an hour
looking for every one, that's a big deal. Especially in the specific
scenario we're talking about. If he spends an hour and there's nothing to
find, that's a pretty big investment for one sentence. If he spends an
hour and *does* find that team, so what? Will it change the point of that
particular column? I don't think so.
>I don't know if he has an intern or something doiung research for him,
>but as Sean said, he has resources most of us drool over.
I sincerely doubt that. He works for ESPNet, not Microsoft.
>> No, because Neyer admits when he's wrong.
>To be honest, he seems to do that quite a bit these days.... More than I
>remembered him doing last year, but that is just my observation, I could
>be wrong.
My recollection is similar to yours, but that hardly proves anything.
After all, when he first started writing, it was so nice to see an
alternative to Gammons Et Al that minor mistakes were unimportant. Now
that we take him for granted, we want to see perfection.
Ted Frank
>I will grant you that if you don't want to keep the gist of those three
>stats (WPA, P, PPPI) in the quick-recall section of your brain, you will
>probably have a hard time enjoying a Drinen article. But out of
>curiosity, how does my paragraph differ from the following:
>
>OBP is the number of times a player reaches base via a hit or a walk as
>a fraction of the number of times he comes to the plate. SLG is a
>weighted batting average, with a single counting one, a double two, and
>so forth. OPS is the sum of OBP and SLG.
>
>The only difference I can see is that we were all exposed to OPS earlier
>and oftener.
Well, just so. It's one thing to have an easily understandable and
calculable metric with an acronym. It's another to take a measure that
throws in everything but the kitchen sink and make that an acronym.
Read some of the abstracts. If Bill James is talking about runs created,
he says "runs created." If he had to make a measure like "RC/27 outs" he
used an acronym that made it intuitive for someone with a basic level of
knowledge: OERA, or Offensive ERA or Offensive Earned Run Average. He'd
say "Park Factor" rather than PF. Occasionally he'd slip, as when he came
up with the Baseball-Digest-style statistic "Victory Important RBI" and
then litter an entire Abstract with the term "VI-RBI". But, in general,
James had a style worth striving for, even if his analysis was sometimes
flawed and careless.
>valuable contribution to sabermetrics. The question now becomes: how
>is one supposed to integrate a new piece of sabermetrics into the
>sabermetric community without using it in places besides a
>self-contained article?
It becomes a question of style. You use language familiar to the reader,
rather than jargon.
If BBBA only aspires to be useful to hard-core sabermetricians, there's
nothing wrong with that. I think it's clear that Baseball Prospectus is
aiming for a wider audience, and BBBA shouldn't be so snarky about it. If
BBBA is really aiming for the Baseball Prospectus audience, then it has to
do some paring of (or segregation of) its sabermetric analysis to reach
that accessibility, and concentrate more on the prose side of things.
In one of his books, Bill James had an excellent essay where he tried to
measure the value of certain baseball statistics, balancing accuracy, ease
of use, and something else. It's something all sabermetricians should
look at.
Ken Bowers
> I'm not a mathematician, or even a sabermetrician, but from what I've seen of
> QMAX, the old adage of "be careful what you wish for; you just might get it"
> would apply. You really want careful scrutiny applied to this thing?
I'm no expert either, but I like QMAX. IMO it shows some basic patterns
of pitcher performance very nicely, and it works both at the game
level and over time. I suspect the BBBA folks sometimes try to wring
more insight out of it than it can really support, but that doesn't
make it a bad, or useless, measure.
> It tells you how many hits *and* how many walks a pitcher gives up in his
> starts--gee, guys, couldn't I just read the box score?
That's not true of SLG and OBP? Like SLG and OBP, QMAX S and C scores
are easy to grasp(*). Their real value comes when they are applied
to sets of games, but you can look at a box score and see where a single
game fits into the bigger pattern. This has been true for batting
stats for years, but I've never seen it done with hit and walk
prevention.
On BBBA's presentation and prose style:
I find that end up skimming some of the background stuff, but I can
figure it out when I try. I think there's a philosophy that the
reader is better off with more information even if it hasn't been
sorted out perfectly. I suspect that one reason it isn't sorted
out perfectly is that BBBA's measures and methods are evolving
pretty quickly. As a result they've constantly got new wrinkles
to show off (and new acronyms to go with them).
I've always seen the big difference between BP and BBBA as BBBA's
fascination with breaking down what happens during games and over
a short period of time. My view is that the BBBA people are very
busy doing this and that they turn around their work pretty quickly
to the website and the book. The result of that can be dense and
a little overwhelming, but that there are a lot of interesting
ideas if the reader works a bit.
Yes, there are times some editing would help, but I don't find
the BBBA stuff all that hard to get through. OTOH, I do
find the tone of the PAP article and some of the recent attacks
on Rob Meyer distressing. Embarassing, really, to someone who
otherwise likes BBBA's work quite a bit.
Ken B.
(*) One thing I don't get is why the QMAX scales are set where they
are. In the QMAX FAQ they are issued without explanation. Maybe
it wouldn't make much difference if they ran, say, 1-14 for hits
and 1-10 for walks, but it'd be interesting to know the reasoning
behind the scales.
Dan Schmidt
| We hear all this talk about how superior QMAX is, but I'm not sure
| what it's supposed to be superior _for._ But it must be, because
| there are lots of charts.
I'm not a big QMAX fan either, but one thing it does have going for it
is that it's per-start, not per-inning. So you can try to talk about
a pitcher's average performance per start, without having your
denominator be something other than starts, like it is with ERA.
I can't think off the top of my head of a non-degenerate case in which
that actually buys you a lot, but at least theoretically it's a good
idea. I think the BBBA 99 even mentioned that point.
--
Dan Schmidt -> df...@harmonixmusic.com, df...@alum.mit.edu
Honest Bob & the http://www2.thecia.net/users/dfan/
Factory-to-Dealer Incentives -> http://www2.thecia.net/users/dfan/hbob/
Gamelan Galak Tika -> http://web.mit.edu/galak-tika/www/
Patrick G. Bridges
> But out of curiosity, how does my paragraph differ from the following:
> OBP is the number of times a player reaches base via a hit or a walk as
> a fraction of the number of times he comes to the plate.
Unfortunately, it doesn't. This should have said something like:
OBP (on-base percentage) is the percentage of plate appearances at
which a player gets a hit or a walk.
One thing BP has done very well is to try and make their statistics
accessible on scales people understand, or, when that's not possible,
to keep things reasonably simple. PERA, SWNL, and EqA are fine
examples of this, and I think there is an importatnt lesson to learn
there. For example, now that I actually understand the QMAX numbers
(it is an interesting idea, BTW), I think you'd do better to actually
sell it as two statistics, hit-prevention and walk-prevention.
That way, you can quote two (related) statistics about a player,
discuss them either together or seperately, and then note that these
two statistics can be used with a nifty chart to better understand
pitchers, and their expected WHIP, ERA, etc. You can then point out
various interesting regions on the chart. Bandying about tuples like
(3,6) and (5,2) just confuses people and doesn't tell them very much
unless the've memorized a 49-element chart. Discussing QMAX as two
related statistics, each with individual, easy-to-understand meaning,
could make it much easier for the average guy to understand.
--
*** Patrick G. Bridges bri...@cs.arizona.edu ***
*** #include <std/disclaimer.h> ***
Chris Dial
<7i13th$cpi$1...@saltmine.radix.net>...
>In article
<Pine.A41.3.95.990519...@green.weeg.uiowa.
edu>,
>Sean Forman <sean-...@uiowa.edu> wrote:
{snip}
>>Many of the questions Neyer wishes could be answered or
believes he knows
>>the answer to in this area are laid out in Doug's reliever
usage matrices
>>and his WPA reports.
>
>Something BBBA may want to consider is readability. I'm
not a genius, but
>I've had five semesters of calculus, three of statistics,
two of financial
>theory, one of econometrics, one of quantitative modelling
and one of
>optimization theory, with a 3.8 or so average amongst the
ten classes, so
>I daresay I'm in the top decile or so of potential BBBA
readers, and *I*
>find VORP and WPA and reliever usage matrices and most of
the analysis of
>BBBA impenetrable. I purchased the 1998 BBBA, but not the
1999 BBBA
>because the only readable part of the earlier edition was
your Iowa
>reports (which I found frustrating because of the black-box
quantitative
>analysis). The acronym-and-jargon-to-prose ratio is
otherwise jarring.
I bought my first of each this year. Personally, I find the
sniping pointless, but mostly I'm surprised BBBA considers
BP a competitor.
Hard to read? I suppose it's not for individuals who don't
have to read technical articles, where acronyms rule, but
the thing I struggle with the most (from both books) is:
why?
BP reads like a guide for Roto players. And they don't
identify who made which comment, so when the author refers
to himself, you don't know WTF is going on.
BBBA provides more analysis and does include author
initials.
>Impenetrable is probably too strong. I'm sure if I put my
mind to it, I
>could figure it out. But I'm past the age when I want to
put that kind of
>effort into a baseball book.
>
>The thing is, it's not even clear what all that complexity
adds to the
>analysis. Does a two-dimensional VORP work substantially
better than one
>that combines hits and walks on one dimension?
You have to be talking QMAX. Woolner's stuff in BP is the
best thing in there. And BP is just as guilty as BBBA for
cryptology.
Of course, from baseball books from the authors as I've come
to know them here, there are FAR too few *numbers*. Thanks
for the analysis, guys, but dammit, provide the raw data
*before* you change them. *I'll* figure out how I want to
use them.
>
>Please don't look at this as a flame: I'd love to see BBBA
succeed. But I
>don't read backatcha any more because what isn't
impenetrable consists
>almost entirely of sniping at other sabermetricians. Chris
Kahrl's BP
>Transaction Reports are alone worth the price of admission,
were there
>one.
Witty writing is.
>
>Disclosure: I'm friends with Chris, worship Gary, have
drank with the BP
>editorial staff, and keep getting asked to participate more
than I have
>with BP (one essay in the 1997 edition). This personal
bias isn't what's
>driving these opinions.
Well, maybe not completely.
My bias lies in untranslated data. The readers of these
books (I'd think) understand park effects. I mean, geez, if
you don't provide the raw data, how can I tell if you did
the freakin' math right. There's plenty of other typos.
Chris Dial
Nelson Lu
David Marc Nieporent <niep...@alumni.princeton.edu> wrote:
>That's only part of it, because they also whined that he plugged STATS
>stuff. I really couldn't figure out what their complaint was, there. I
>understand their jealousy of BP, but BBBA and STATS aren't exactly direct
>competitors.
Disclaimer: I have contributed in small parts to the last two BBBA, but am not
a major contributor.
As far as I know (I hope that Don won't be upset about my saying this), a
reason why there might be animosity toward STATS is because STATS has
continuously sidetracked BBBA's attempts to purchase data from them; no quotes,
no outright refusals, no nothing, just continous delays.
Personally, we have also run into this poor treatment by STATS while trying to
purchase data for The IBL Game (for more details, see http://www.ibl.org --
it's a not-for-profit simulation game for use in our league). I'm thus also a
bit upset at STATS in general.
===============================================================================
GO ANAHEIM ANGELS!
===============================================================================
Nelson Lu (n...@cs.stanford.edu)
Cris Whetstone
>
> On 20 May 1999, NawrockiT wrote:
>
>
> > : You mean QMAX. VORP is a very straightforward value
> > : over replacement
> > : player, and I don't even think it has anything to do with
> > : BBBA.
>
> > : QMAX is, but I'm
> > : not sure what it's supposed to be superior _for._ But it
> > : must be,
> > : because there are lots of charts.
>
>
> > It tells you how many hits *and* how many walks a pitcher gives up in his
> > starts--gee, guys, couldn't I just read the box score?
>
>
> I don't mean to be coy, but which is it "dense and unreadable" or too
> simple and obvious? David calls VORP very straightforward compared to
> QMAX and Tom you are saying that it is just rearranging the box score.
As Mr Strong has noted you have hit the nail on the head. Which is it? I
think it's up to BBBA to tell us. My experience reading about Q-MAX is
somewhat the same as David and Tom's who seem to be very knowledgeable
guys who have posted here for quite some time. When I first read about
it made little sense and seemed like a long climb to get anything useful
out of it without extended work. Upon further reading it seemed to boil
down to walk and hit prevention. So what? Is this supposed to be
something I _can't_ understand without jumping through a bunch of hoops
to get there? As Tom said, I do that everday when I read boxscores and
every week when I read player stat lines.
--
Cris Whetstone ** To reply remove the * from my address**
cris...@ix.netcom.com Frank Catalanotto Fan Club
jwhe...@engr.csulb.edu
"Work is the curse of the drinking class." - Oscar Wilde
David Brazeal
Ted Frank wrote in message <7i211s$agt$1...@saltmine.radix.net>...
>In article
<Pine.A41.4.10.990520...@homer39.u.washington.edu>,
>B. David Harrison <bdha...@u.warshinton.edu> wrote:
>I'll agree with that principle in general, and state that it is
>inapplicable to the complaint about Rob Neyer in particular.
>
>As I wrote to Sean Forman in e-mail:
>But Neyer's exact quote was "You have to wonder if any team in the
>history of the game has more than doubled its walk rate from one season to
>the next." C'mon! How much would it have added to this column to mention
>four teams 90 to 110 years ago? Nothing.
I disagree here. Given the state of information nowadays, it's not that big
a job to find that kind of thing out. You're assessing the lack of
information in hindsight. Sure, now that we KNOW four teams 90 to 110 years
ago doubled their walks rate, it's not so interesting. But in the first
column, it was left wide open. For all I knew, it happened 15 years ago, or
25, or three times in the 1950s. It's only after we found out the answer
that we realized it didn't add much to the column. It was the lack of an
answer that _detracted_ from the column. I remember thinking at the time,
"Why doesn't he just check?" It doesn't even have to be a thorough
check--at least go back 20 years and tell us it hasn't been done in that
period of time. That gives us some idea, rather than leaving it wide open.
I really like Neyer, but he _does_ tend to leave things like that
unanswered, instead of just looking them up to give us some context. And it
IS his job, as someone said above.
davidb
Doug Drinen
By all means.
> The problem I have with WPA (and I'll admit that I haven't read the
> BBBA)
> is that I don't understand the metric for it.
> with SNWL, break it down into SNW and SNL and the metric for each is
> wins
> and losses. Can that be done with WPA? is it? if so, then it makes
> sense,
> if not, then it doesn't. All of the numbers in BP may be translated,
> but
> at least they have the same metrics as those of any normal sports fan.
> Translated ERA may be different than Raw ERA, but its still in the
> same
> metric.
I'm not sure I understand your use of the word "metric." I think I can
answer your question, though.
WPA's units are wins. More precisely, wins above average. If your WPA
is 0, you're average. If your WPA is +1, you added one win beyond what
an average pitcher would have done. If your WPA is +2.2 and it's
mid-May, then you're Jeff Zimmerman. (Gratuitous Rangers reference)
WPA is directly comparable to Total Baseball's Pitching Wins and, if I'm
not mistaken, SNVA. I don't know if "any sports fan" is willing and
able to appreciate Pitching Wins, but certainly any
sabermetrically-inclined one is. I don't see why *every* sabermetric
stat needs to be directly comparable to a traditional one.
Doug Drinen
m...@Radix.Net (Ted Frank) wrote:
> Well, just so. It's one thing to have an easily understandable and
> calculable metric with an acronym. It's another to take a measure
> that
> throws in everything but the kitchen sink and make that an acronym.
>
> Read some of the abstracts. If Bill James is talking about runs
> created,
> he says "runs created." If he had to make a measure like "RC/27 outs"
> he
> used an acronym that made it intuitive for someone with a basic level
> of
> knowledge: OERA, or Offensive ERA or Offensive Earned Run Average.
> He'd
> say "Park Factor" rather than PF. Occasionally he'd slip, as when he
> came
> up with the Baseball-Digest-style statistic "Victory Important RBI"
> and
> then litter an entire Abstract with the term "VI-RBI". But, in
> general,
> James had a style worth striving for, even if his analysis was
> sometimes
> flawed and careless.
Well, by all means, if not being as good a writer as Bill James is a
crime, I'm guilty as hell. I couldn't agree more that his style is
worth striving for.
You (and others) have given me lots to think about in this thread.
Thanks for your comments.
Sean Forman
> As Mr Strong has noted you have hit the nail on the head. Which is it? I
> think it's up to BBBA to tell us. My experience reading about Q-MAX is
> somewhat the same as David and Tom's who seem to be very knowledgeable
> guys who have posted here for quite some time. When I first read about
> it made little sense and seemed like a long climb to get anything useful
> out of it without extended work. Upon further reading it seemed to boil
> down to walk and hit prevention. So what? Is this supposed to be
> something I _can't_ understand without jumping through a bunch of hoops
> to get there? As Tom said, I do that everday when I read boxscores and
> every week when I read player stat lines.
> Cris Whetstone ** To reply remove the * from my address**
IMO the value of Qmax comes in showing the variety in starting pitchers
where traditional season long statistics may show none. A pitcher who
oscillates between game logs of
IP H R ER BB SO
8 4 0 0 4 7
and 4 8 6 6 2 3
may appear to be the same as the pitcher who throws
6 6 3 3 3 5
every game, but they really aren't. Now SNWL does a nice job of handling
this effect, but it treats Steve Woodard and Kerry Wood with a similar
result. Qmax teases apart those differences and brings out differences
that may not have been visible on a gander at the end of the year stat
line. Knowing that so and so had a wide variance in his effectiveness
probably means something. And knowing that Greg Maddux's control fell off
the table last September appears to have some meaning as well.
Now I know you can argue that, "Well you don't need Qmax when you can look
at the raw numbers." Well, that may be right, but by discretizing the raw
numbers and looking at things start by start, I've found that you become
much more attuned to such patterns and changes in performance.
Greg Spira
says...
>
> As far as I know (I hope that Don won't be upset about my saying this), a
> reason why there might be animosity toward STATS is because STATS has
> continuously sidetracked BBBA's attempts to purchase data from them; no quotes,
> no outright refusals, no nothing, just continous delays.
Oh, that's just STATS' normal way of doing business. Nobody should think
there's anything personal about STATS being uncooperative and
incompetent; that's just their corporate policy.
>
> Personally, we have also run into this poor treatment by STATS while trying to
> purchase data for The IBL Game (for more details, see http://www.ibl.org --
> it's a not-for-profit simulation game for use in our league). I'm thus also a
> bit upset at STATS in general.
Oh, there are so many stories I could tell.....
It is really sad, as far as I'm concerned, but at this point STATS is a
company with little vision and no respect for its non-corporate
customers or other people in the "industry." And it's never going to
change until its leadership does.
I still like STATS' books a lot - they may be shovelware, but they are
definitely useful shovelware - but I make it a point of not buying any
book directly from STATS that I don't have to; the aggravation just isn't
worth it. It's still hard for me to believe that a company which,
because of my interest in their output, so easily captured me as a
customer, worked so hard for so long to drive me away.
Greg
Greg Spira
>
>
> Of course, from baseball books from the authors as I've come
> to know them here, there are FAR too few *numbers*. Thanks
> for the analysis, guys, but dammit, provide the raw data
> *before* you change them. *I'll* figure out how I want to
> use them.
But, like, why? There are a zillion other places and books to get the
data. Why should either book waste space and money by presenting the raw
data when the entire purpose of the books is to adjust them?
It's like asking a movie be a drama instead of a comedy.
>
>
> My bias lies in untranslated data. The readers of these
> books (I'd think) understand park effects.
Yes, most probably understand park effects. But I don't know anyone who
can do translations in their heads. When I look at the BP during the
season, it's almost always to look at the translation of someone's
numbers at Wichita into a form that I can easily understand.
> I mean, geez, if
> you don't provide the raw data, how can I tell if you did
> the freakin' math right. There's plenty of other typos.
>
The math in these books is produced totally separate processes than the
writing; I suspect that there are almost no errors of this type in either
book.
Greg
Greg Spira
100...@green.weeg.uiowa.edu>, for...@blue.weeg.uiowa.edu says...
>
> What I have noticed about Neyer that bothers me is how lax he is on
> checking out the numbers. He has data that many of us would kill for
> one season to the next?
Doees Rob really have access to all that much more than we do? Maybe
current stats, somewhat, if he still has full STATS access, but certainly
not historical data.
> I'm sure someone can dig up an example of me
> doing this, but I dislike it when baseball writers (sabermetricians in
> particular) make minor unsubstantiated claims that would be easy to back
> up or major unresearched claims that they choose not to spend the time
> backing up.
Well, yes. I try to avoid that by often putting qualifiers on what I
say, unless I'm sure. Neyer just the other day used the mythical 600
Club decreased Fenway homers story; I sent along a correction. It
happens; I don't think it's that big a deal. I worry more about writers
getting the bigger picture right.
>
> My other pet peeve is more particular. As many times as Neyer has talked
> about closers and relievers in the last two years, he has not once
> mentioned Doug Drinen's tireless and groundbreaking work in this area.
I know you may find this hard to believe in light of Neyer's mentions of
the BP, but ESPN.com is very anti-link to things off their site, so I
think he does work under some restrictions. I've only heard good things
about Doug's work, myself, but my impression (last season) was that it
doesn't get updated often during the season, and for that reason I
haven't looked at it in-depth.
Greg
NawrockiT
: IMO the value of Qmax comes in showing the variety in starting pitchers
: where traditional season long statistics may show none. A pitcher who
: oscillates between game logs of
: IP H R ER BB SO
: 8 4 0 0 4 7
: and 4 8 6 6 2 3
: may appear to be the same as the pitcher who throws
: 6 6 3 3 3 5
: every game, but they really aren't. Now SNWL does a nice job of handling
: this effect, but it treats Steve Woodard and Kerry Wood with a similar
: result. Qmax teases apart those differences and brings out differences
: that may not have been visible on a gander at the end of the year stat
: line. Knowing that so and so had a wide variance in his effectiveness
: probably means something.
Sure, that's nice, but how many pitchers is it relevant to? Are there a lot of
cases where, say, it turns out that Pat Rapp was actually a more effective
pitcher than El Duque? And if the whole point of the thing is to discern
consistency, then the claims being made for it are overblown.
: And knowing that Greg Maddux's control fell off
: the table last September appears to have some meaning as well.
Of course it does. I can also look at Greg Maddux's numbers on the STATS AOL
site and see that his walks by month went 2, 8, 3, 8, 9, 12, despite the fact
that he made one fewer start those last two months. Now that means something to
me, more than "his September starts produced a C rating of 1,7 and fell into
the Q13 zone" or whatever Don Malcolm would say about it.
The point is not that QMAX is invalid or useless. The point is that if it has a
benefit, it is a marginally incremental benefit over more traditional and
accessible numbers. So except for those few instances where it's more useful
than traditional measures, why bother?
: Now I know you can argue that, "Well you don't need Qmax when you can look
: at the raw numbers." Well, that may be right, but by discretizing the raw
: numbers and looking at things start by start, I've found that you become
: much more attuned to such patterns and changes in performance.
I'd be able to respond better if I knew what "discretizing" meant. But again,
on the STATS site, I can look at Maddux's game logs for each start and see that
he walked 4, 2, 3, and 3 guys in his last four starts in 1998. And I'm
completely attuned to his changes in performance.
Part of the problem is that having invented the thing, Don Malcolm now wants to
use it to prove everything there is to know about pitching. He can't use
anything else, even when something else makes more sense. From the BBBA site:
"As our display of his 1995-98 QMAX matrix boxes confirms, Nomo actually had a
higher percentage of S12 games in 1998 than in either 1996 or 1997, but his
command was completely AWOL more than a third of the time." Here's what I'd
say: "His walks per nine innings since 1995 go 3.67, 3.35, 3.99, 5.37. His hits
per nine innings were a still-good 7.44 in 1998. Obviously, his control is the
problem." Which is more understandable?
Also from the BBBA site: "QMAX may have some additional predictive value that
would have a significant impact on notions of pitcher development. Currently,
the state of this research is very limited in scope and acknowledged to be
problematic at best. QMAX may be a tool that can help clarify the issues
involved in pitcher development. Here at BBBA, we're hard at work on conducting
additional research to determine just how helpful QMAX may prove to be."
In other words, this thing might just turn out to be useful, but we haven't
figured out yet if it is. This makes their criticism of Pitcher Abuse Points as
not peer-reviewed and fully tested simply churlish. If you don't know if QMAX
is "problematic" or "helpful," why is it splashed all over your work?
: later,
: sean
Your Iowa Farm Report, Sean, is the best thing on the site. Readable and
useful. I really enjoy it.
Tom Nawrocki
Mosey
niep...@alumni.princeton.edu wrote:
> >Frankly, I don't see this as too much of an excuse. Say he puts in
your
> >average 8 hour day. If he checks that one (admitdly harder than most
> >facts) fact over an hour, he has 7 other hours to do his thing.
>
> Well, out of curiosity, do we know what "his thing" is? I'm curious.
I
> know Gammons is a full-time columnist. All he has to do is talk to
> sources and write articles. OTOH, I can't imagine that it's a
full-time
> job for Neyer to write his columns. (If it is, I want it.)
Neyer does write more than one column. Not on a daily basis, but he
pulls his weight around the website. Stats class, Doesn't he still do a
point-counterpoint for the Midweek match up? Fantasy articles etc. Five
or six columns, one daily, I could see a full-time job there.
From what I can gather he moved to Seattle a year or two ago from he
east coast to do his job better. If that is true, I don't see him moving
3000 miles away for a part-time job or because he is a Mariners fan.
<snip>
But
> there are *lots* of factoids in a column, and if he has to spend an
hour
> looking for every one, that's a big deal. Especially in the specific
> scenario we're talking about. If he spends an hour and there's
nothing to
> find, that's a pretty big investment for one sentence. If he spends
an
> hour and *does* find that team, so what? Will it change the point of
that
> particular column? I don't think so.
No, but I didn't mean this factoid in particular. I was referring to his
general errors that he seems to be making. You and I both know that most
of the factoids he has take less than an hour. How many teams have lost
more than X games in a season, best F% for a RF and so on. Some are real
doozies obviously, but those are also the ones he writes whole columns
on.
> >I don't know if he has an intern or something doiung research for
him,
> >but as Sean said, he has resources most of us drool over.
>
> I sincerely doubt that. He works for ESPNet, not Microsoft.
Doubt what? The intern or the material? He may not have an intern, but
is isn't as if he couldn't get a pro-bono one. I'm sure he he sent out a
calling for interns he would get a ton.
It's not uncommon, he is a direct James prodigy, and if I remember James
had a few freebe researchers.
As for the extra stats. I'd be prtty sure he has full Stats access,
probably half a dozen 2000 page resources, his personal collection of
material, and of course, direct access to people who can look things up
for him and probably jump at he chance (he has done that to a couple of
regulars here if I remember). I don't know if anyone else has that kind
of thing laying around, the last one in particular.
>
> My recollection is similar to yours, but that hardly proves anything.
> After all, when he first started writing, it was so nice to see an
> alternative to Gammons Et Al that minor mistakes were unimportant.
Now
> that we take him for granted, we want to see perfection.
Good point. I wouldn't think of scratching my head on "the polar bear
affect" or his small market stuff two years ago.
--
John Mosey FIRE BANDO!
"I drive too fast, talk too loud and I'm an
asshole. Deal with it."
http://www.mosey.com Mosey's Fantasy Baseball
Sean Forman
> 100...@green.weeg.uiowa.edu>, for...@blue.weeg.uiowa.edu says...
> > What I have noticed about Neyer that bothers me is how lax he is on
> > checking out the numbers. He has data that many of us would kill for
> > and yet he can't check if any team has ever doubled their walk totals from
> > one season to the next?
> Doees Rob really have access to all that much more than we do? Maybe
> current stats, somewhat, if he still has full STATS access, but certainly
> not historical data.
> Greg
That's an interesting question that I certainly don't have an answer to.
I know Art Martone has referred to a STATS inc package that his newspaper
has. I would imagine he has something like that. I know that ESPN
broadcasters seem to be able to get full game log data on all players
majors and minors. They are always quoting esoteric split data on
SportsCenter that I wouldn't be able to get, and stuff like he is the
first player since 1950 to do this. I'm not sure what he has access to,
but I would be a bit surprised if he didn't have access to everything ESPN
proper has. Also, I realize that not everyone knows SQL (standard
querying language) and doesn't realize that on a good database you can do
stuff like find if any teams doubled their walk totals in about five
minutes.
My in-laws-to-be lived in Evanston, Illinois for up until a year ago.
During one visit I called STATS,inc. to see if they had company tours, you
know to see the data. The receptionist didn't quite know what to make of
me.
Hijacking this thread further, I think there needs to be an effort to
fully define biographical information for every professional baseball
player ever. In addition to this, someone should develop a standard set
of team and league abbreviations that uniquely define every league and
baseball city. This information could then be used to define standard
keys for every player and team. If one could convince people to use this
you would then have powerful compatibility across various projects like
Retrosheet, Lahman's DB, Pat Doyle's DB and other sources. You would be a
simple query from knowing the age of every player in the Eastern League in
1980. Just something on my wish list.
later,
sean
Sean Forman Program in Applied Mathematics U. Of Iowa
"When I hear their banjo music at the beginning, I have a
sudden burst of happiness." --Sylvia speaking of CarTalk
vi...@baseball.org
> In article <92724151...@news.remarQ.com>, acd...@intrex.net says...
> > I mean, geez, if
> > you don't provide the raw data, how can I tell if you did
> > the freakin' math right. There's plenty of other typos.
>
> The math in these books is produced totally separate processes than the
> writing; I suspect that there are almost no errors of this type in either
> book.
Well, there was Travis Lee's Vlad projection a year or two back, which
was based on the wrong age. I don't have my BP '98 with me, but the
writer of the player comment line noted the error, but said something
like, "But I'm not going to correct the error, because this line looks
pretty good to me."
--
/---------------------------------------------------------------\
| Vinay Kumar |
| vi...@baseball.org http://www.baseball.org/~vinay |
\---------------------------------------------------------------/
vi...@baseball.org
> Part of the problem is that having invented the thing, Don Malcolm
> now wants to use it to prove everything there is to know about
> pitching. He can't use anything else, even when something else makes
> more sense. From the BBBA site: "As our display of his 1995-98 QMAX
> matrix boxes confirms, Nomo actually had a higher percentage of S12
> games in 1998 than in either 1996 or 1997, but his command was
> completely AWOL more than a third of the time."
My favorite was a comparison between two pitchers in BBBA '99 (I don't
remember who; I didn't buy the book, but was thumbing through it in a
bookstore). They had very similar QMAX scores, but their ERA's
differed by quite a bit. The author said something like, "See, ERA
doesn't tell you everything. By looking at the QMAX you can see that
these guys were actually pretty similar last year." Of course, the
real difference between the two was extra-base hit prevention, as one
pitcher had some trouble with the gopher ball and the other didn't. I
was laughing hysterically at that one. I wish I could find it again.
> Your Iowa Farm Report, Sean, is the best thing on the site. Readable and
> useful. I really enjoy it.
Agreed.
And Don Malcolm's charts about 3-start pitch counts were actually very
interesting. Of course, they were buried beneath a few hundred words
of completely disgusting PAP-bashing.
Greg Spira
> nawr...@aol.com.not (NawrockiT) writes:
>
>
> > Your Iowa Farm Report, Sean, is the best thing on the site. Readable and
> > useful. I really enjoy it.
>
> Agreed.
>
magnifying glass I assume everybody else's copy came with, so I found
that I couldn't read any of Sean's stuff in the book.
Greg
Ron Johnson
Likewise.
>
>It is really sad, as far as I'm concerned, but at this point STATS is a
>company with little vision and no respect for its non-corporate
>customers or other people in the "industry." And it's never going to
>change until its leadership does.
>
>I still like STATS' books a lot - they may be shovelware, but they are
>definitely useful shovelware - but I make it a point of not buying any
>book directly from STATS that I don't have to; the aggravation just isn't
>worth it. It's still hard for me to believe that a company which,
>because of my interest in their output, so easily captured me as a
>customer, worked so hard for so long to drive me away.
Funny. There are people who can order from them and get what
they ordered (not a given) on a timely basis.
Never happened to me.
To bring this back to BBBA though. It's worth noting that Don
edited out a bit of an anti-STATS rant of mine.
Essentially STATS has become Elias. And it's going to meet the same
response. Sooner or later there's going to be a Project Scoresheet II.
--
RNJ
Ben Flieger
Greg Spira wrote in message ...
>But, like, why? There are a zillion other places and books to get
the
>data.
Indeed. But I would pay good money to either have A)more adjusted
lines(in BP '98 it goes back to 1994) or B)unadjusted stats. I can
live without a witty comment regarding Mike Glavine getting the same
strike zone his brother does if I can see more lines for an older
player or can see what Player X really did.
Oh, this might be in BP '99, mine is on backorder, but could you
please have an adjusted stats leader board? 1 page, that's all. You
managed to essentially waste a page telling me such facts that Jeff
Berblinger was the second best 2B in the American Association with the
glove, I'm more interested in where Edgar Martinez ranks in EqA.
Why should either book waste space and money by presenting the raw
>data when the entire purpose of the books is to adjust them?
>
>It's like asking a movie be a drama instead of a comedy.
I think it's just asking for fleshed out adjustments. I suppose I
could flip between Neft+Cohen and the BP, but it would be easier if
you had the unadjusted stats. Adjusted FROM what?
>> I mean, geez, if
>> you don't provide the raw data, how can I tell if you did
>> the freakin' math right. There's plenty of other typos.
>>
>The math in these books is produced totally separate processes than
the
>writing; I suspect that there are almost no errors of this type in
either
>book.
I don't think your accuracy is bad, I just like the data.
Ben Flieger
vi...@baseball.org wrote in message ...
>My favorite was a comparison between two pitchers in BBBA '99 (I
don't
>remember who; I didn't buy the book, but was thumbing through it in
a
>bookstore).
Jose Lima and Randy Johnson, IIRC.
They had very similar QMAX scores, but their ERA's
>differed by quite a bit. The author said something like, "See, ERA
>doesn't tell you everything. By looking at the QMAX you can see that
>these guys were actually pretty similar last year." Of course, the
>real difference between the two was extra-base hit prevention, as one
>pitcher had some trouble with the gopher ball and the other didn't.
I
>was laughing hysterically at that one. I wish I could find it again.
I don't quite understand the point of QMAX.
It's a start by start(rather than average per inning) WHIP, but that's
only important in telling me whose WHIP is consistant. It doesn't tell
me anything about quality, because the goal of pitching is not to
prevent hits and walks, but to prevent runs. Preventing hits and walks
are means to an end. But Don Malcolm will go on and on about the
virtues of QMAX. I don't get it, and this thread hasn't enlightened
me, it's only shown that other people are as confused as I am
regarding QMAX's value. To top it off, QMAX is frighteningly dense for
such a simple stat. When I read passages with tons of (x, y) couples
and numerous references to QMAX S or C, I can't get to the point I
presume was going to be made.
Of course, it's not hard to solve the problem you cite. Replace hits
with total bases and you've covered the power aspect. That's actually
something that I haven't seen before, and I think TB is a more
accurate "Stuff" stat.
dtk
Why doesn't anybody tabulate pitchers' OPS allowed?
ddr...@imap1.asu.edu
: Why doesn't anybody tabulate pitchers' OPS allowed?
The BBBA does, for relievers.
Doug
Dan Schmidt
| Why doesn't anybody tabulate pitchers' OPS allowed?
OPS is an attempt to combine lots of information into one big number
that summarizes how good a hitter is.
In the case of pitching, we already have a fine 'one big number that
summarizes how good a pitcher is'; it's ERA (though some people prefer
to just use Run Average). So the main reason for OPS doesn't really
exist.
I can think of a couple reasons to look at POPS, though:
1) It might be interesting to compare POPS with ERA, and find out if
any pitchers consistently allow more or fewer runs than their POPS
would predict.
2) ERA isn't the greatest metric for relievers, who should probably be
penalized to some degree for inherited runners who score. So POPS
might be useful for measuring their performance.
J. Edward Tuttle
> > All of the numbers in BP may be translated, but at least they have
> > the same metrics as those of any normal sports fan. Translated ERA
> > may be different than Raw ERA, but its still in the same metric.
and Doug Drinen replied:
> I'm not sure I understand your use of the word "metric." I think I
> can answer your question, though.
Ira is obviously an educated person. I used to think "metric" meant the
metric system -- until I took some geometry course where the guy kept
talking about "metrics." Then I had to look it up.
Here, I think Ira means "units of measure." No problema. When in Rome,
talk like the Romans do.
Same thing goes for "rubric." Once it had something to do with
prescribed procedures in church ... until it became part of
education-speak. Now it has to do with rules for grading exams.
Once upon a time, "to obtain" meant to get something. Then the academic
social scientists started talking about things that "obtain." For them,
the word means "exist."
All those are marks of the intelligentsia. You got to learn the
jargon. And it's not limited to academia, either. In bizness-speak,
people "utilize" things. Once upon a time, they just "used" things.
But no harm, no foul. Nothing wrong with that stuff. Much worse are
those writers who can't distinguish between "affect" and "effect." That
one really rankles me, and I see 'em mixed up all the time. Do you
"affect" something, or do you "effect" it?
And then there's "ensure" versus "insure." One of 'em is a dietary
supplement, I think. Maybe that's what McGwire was on last year.
James
J. Edward Tuttle
>
> dtk <kur...@musc.edu> writes:
>
> | Why doesn't anybody tabulate pitchers' OPS allowed?
>
> OPS is an attempt to combine lots of information into one big number
> that summarizes how good a hitter is.
>
> In the case of pitching, we already have a fine 'one big number that
> summarizes how good a pitcher is'; it's ERA (though some people prefer
> to just use Run Average). So the main reason for OPS doesn't really
> exist.
That's wrong, because the ERA has several flaws.
1. The "inherited runner" problem.
2. The possibility that a two-out error can lead to a half-dozen
unearned runs.
3. The fact that the results of fielding errors should be measured in
unearned bases, not unitary unearned runs.
Doug Drinen addresses these, and so do I, in my Earned Runs Per Game
(ERPG) stat in Base Production.
J. Edward Tuttle
> I don't quite understand the point of QMAX. [snip]
> ... the goal of pitching is not to prevent hits and walks, but to
> prevent runs. Preventing hits and walks are means to an end.
The pitcher's objective is to prevent batters from becoming runners, and
to prevent runners from advancing on the basepaths. To the extent
pitchers are able to do that, they prevent runs from scoring.
> Of course, it's not hard to solve the problem you cite. Replace hits
> with total bases and you've covered the power aspect. That's actually
> something that I haven't seen before, and I think TB is a more
> accurate "Stuff" stat.
Try Total Bases Allowed (including walks and baserunner advancement)
minus Stranded Runners. That correlates directly with Runs Allowed. In
fact, that's the rationale behind my Base Production method.
Jim Furtado
1) Since part of this thread revolves about Rob vs. Robb, I'd like to point
out that Don's last name is Malcolm.
2) Like Sean and Doug, I contributed to the 1999 BBBA. Unlike them,
however, I won't admit to having skin any thinner than your average,
run-of-the-mill, circus elephant.
3) It's too bad that Don isn't around because I'm sure he'd have a few
comments. I guess we'll all have to wait a few more days until this thread
really gets going.
4) BP and BBBA are two entirely different books.
5) Yes, there are a too many acronyms in the 1999 BBBA.
6) There is a lot of stuff in BBBA. I think this is because Don wants to
present different views on player analysis. He doesn't want to just
regurgitate the same perspectives that readers can find in other places.
Does this lead to acronym overload? See number 5.
7) Don assumes responsibility for more tasks involving the book than any one
person can possibly handle.
8) The book needs a good copy-editor. See number 7.
9) Bill James' Abstacts were a lot less technical than BBBA.
10) James' Abstracts were a lot easier to read than many parts of BBBA.
11) Bill James is a much, much better writer than I am.
12) I expect to have a lot of fun watching Andruw Jones play over the next
few years.
13) There is a lot of stuff in BBBA. I know I said this already, but I
think this point gets lost.
Well, I have a few other thoughts, but I'm tired and my wife wants a little
attention.
I look forward to reading any and all comments (See #2.)
Gotta go.
Jim Furtado
jfur...@baseballstuff.com
Keri Olsen and Arne Olson
Ben Flieger wrote:
>
> Greg Spira wrote in message ...
> >But, like, why? There are a zillion other places and books to get
> the
> >data.
>
> Indeed. But I would pay good money to either have A)more adjusted
> lines(in BP '98 it goes back to 1994) or B)unadjusted stats. I can
> live without a witty comment regarding Mike Glavine getting the same
> strike zone his brother does if I can see more lines for an older
> player or can see what Player X really did.
>
> Oh, this might be in BP '99, mine is on backorder, but could you
> please have an adjusted stats leader board? 1 page, that's all. You
> managed to essentially waste a page telling me such facts that Jeff
> Berblinger was the second best 2B in the American Association with the
> glove, I'm more interested in where Edgar Martinez ranks in EqA.
This is a good suggestion, and I'd take it a step further and ask for a
position-by-position leader board.
>
> Why should either book waste space and money by presenting the raw
> >data when the entire purpose of the books is to adjust them?
> >
> >It's like asking a movie be a drama instead of a comedy.
>
> I think it's just asking for fleshed out adjustments. I suppose I
> could flip between Neft+Cohen and the BP, but it would be easier if
> you had the unadjusted stats. Adjusted FROM what?
FWIW, I'd like to chime in here in favor of the adjusted stats. There
are literally dozens of places to get the raw stats. I enjoy being able
to compare everyone in both major leagues and several minor leagues all
on the same scale. I find it makes it really easy to compare two guys
who are rumored to be involved in a trade, or to know what to expect
when someone switches teams in the off-season. I've fully internalized
the concept of park and league adjustments (I think one has to have in
order to appreciate the analysis in BP), but I still can't do them in my
head. I don't think including the raw stats would add much except bulk.
Arne
Chris Conley
Jim Furtado <jfur...@baseballstuff.com> wrote:
[ Lots of reasonable comments about BBBA and this thread. ]
>13) There is a lot of stuff in BBBA. I know I said this already, but I
>think this point gets lost.
It seems to me that many of the complaints, particularly those about the
web site, can be summarizes as the inverse of this statement:
There is a lot of stuff that is *not* in BBBA -- and some of the authors
of that work seem loathe to admit that.
The whole jihad over PAP is an example. The concept behind PAP is, to
put words in Rany's mouth, that the risk of damage to a pitcher's arm
gets progressively higher per pitch as the pitch count increases within
a single start. The concept between the three-start metric is that the
risk of damage to a pitcher's arm is related to the stress placed on the
arm across a sequence of consecutive starts.
These are *NOT* conflicting viewpoints; it's pretty simple to see that
they could possibly be combined into a three-start PAP count, which may
be the best of both worlds. Or, of course, it may not; it may well be
the case that one or the other adds nothing to the overall study of the
impact of heavy workloads on pitchers. In any case, it's certainly too
soon to laud or dismiss either metric; time will tell which aspects of
each are actually important and which provide little useful information.
But, as far as I can tell, the author of that piece didn't care to
consider any of the possible advantages that PAP might provide to an
analyst. Rather, he went straight to vague implications about the (lack
of) meaning of the measure and even attacks on the integrity of its
creator rather than intelligently disecting the measure.
That's NIH (Not Invented Here) syndrome at its finest, and it's an ugly
sight, especially from a source which generally does produce fine work
and intelligent commentary.
- Chris
Keri Olsen and Arne Olson
FYI, I've cross-posted below a discussion from
alt.sports.baseball.sea-mariners. The PAP-bashing has really irritated
me, so I'm glad to see it being raised here. I don't think PAP is "The
Answer", but I do think it's a good start towards putting together a
possible diagnostic tool. I agree completely with Chris that PAP and
the 3-start model are based on different views of how pitching damages
the arm. One view is that a single start can be damaging, the other
that a sustained workload over a number of starts is what does the
damage. I, for one, haven't seen any empirical evidence that one method
is superior to the other, and would really like to wait for about five
years before we start claiming that the other guy's method sucks. And
I'd be happy if we *never* got into personal attacks.
Dylan Bumbarger:
>
> Ken Adams wrote an article last week, at
>
> http://www.backatcha.com/homepage/ohrobb01.html
>
> which deals with this subject. I think we're overreacting way too
> much on this stuff.
Arne Olson:
Whatever. As part of an ongoing attempt to discredit both "Robb" Neyer
and BP, Ken Adams takes Neyer to task for using the PAP method to
criticize Bobby Cox. The main complaint is that there is no empirical
basis for concluding that a single 120 pitch game is dangerous to a
young pitcher. Not said, but implied, is that an alternative model, one
that relied on, say, a three-start running average, would be superior.
Of course, there's no empirical evidence that three starts is the
appropriate increment, but that's not talked about.
This might be more credible if Adams didn't attempt to attribute the
outing in question, also with no evidence, to Cox's "testing" Perez.
According to Adams, "that's part of what a manager does." Maybe, maybe
not. There's evidence that accumulations of pitch counts above 100 are
bad, but no research on whether there's some minimum threshold of
exposure that's safe. That's like saying we know cigarettes cause
cancer, but we haven't ruled out that only smoking 3 a day is safe.
That still doesn't make smoking 3 a day a good idea.
PAP is a start towards accumulating that body of evidence. For BBBA to
sneer so contemptously at it strikes me as a case of throwing rocks in
glass houses.
Dylan Bumbarger:
>
> :PAP is a start towards accumulating that body of evidence.
>
> OK. Despite the fact that you took the most negative view possible of
> the article, your statements are quite reasonable. However, I totally
> disagree with the above. *Pitch counts* are a start toward
> accumulating that body of evidence. PAP, because of the nature of
> their construction, are not.
Arne Olson:
Pitch counts are the raw data. By themselves, they don't tell us
anything. We can use them, however, to construct statistics, like mean,
median, standard deviation, etc. How you do this depends on how you
think the physical process works. If you think each pitch does
irreparable damage, for example, you'd just look at cumulative pitch
counts over a career. If you think the "loading" of pitches over a
fixed period of time is what causes damage, you'd come up with some
mean-based system, perhaps a running average.
Jazayerli used a fairly common concept in epidemiology, which is the
"acceptable loading". The idea is that there is some level of exposure
to the hazard below which essentially no permanent harm is done. EPA
uses this method to determine acceptable levels of pollutants in air and
water. There can be and often is disagreement about just what level of
exposure is harmful, and that's generally the subject of lots of
research. But it's pretty well accepted that the marginal gains from
reducing pollution not only decline but in fact go to zero at a given
point.
Jazayerli used a similar approach to design PAP. He assumed that
there's some number of pitches that is "safe", i.e., that results in no
incremental injury risk. I don't think there's any disagreement about
that in principle. I believe he cites Craig Wright's work for the
assumption that 100 pitches is safe. He also basically assumes that
risk increases geometrically beyond that point. Again, this is not
inconsistent with other epidemiological studies. Both of these
assumptions are subject to questioning and, hopefully, verification.
The only thing that will show whether PAP, or the three-start model for
that matter, is a useful statistic is whether it can be used to predict
injury. We won't know that for several years. If we find that it
doesn't predict injury very well, some modification would obviously be
in order. Maybe 95 pitches is a better cutoff point, or maybe a
three-start model is better, or maybe sort of a combined three-start
PAP, where there's a logarythmic scale above 95 pitches.
Again, at this point we're sort of at that 3-cigarette phase, where we
know overwork is bad, but we don't know much beyond that. There's no
empirical evidence that either method is useful at all, much less than
one is superior to the other. I don't see any reason for one party to
assume a sneeringly superior pose or try to run down the work of the
other. Why don't we just let the better model win?
Arne
Chris Dial
<374644B5...@ix.netcom.com>...
Where can I get them to read on a plane? And I'd like the
data available in a single tome. Yes, I have both of STATS
books as well, but I'd rather that as one book.
>I enjoy being able
>to compare everyone in both major leagues and several minor
leagues all
>on the same scale. I find it makes it really easy to
compare two guys
>who are rumored to be involved in a trade, or to know what
to expect
>when someone switches teams in the off-season. I've fully
internalized
>the concept of park and league adjustments (I think one has
to have in
>order to appreciate the analysis in BP), but I still can't
do them in my
>head. I don't think including the raw stats would add much
except bulk.
Well, I can do that in my head. I usually adjust by
park/town rather than just league. The acceptable range for
adjusting stats is wide, and I'd like more analysis. Keith
Woolner's stuff *was not* adjusted and hands down the best
information in the book.
Maybe I'm just the wrong audience, but I would much prefer a
book filled with articles by Woolner, Grabiner, Johnson.
The best part of the Abstracts and Elias' Analysts is the
articles...I don't buy them to look at the pictures.
Personally, I wish one of the three would bother
*publishing* those articles rather than just at
stathead.com. Heck, Arne's work with the RBI vulture would
be useful to have bound (although *I* think it provides a
picture of what may happen, rather than describing what does
happen).
Chris Dial
>
>
>
>Arne
Chris Dial
>>(misplaced attribute, but someone else said:)
>>
>> I sincerely doubt that. He works for ESPNet, not
Microsoft.
>
>Doubt what? The intern or the material? He may not have an
intern, but
>is isn't as if he couldn't get a pro-bono one. I'm sure he
he sent out a
STATS pays 5 dollars for guys to score, and I'm betting
dollars to donuts ESPN has a database (like Lahman's) out
the wazoo. I used to write my own cheesy newsletter, and
I'd have the resources to look that up.
>
>It's not uncommon, he is a direct James prodigy, and if I
remember James
>had a few freebe researchers.
>
>As for the extra stats. I'd be prtty sure he has full Stats
access,
>probably half a dozen 2000 page resources, his personal
collection of
>material, and of course, direct access to people who can
look things up
>for him and probably jump at he chance (he has done that to
a couple of
>regulars here if I remember). I don't know if anyone else
has that kind
>of thing laying around, the last one in particular.
Here I may disagree. I've checked a few stathead mantras
(like minor league stats predictive value), and get plenty
of help from Dan Szymborski, Ron Johnson and Dale
Stephenson. I'd think any of us working on something should
feel comfortable asking one another for research assistance
(it's definitely a SABR thing).
Chris Dial
Chris Dial
Hey! I've got dibs on this complaint!
>
>OPS is an attempt to combine lots of information into one
big number
>that summarizes how good a hitter is.
>
>In the case of pitching, we already have a fine 'one big
number that
>summarizes how good a pitcher is'; it's ERA (though some
people prefer
>to just use Run Average). So the main reason for OPS
doesn't really
>exist.
>
>I can think of a couple reasons to look at POPS, though:
>
>1) It might be interesting to compare POPS with ERA, and
find out if
> any pitchers consistently allow more or fewer runs than
their POPS
> would predict.
>
>2) ERA isn't the greatest metric for relievers, who should
probably be
> penalized to some degree for inherited runners who
score. So POPS
> might be useful for measuring their performance.
I've gotten an explanation from David Grabiner before, but
my accepting his explanation is based solely on my respect
for his knowledge, so since I've never met him, occasionally
I doubt my faith, and ask: If OPS is the best q&d number for
hitters, then logic dictates that POPS (as named above)
*must* be the best q&d number for pitchers.
Many statheads say that a pitcher's job is to prevent runs,
not baserunners. Then the job of hitters is to score runs,
not get on base. Both are a means to an end, and one
requires the other.
Other than it isn't the easiest stat to come by, is there
really a reason not to prefer POPS? (Or as Ron points out
regularly in these times, OBP*2+SLG).
And *this* MF-stat is one I'd like in these books, that is
NOT "in a zillion places already". And it sure as shit
isn't in tabular form for ready comparison (park-adjusted
etc.)
Chris Dial
Derek
My two centavos worth:
As a layman in the worst sense of the word, I would be apprehensive to
take any of the theories given for pitchers whose arms end up requiring
surgery like Kerry Wood. All of this data that's being collected and put
into models, doesn't address some other important issues.
1. What are the arm mechanics of the pitcher in question?
Often it is poor arm mechanics that can lead to a pitcher's arm breaking
down. How will this be factored into PAP or the 3 start theory proposed
by BBBA?
2. What was the workload of the injured pitcher at a young age?
Pitch counts are given these days in just about every box score you see
online, but what about the workload of a pitcher when he's in high
school, or even further back to Little League? I know that some, if not
all, of the youth levels in baseball have rules to prevent this kind of
abuse, but I've also heard that one of the reasons proposed for Woods'
injury was that he was "abused" in high school. Will there be more than
just anectdotal evidence for this part of the equation?
I have also heard that it is during these times when a young pitcher's
body goes through his growing years that care be taken not to overwork
them, and even limit the number of breaking balls thrown because of the
pressure it puts on an elbow.
I'm sure there's more possible reasons for an injured pitcher's arm than
the two I've noted, but just with those two reasons cited, I would be
skeptical of any conclusions reached in a study that only had the data
of a pitcher at the major league level. It seems to me you'd be leaving
the majority of the years the pitcher in question has been throwing a
baseball. For instance, take the case of Kerry Wood. I don't know
exactly when he started pitching, but I would guesstimate that it was
around 11 or 12. So if you only take his major league stats into some
abuse theory, you're leaving out about 10 years worth of data.
As another poster likes to say whose name escapes me....
"Any and all corrections are appreciated", or something to that effect.
Derek
Keith Woolner
cited author.]
In article <7i78m0$85d$1...@remarQ.com>, acd...@intrex.net says...
> I've gotten an explanation from David Grabiner before, but
> my accepting his explanation is based solely on my respect
> for his knowledge, so since I've never met him, occasionally
> I doubt my faith, and ask: If OPS is the best q&d number for
> hitters, then logic dictates that POPS (as named above)
> *must* be the best q&d number for pitchers.
I originally had the same point of view as Chris -- use the same stats
to measure pitchers and batters (I think one of my early posts to r.s.b
advocated using pitcher's RC instead of OPS, as Chris has done). I
eventually changed my mind, and internally I understood it using the
kind of reasoning below.
Restating Chris's key point:
> If OPS is the best q&d number for hitters, then logic dictates that
> POPS *must* be the best q&d number for pitchers.
Not necessarily true. One of the key differences between pitchers
and hitters is that hitters only get their opportunties sporadically
during a game, and when they come up to bat, they face a situational
context that they had little to no influence on (i.e. the base-out
situation is due to the work of the other batters on the team and
the opposing pitchers).
A pitcher, on the other hand, faces a consecutive sequence of
batters, so any situation faced by the pitcher is substantially
dependent on his own performance (let's say starting pitcher, for
the sake of argument. It's less true for relievers coming into
the game with runners on bas).
IOW, opportunities an individual batter faces are isolated
and unrelated to his own skill, but the opportunities a pitcher
faces are part of a a sequence where he directly participates in
each step in a significant way.
Using OPS removes the situational context from the batter's
performance by looking simply at the value-weighted outcomes of
each plate appearance. Since the context is beyond the control
of the batter, this is appropriate.
Using ERA (or RA) preserves the situational context of the pitcher's
performance by considering the outcomes of the entire sequence
of events that the pitcher participated in. This is appropriate
since the pitcher had a direct hand in creating the string of events
that led to run scoring or prevention.
> Many statheads say that a pitcher's job is to prevent runs,
> not baserunners. Then the job of hitters is to score runs,
> not get on base. Both are a means to an end, and one
> requires the other.
I wouldn't argue with either end of the statement (pitcher = preventing
runs, hitters = scoring runs), unless we nitpick and say that their
job is to maximize the likelihood of the team winning the
game/season/World Series. But each job faces a different set of
circumstances in trying to accomplish their jobs, so it's not
inconceivable that different metrics are better suited for assessing
the success or failure for each role.
> Other than it isn't the easiest stat to come by, is there
> really a reason not to prefer POPS? (Or as Ron points out
> regularly in these times, OBP*2+SLG).
I think the situational-context difference captures most of the reasons.
What I like to do when confronting a question like this
is to imagine a variant of baseball where the rules are different,
and would lead us to a different answer. Contrasting the two
often yields insight.
So let's try to imagine a variant of baseball where OPS isn't
the best q&d way to measure batter performance. Imagine a
game of baseball in which there are nine batters A B C ... H I, but
instead of batting one after another, each batter is solely responsible
for all plate appearances for an inning. That is, A bats in the first inning,
and keeps batting until three outs are made (imagine that either pinch
runners are used on the basepaths, or that clones of A occupy the bases,
or use the "ghost runners" rule that we often used as kids when we were
short-handed). After three outs are made, A doesn't bat again during the
game. B gets all of the plate appearances in the 2nd inning, C in the 3rd,
and so on.
In this variation, each batter participates in a sequence of plate apperances
that leads to run scoring, and the context of each plate appearance depends
solely on his own prior performance. There's no interaction with the
offensive prowess of his teammates. You can look at a batter's inning in
isolation and know exactly how many runs were produced, and attribute them
entirely to that batter's efforts.
In a game like this, there's no reason to rely on OPS, since run scoring
can be measured directly. We'd probably have a stat called something like
BRA (Batter's Run Average) that it analagous to ERA or RA for pitchers.
However, in real baseball, batters don't control the situational contexts
they face, so we'd want a measurement that doesn't depend on them. OPS
is one such option, and one that's been widely adopted for other reasons
(availability of OBP and SLG in the media, and the mathematical simplicity
of simply adding them together).
Hope this helps.
--
Keith Woolner kwoo...@best.com http://www.best.com/~kwoolner
Baseball Engineering Library, http://www.stathead.com
List Owner, Red Sox mailing list, http://www.best.com/~kwoolner/redsox/list
Newsgroup Moderator, rec.sport.baseball.analysis
Author, Baseball Prospectus 1999 http://www.baseballprospectus.com
J. Edward Tuttle
inventing hypothetical rules that lend themselves to insightful thought
experiments. The following impresses me as a vintage Woolner tour de
force.
Keith Woolner wrote:
>
> Chris Dial writes:
> >
> > I've gotten an explanation from David Grabiner before, but
> > my accepting his explanation is based solely on my respect
> > for his knowledge, so since I've never met him, occasionally
> > I doubt my faith, and ask: If OPS is the best q&d number for
> > hitters, then logic dictates that POPS (as named above)
> > *must* be the best q&d number for pitchers.
>
> I originally had the same point of view as Chris -- use the same stats
> to measure pitchers and batters (I think one of my early posts to r.s.b
> advocated using pitcher's RC instead of OPS, as Chris has done). I
> eventually changed my mind, and internally I understood it using the
> kind of reasoning below.
I've thought through this stuff several times myself, and faced the same
issues.
> Restating Chris's key point:
>
> > If OPS is the best q&d number for hitters, then logic dictates that
> > POPS *must* be the best q&d number for pitchers.
>
> Not necessarily true.
Keith is correct here. The word "must" is too strong. "May" is
better. By rejecting the syllogism and the inevitability of the
conclusion, we should not reject the possibility that the conclusion is
nevertheless correct.
> One of the key differences between pitchers
> and hitters is that hitters only get their opportunties sporadically
> during a game, and when they come up to bat, they face a situational
> context that they had little to no influence on (i.e. the base-out
> situation is due to the work of the other batters on the team and
> the opposing pitchers).
>
> A pitcher, on the other hand, faces a consecutive sequence of
> batters, so any situation faced by the pitcher is substantially
> dependent on his own performance (let's say starting pitcher, for
> the sake of argument. It's less true for relievers coming into
> the game with runners on bas).
>
> IOW, opportunities an individual batter faces are isolated
> and unrelated to his own skill, but the opportunities a pitcher
> faces are part of a a sequence where he directly participates in
> each step in a significant way.
The observations in the three preceding paragraphs are correct, but they
don't necessarily support the conclusion that will follow.
> Using OPS removes the situational context from the batter's
> performance by looking simply at the value-weighted outcomes of
> each plate appearance. Since the context is beyond the control
> of the batter, this is appropriate.
>
> Using ERA (or RA) preserves the situational context of the pitcher's
> performance by considering the outcomes of the entire sequence
> of events that the pitcher participated in. This is appropriate
> since the pitcher had a direct hand in creating the string of events
> that led to run scoring or prevention.
These two paragraphs don't necessarily follow from the preceding three.
Imagine a game consisting of 36 PAs with no player substitutions. Each
of nine batters has four PAs, and each of four pitchers faces exactly
nine batters in a rotation. The starting pitcher faces batters 1, 5, 9,
4, 8, ...; pitcher 2 faces batter 2, 6, 1, 5, 9, ...
With these rules, pitchers don't create the situations they face. Now,
you say, this isn't baseball. Well, yeah. My only point is that you
aren't required to use the sequential nature of pitching; it's possible
to dissociate the pitcher from the situations he faces. And that means
that each pitcher-batter confrontation, including the concomitant
base-out situation, can be treated independently.
> > Many statheads say that a pitcher's job is to prevent runs,
> > not baserunners. Then the job of hitters is to score runs,
> > not get on base. Both are a means to an end, and one
> > requires the other.
>
> I wouldn't argue with either end of the statement (pitcher = preventing
> runs, hitters = scoring runs), unless we nitpick and say that their
> job is to maximize the likelihood of the team winning the
> game/season/World Series. But each job faces a different set of
> circumstances in trying to accomplish their jobs, so it's not
> inconceivable that different metrics are better suited for assessing
> the success or failure for each role.
Rule 5.04. "The offensive team's objective is to have its batter become
a runner and its runners advance."
Rule 5.05. "The defensive team's objective is to prevent offensive
players from becoming runners, and to prevent their advance around the
bases."
> > Other than it isn't the easiest stat to come by, is there
> > really a reason not to prefer POPS? (Or as Ron points out
> > regularly in these times, OBP*2+SLG).
>
> I think the situational-context difference captures most of the reasons.
Again, not necessarily, especially if you can account for or negate
those situational-context differences.
> What I like to do when confronting a question like this
> is to imagine a variant of baseball where the rules are different,
> and would lead us to a different answer. Contrasting the two
> often yields insight.
Yes, and Keith is a master at this.
> So let's try to imagine a variant of baseball where OPS isn't the best
> q&d way to measure batter performance. Imagine a game of baseball
> in which there are nine batters A B C ... H I, but instead of batting
> one after another, each batter is solely responsible for all plate
> appearances for an inning. That is, A bats in the first inning,
> and keeps batting until three outs are made (imagine that either pinch
> runners are used on the basepaths, or that clones of A occupy the bases,
> or use the "ghost runners" rule that we often used as kids when we were
> short-handed). After three outs are made, A doesn't bat again during the
> game. B gets all of the plate appearances in the 2nd inning, C in the
> 3rd, and so on.
>
> In this variation, each batter participates in a sequence of plate
> appearances that leads to run scoring, and the context of each plate
> appearance depends solely on his own prior performance. There's no
> interaction with the offensive prowess of his teammates. You can look at
> a batter's inning in isolation and know exactly how many runs were
> produced, and attribute them entirely to that batter's efforts.
>
> In a game like this, there's no reason to rely on OPS, since run scoring
> can be measured directly. We'd probably have a stat called something
> like BRA (Batter's Run Average) that it analagous to ERA or RA for
> pitchers.
>
> However, in real baseball, batters don't control the situational contexts
> they face, so we'd want a measurement that doesn't depend on them. OPS
> is one such option, and one that's been widely adopted for other reasons
> (availability of OBP and SLG in the media, and the mathematical
> simplicity of simply adding them together).
Well, that's one way to look at it. But not the only way. Hark back to
my hypo where each of four pitchers faces every fourth batter (or each
of five pitchers faces every fifth batter). With this paradigm, each
plate appearance, including its base-out situation, is independent with
respect to the pitcher-batter confrontation; that's analogous to each
plate appearance representing a pitching change, possibly accompanied by
inherited runners.
Pitching performance in this environment can be evaluated, perhaps by
OPS (POPS). We might say that a pitcher faced nine batters in a game,
with some result.
Now get back to reality where pitchers face successive batters, but
apply the paradigm we just developed, whereby each PA is evaluated
independently. A starting pitcher may go several innings and face 25
batters. Each PA, including the prior and post situations, can be
evaluated, and the pitcher's overall performance can be evaluated by
summing up.
Do you get a valid result this way? Sure you do. Suppose some inning
goes out, double, K, RBI single, out. Is there any reason not to say
that the pitcher, independently, (1) got a leadoff out; (2) gave up a
one-out bases-empty double; (3) got a one-out strikeout with a runner on
second; (4) gave up a two-out RBI single with a runner at second; and
(5) got a third out while stranding a runner at first?
There's nothing wrong with that. Whether the pitcher faced these
situations and got these results serially or spread out is immaterial.
Given the five base-out situations, the pitcher faced five batters and
got five results. He's charged or credited with each result as
appropriate, and his overall performance is the sum of the five.
Nothing at all wrong with this, and the fact that he created four of the
five situations is immaterial.
Therefore it's legitimate to evaluate pitchers using the same metric as
you'd use for batters.
"The RSBB Poster Formerly Known as James"
james withrow
kwoo...@best.com (Keith Woolner):
>Restating Chris's key point:
>>If OPS is the best q&d number for hitters, then
>>logic dictates that POPS *must* be the best q&d
>>number for pitchers.
>Not necessarily true. One of the key differences
>between pitchers and hitters is that hitters only
>get their opportunties sporadically during a
>game, and when they come up to bat, they face
>a situational context that they had little to no
>influence on (i.e. the base-out situation is due to
>the work of the other batters on the team and
>the opposing pitchers).
>A pitcher, on the other hand, faces a
>consecutive sequence of batters, so any
>situation faced by the pitcher is substantially
>dependent on his own performance (let's say >starting pitcher, for the
sake of argument. It's
>less true for relievers coming into the game with
>runners on base).
I'm not a statistician, but I play one on WebTV.
Is this the issue of "grouping"? Walks and hits grouped together score
more runs than if they're spread out. That may or may not have
something to do with my next paragraph.
IIRC QMAX scores suggest that a control pitcher who gives up more than 2
walks in an outing suddenly becomes vulnerable to a higher ERA. The
extra walks seem to make more of a difference than extra hits. The
formulas we use for hitting don't seem to explain this phenomenon, if I
have remembered and repeated it correctly. Maybe, a starter's control
tends to desert him over a stretch of batters rather than sporadically
throughout a game. If such is the case, grouping those walks with hits
would produce scoring outbursts.
A separate question I wish genuine staheads would focus on is which
stats explain value and which predict future success. Strike outs, for
instance, aren't worse for hitters than other outs, but they seem to be
of value predicting pitchers' futures. Of the two uses for stats--
explaining value and predicting the future-- the latter seems much more
important to me.
Withrow
John R. Tedrow
: Restating Chris's key point:
:> If OPS is the best q&d number for hitters, then logic dictates that
:> POPS *must* be the best q&d number for pitchers.
: Not necessarily true. One of the key differences between pitchers
: and hitters is that hitters only get their opportunties sporadically
: during a game, and when they come up to bat, they face a situational
: context that they had little to no influence on (i.e. the base-out
: situation is due to the work of the other batters on the team and
: the opposing pitchers).
One other difference between POPS and RA has do with the question
that you are trying to answer by using a particular stat. For me,
RA or whatever analogue you like is the number to look at when
you're trying to determine the value of a pitcher's contribution
in games that have already occurred. However, POPS may be a better
tool of determining if a pitcher is likely to succeed in the future.
For instance, Steve Avery has been very good at preventing runs
thus far, but he's walking roughly 5 guys per 9 IP, so you have a
feeling that something is going to change in the future. That
ill feeling, however, does not take away from the fact that
he has been a good pitcher for Cincinnati thus far.
- John Tedrow
"I'm through with patience; I'll wait no longer; You never
waited for me anyway; This thing inside me is growing stronger
It's going to tear, tear until it sees the light of day.
Clean the mind off this plastic baby." - Floater
zene...@wwa.com
> Of the two uses for stats-- explaining value and predicting the future--
> the latter seems much more important to me.
But predicting future stats does you no good if you can't evaluate
performance based on them, so the first use of stats (explaining value) is
a necessary prerequisite for the second use.
--
Jeff
Keith Woolner
cited author.]
In article <374705...@daedal.net>, j...@daedal.net says...
> Chris Dial asks good questions, and Keith Woolner has a knack for
> inventing hypothetical rules that lend themselves to insightful thought
> experiments. The following impresses me as a vintage Woolner tour de
> force.
Such praise from the esteemed Mr. Tuttle can only mean he intends
to disagree with me later in the article. :-)
> Keith Woolner wrote:
> >
> > Chris Dial writes:
> > Restating Chris's key point:
> >
> > > If OPS is the best q&d number for hitters, then logic dictates that
> > > POPS *must* be the best q&d number for pitchers.
> >
> > Not necessarily true.
>
> Keith is correct here. The word "must" is too strong. "May" is
> better. By rejecting the syllogism and the inevitability of the
> conclusion, we should not reject the possibility that the conclusion is
> nevertheless correct.
True, but let me reiterate that we are talking about which measure
is a better measure of pitching value. That is, "if OPS is the best for
hitters, then POPS is the best for pitchers". I *do* think that
POPS is a decent measure for pitcher performance, and it's not
without value. I just think that for assessment of past performance,
(E)RA is superior, as it incorporates a greater amount of relevant
data about the pitcher's influence on run scoring. I have no
philosophical objection to publishing or analyzing POPS, and it may have
specific extra value or information in diagnostic or predictive situations.
I'm a little confused, former-James. You seem to be disagreeing with
me, but your example supports my original contention.
In the game you describe, I would agree that POPS is probably a better
solution for measuring performance, again because the situational
contexts faced are largely independent of the pitcher's own performance.
It is, in fact, the reverse thought-experiment to the one I posted --
instead of making a batter's PAs less isolated, you are increasing
the isolation of a pitcher's PAs.
While you could extend that isolated-PA model that to real baseball, you
are losing information about a pitcher's productiveness that is preserved
in (E)RA. The fundamental differences between real baseball and either
of the pseudo-games we've described is whether the batter or pitcher
faces a sequence of situations of their own making. If they do, then
direct measurement of run scoring is a better measure. If they don't
control or influence the situational context, then we ought to try
to minimze its impact on our assessment of their production.
[...]
> > > Other than it isn't the easiest stat to come by, is there
> > > really a reason not to prefer POPS? (Or as Ron points out
> > > regularly in these times, OBP*2+SLG).
> >
> > I think the situational-context difference captures most of the reasons.
>
> Again, not necessarily, especially if you can account for or negate
> those situational-context differences.
I don't *want* to negate it in the cases where the context differences
are attributable to the player himself. It's a potentially significant
part of the process of run scoring, and worth preserving.
> > What I like to do when confronting a question like this
> > is to imagine a variant of baseball where the rules are different,
> > and would lead us to a different answer. Contrasting the two
> > often yields insight.
>
> Yes, and Keith is a master at this.
You're too kind. :-)
Yes, and in this hypothetical game POPS would be a better way to evaluate
pitchers, precisely because the situational context is not their own
creation.
> Now get back to reality where pitchers face successive batters, but
> apply the paradigm we just developed, whereby each PA is evaluated
> independently. A starting pitcher may go several innings and face 25
> batters. Each PA, including the prior and post situations, can be
> evaluated, and the pitcher's overall performance can be evaluated by
> summing up.
You can, but you lose information about the interaction between the
states created by the pitcher, and the effectiveness of the pitcher
in those states.
> Do you get a valid result this way? Sure you do. Suppose some inning
> goes out, double, K, RBI single, out. Is there any reason not to say
> that the pitcher, independently, (1) got a leadoff out; (2) gave up a
> one-out bases-empty double; (3) got a one-out strikeout with a runner on
> second; (4) gave up a two-out RBI single with a runner at second; and
> (5) got a third out while stranding a runner at first?
[...]
> Nothing at all wrong with this, and the fact that he created four of the
> five situations is immaterial.
Yes, you lose the transitional probabilities that describe how you
end up with those precise situations in that order. Taking them
independently without recognizing the sequential relationship and the
pitcher's responsibility for them you might be tempted. If a pitcher's
performance changes with the base-out situation, then implicitly assuming
a league average distribution of base-out situations faced (as a
PA-isolating measure such as OPS sort of does) skews the results. With
(E)RA, those probabilities are already automatically encoded in the
actual run scoring results.
>
> Therefore it's legitimate to evaluate pitchers using the same metric as
> you'd use for batters.
Legitimate? Yes. As good as using actual run scoring (i.e. (E)RA)? No.
You lose some accuracy and information in using OPS, but it will largely
give you the right range of values. And, as I said earlier, there may
be other reasons why you might want to look at OPS. It's not without merit,
but for the purposes Chris originally described, it is not as good as
(E)RA.
Keith Woolner
cited author.]
In article <6199-374...@newsd-112.bryant.webtv.net>, with...@webtv.net
says...
>
> kwoo...@best.com (Keith=A0Woolner):
>
> >Restating Chris's key point:
>
> >>If OPS is the best q&d number for hitters, then
> >>logic dictates that POPS *must* be the best q&d
> >>number for pitchers.
>
>
> >Not necessarily true. One of the key differences
> >between pitchers and hitters is that hitters only
> >get their opportunties sporadically during a
> >game, and when they come up to bat, they face
> >a situational context that they had little to no
> >influence on (i.e. the base-out situation is due to
> >the work of the other batters on the team and
> >the opposing pitchers).
>
> >A pitcher, on the other hand, faces a
> >consecutive sequence of batters, so any
> >situation faced by the pitcher is substantially
> >dependent on his own performance (let's say >starting pitcher, for the
> sake of argument. It's
> >less true for relievers coming into the game with
> >runners on base).
[...]
> Is this the issue of "grouping"? Walks and hits grouped together score
> more runs than if they're spread out.
Yes, grouping (or, mathematically speaking, changes in conditional
probability depending on base-out state). Clustering offensive events
will generally lead to more run scoring that an even distribution (unless
they're all home runs, of course). A pitcher who loses effectiveness
working from the stretch instead of the full windup should give up more
runs than you'd expect from his peripherals stats, because he's more
likely to surrender a bunch of hits together once a runner reaches
base.
[...]
> A separate question I wish genuine staheads would focus on is which
> stats explain value and which predict future success. Strike outs, for
> instance, aren't worse for hitters than other outs, but they seem to be
> of value predicting pitchers' futures. Of the two uses for stats--
> explaining value and predicting the future-- the latter seems much more
> important to me.
Generally speaking, the stats that explain value *are* the ones that
are most useful in predicting future success. The so-called predictive
stats or supporting stats, such as pitcher's strikeout rate, as used
to "tune" or adjust the prediction model, which still uses past
performance as an input.
I.e. If two pitchers had a 3.00 ERA and a 5.00 ERA last year, but
both struck out 8 batters per 9 innings, you don't look at the strikeout
rate and assume they will be equal in performance next year. You might
expect each to at least maintain or perhaps improve on their performance
because of their overpowering stuff indicated by their strikeout rate,
but you're still using their actual performance as a baseline for adjusting
your predicitons.
Where predictive stats are also useful is looking at two pitchers, each of
whom have a 3.00 ERA last year, but one had a strikeour rate of 9.0 K/9IP
and the other had a rate of 3.0 K/9IP, the former is more likely to retain
his above-average level of performance than the latter (all else being
equal).
Hope that helps.
David Grabiner
> I've gotten an explanation from David Grabiner before, but
> my accepting his explanation is based solely on my respect
> for his knowledge, so since I've never met him, occasionally
> I doubt my faith, and ask: If OPS is the best q&d number for
> hitters, then logic dictates that POPS (as named above)
> *must* be the best q&d number for pitchers.
The reason that this does not work is that pitchers create their own
situations, while hitters do not. A linear model of offense, such as
OPS or LWTS, assumes that the difference between a home run and a
strikeout is independent of what the player does in his other
situations.
For a hitter, this is a good assumption, with the exception of lineup
effects. A home run will lead to more runs if there are more runners on
base, but a hitter can do very little to affect the number of runners on
base for his home runs.
For a pitcher or a team, this assumption does not work. A home run will
lead to fewer runs if it is hit by a team with a low OBP, or off a
pitcher with a low opposition OBP. Thus a proper team model is based on
the interaction between runners on base and advancement; RC is an
example of such a model.
OPS (or k*OBP+SLG for an appropriate value of k) works for hitters
because it measures the hitter's contribution to his team's product
model. 1.2*OBP+SLG is a hitter's contribution to his team's OBP*SLG,
and 1.7*OBP+SLG is a hitter's contribution to his team's RC or LWTS.
--
David Grabiner, grab...@math.lsa.umich.edu
http://www.math.lsa.umich.edu/~grabiner
Shop at the Mobius Strip Mall: Always on the same side of the street!
Klein Glassworks, Torus Coffee and Donuts, Projective Airlines, etc.
James Weisberg
>I've gotten an explanation from David Grabiner before, but
>my accepting his explanation is based solely on my respect
>for his knowledge, so since I've never met him, occasionally
>I doubt my faith, and ask: If OPS is the best q&d number for
>hitters, then logic dictates that POPS (as named above)
>*must* be the best q&d number for pitchers.
>
>not baserunners. Then the job of hitters is to score runs,
>not get on base. Both are a means to an end, and one
>requires the other.
I've been thinking about this a little bit and I'm wondering if
there's a way to come up with a pitching stat which reflects both POPS
and ERA. I don't know if anyone's done this yet, but what about
scaling ERA by OPS? As in, ER*OPS/IP? I've called this stat OPERA
or OPRA (for RA vs ERA) and made the following chart:
Yr H BB R ER IP ERA OBP SLG OPRA OPERA
Maddux 98 201 45 75 62 251.0 2.22 .260 .299 .167 .138
Glavine 98 202 74 67 63 229.1 2.47 .300 .325 .183 .172
Smoltz 98 145 44 58 54 167.2 2.90 .285 .337 .215 .200
Brown 98 225 53 77 68 257.0 2.38 .279 .294 .172 .152
Schilling 98 236 64 101 97 268.2 3.25 .282 .373 .246 .236
Clemens 98 169 88 78 69 234.2 2.65 .277 .296 .190 .168
Martinez 98 188 70 82 75 233.2 2.89 .278 .347 .219 .201
Johnson 98 203 87 102 89 244.1 3.28 .300 .344 .267 .235
Maddux 97 200 20 58 57 232.2 2.20 .256 .311 .141 .139
Glavine 97 197 79 86 79 240.0 2.96 .292 .333 .224 .206
Smoltz 97 234 63 97 86 256.0 3.02 .288 .359 .245 .217
Brown 97 214 73 77 71 237.1 2.69 .303 .319 .202 .186
Schilling 97 208 61 96 84 254.1 2.97 .271 .372 .243 .212
Clemens 97 204 69 65 60 264.0 2.05 .273 .290 .139 .128
Martinez 97 158 72 65 51 241.3 1.90 .250 .277 .142 .111
Johnson 97 147 79 60 54 213.0 2.28 .277 .318 .168 .151
It basically puts the results in an order which you'd expect
from ranking by ERA alone. Though, for instance, we see that Curt
Schilling is hurt by allowing inflated SLG numbers against his
opponents. This numbers could be further adjusted for league and
park. Then you'd have Martinez and Clemens with even more staggering
numbers.
If ya'll like OPERA (insert joke here) and think it adds
something to the pitching evaluation, maybe we can do some more
detailed analysis of it, perhaps using 1.2*OBP+SLG and use the
park/league adjustments.
--
World's Greatest Living Poster
Gerry Myerson
> It's not uncommon, he is a direct James prodigy...
Neyer may well be a prodigy, but I think the word you were looking for here
was "protege".
Gerry Myerson (ge...@mpce.mq.edu.au)
Jeremy Billones
statheads keep reinventing cricket when they need to test hypotheses :)
--
Jeremy Billones http://www.primenet.com/~billones/
"You bust through the door and create a diversion. They all turn and aim
at you. You try and sweet talk them out of blowing your brains out while
I sneak around back, bust in, and *really* surprise 'em!"
Dan Schmidt
| Dan Schmidt wrote:
| >
| > dtk <kur...@musc.edu> writes:
| >
| > | Why doesn't anybody tabulate pitchers' OPS allowed?
| >
| > OPS is an attempt to combine lots of information into one big number
| > that summarizes how good a hitter is.
| >
| > In the case of pitching, we already have a fine 'one big number that
| > summarizes how good a pitcher is'; it's ERA (though some people prefer
| > to just use Run Average). So the main reason for OPS doesn't really
| > exist.
|
| That's wrong, because the ERA has several flaws.
|
| 1. The "inherited runner" problem.
Yes, I mentioned this in the part of my post you snipped.
| 2. The possibility that a two-out error can lead to a half-dozen
| unearned runs.
Yes, I mentioned this in passing. I guess I should have explained why
some people prefer Run Average.
| 3. The fact that the results of fielding errors should be measured in
| unearned bases, not unitary unearned runs.
That's a good point, but it's arguable whether one should change the
fundamental unit of pitcher measurement from runs to bases just to
address this issue.
| Doug Drinen addresses these, and so do I, in my Earned Runs Per Game
| (ERPG) stat in Base Production.
I'm not familiar with ERPG, but I admire Doug's work.
Eric Roush
: Maybe I'm just the wrong audience, but I would much prefer a
: book filled with articles by Woolner, Grabiner, Johnson.
: The best part of the Abstracts and Elias' Analysts is the
: articles...I don't buy them to look at the pictures.
Well, I suppose you could just print out r.s.b. :-)
--
Eric Roush "I can show you how to make a bomb out of
ero...@phl.vet.upenn.edu a roll of toilet paper and a stick of
dynamite." Dale Gribble
This program has undergone a Y2K upgrade. Have a nice dak.
Eric Roush
: First and least importantly, what acronym *isn't* non-intuitive? BASIC
: and ICQ are the only two I can think of, and I'm not sure ICQ is
: actually an acronym.
SNAFU?
Eric Roush
: As a layman in the worst sense of the word, I would be apprehensive to
: take any of the theories given for pitchers whose arms end up requiring
: surgery like Kerry Wood. All of this data that's being collected and put
: into models, doesn't address some other important issues.
: 1. What are the arm mechanics of the pitcher in question?
: Often it is poor arm mechanics that can lead to a pitcher's arm breaking
: down. How will this be factored into PAP or the 3 start theory proposed
: by BBBA?
This doesn't really affect the question being asked, which
is "What is the effect of a certain workload on the health
of pitching arms?" If you can gather data showing that
given workload x at age y, n% of pitchers suffer a prolonged
loss of effectiveness, you can then search the group
of injured pitchers for contributory factors, such as
bad mechanics. My gut feeling is that poor mechanics
will be a minor factor in arm injuries relative to
age/useage. Certainly, the anecdotal evidence can support
this gut feeling, as plenty of guys such as Steve Avery
suffered through arm troubles despite having excellent mechanics.
: 2. What was the workload of the injured pitcher at a young age?
: Pitch counts are given these days in just about every box score you see
: online, but what about the workload of a pitcher when he's in high
: school, or even further back to Little League? I know that some, if not
: all, of the youth levels in baseball have rules to prevent this kind of
: abuse, but I've also heard that one of the reasons proposed for Woods'
: injury was that he was "abused" in high school. Will there be more than
: just anectdotal evidence for this part of the equation?
You can only work with the data available, and getting this
sort of data would be quite problematic. However, if any
of us dads think we've got a budding major-league arm
in the family, we might want to keep an eye on workloads.
vi...@baseball.org
> Why doesn't anybody tabulate pitchers' OPS allowed?
We seem to have this debate about 3 times/year. I've read somewhere
(though I can't remember where) that OPS allowed correlates better
with the following year's ERA than current ERA does. That's a pretty
hefty statement. I'd like to do a study to back this up, but I don't
have the data in any handy electronic format.
For my own purposes (fantasy or whatever), I look at opponents' OBA/SA
a lot, especially for relievers. I especially try to find pitchers
whose OBA/SA allowed looks good, but their ERA and W-L doesn't,
because those pitchers tend to be undervalued. While I think that
run-based measures (especially with adjustments for inherited runners,
etc) are better at measuring how valuable a pitcher has been in the
best, I don't necessarily think those are the best predictive
measures. Two different purposes, and a distinction I think people
should keep clear when discussing how useful a metric is.
--
/---------------------------------------------------------------\
| Vinay Kumar |
| vi...@baseball.org http://www.baseball.org/~vinay |
\---------------------------------------------------------------/
David Grabiner
>
> "J. Edward Tuttle" <j...@daedal.net> writes:
> | 3. The fact that the results of fielding errors should be measured in
> | unearned bases, not unitary unearned runs.
> That's a good point, but it's arguable whether one should change the
> fundamental unit of pitcher measurement from runs to bases just to
> address this issue.
> | Doug Drinen addresses these, and so do I, in my Earned Runs Per Game
> | (ERPG) stat in Base Production.
It's not necessary to treat a base as the fundamental unit; you can
convert everything to runs or wins. Doug's work uses wins as the
fundamental unit; a reliever who enters in a situation in which his team
had a .500 probability of winning and leaves in a situation in which his
team had a .625 probability of winning is credited with .125 of a run.
The same thing can be done with runs, using Palmer's table of expected
future runs. If a hitter reaches first on an error with two out, the
team has .209 expected runs rather than zero, so the error allows .209
of an unearned run, and the pitcher has prevented .095 of an earned run
just as if he geot the last batter out. If the inning ends with one run
scored, the pitcher has allowed .791 of a run, and the defense has
allowed .209; if the pitcher works out of the jam, he has allowed -.209
of an earned run. (Thus a shutout with errors gives more credit to the
pitcher than an errorless shutout.)
vi...@baseball.org
> > the latter seems much more important to me.
>
> performance based on them, so the first use of stats (explaining value) is
> a necessary prerequisite for the second use.
But we don't need to use the same statistics for both uses. If I know
that Wally Joyner hit like a banshee in clutch situations last year, I
know that he was more valuable to the Padres than your average
.298/.370/.453 hitter. But since I don't think that clutch hitting is
an ability that's duplicated from year to year, I wouldn't let that
knowledge enter my thinking of decisions based on future value (ie,
trade talk).
Similarly, though I think run-based measures are the best way to judge
a pitcher's value, I think OPS-like measures might predict his future
performance better.
kin...@sp2n23-t.missouri.edu
Gerry Myerson <ge...@mpce.mq.edu.au> wrote:
>In article <7i40s8$18a$1...@nnrp1.deja.com>,
>Mosey <jmo...@my-dejanews.com> wrote:
>>
>> It's not uncommon, he is a direct James prodigy...
>
>Neyer may well be a prodigy, but I think the word you were looking for
>here was "protege".
...but presumably with at least one accent mark (and an unnecessary umlaut
would be nice :-)).
On the other hand, I thought the poster meant to say "progeny". The
"direct" part had me stumped until I considered the possibility of asexual
reproduction here. I mean, we are talking about geeks here...
jking
Roger Moore
>Derek (heck...@yahoo.com) wrote:
>: As a layman in the worst sense of the word, I would be apprehensive to
>: take any of the theories given for pitchers whose arms end up requiring
>: surgery like Kerry Wood. All of this data that's being collected and put
>: into models, doesn't address some other important issues.
>: 1. What are the arm mechanics of the pitcher in question?
>: Often it is poor arm mechanics that can lead to a pitcher's arm breaking
>: down. How will this be factored into PAP or the 3 start theory proposed
>: by BBBA?
>This doesn't really affect the question being asked, which
>is "What is the effect of a certain workload on the health
>of pitching arms?" If you can gather data showing that
>given workload x at age y, n% of pitchers suffer a prolonged
>loss of effectiveness, you can then search the group
>of injured pitchers for contributory factors, such as
>bad mechanics. My gut feeling is that poor mechanics
>will be a minor factor in arm injuries relative to
>age/useage. Certainly, the anecdotal evidence can support
>this gut feeling, as plenty of guys such as Steve Avery
>suffered through arm troubles despite having excellent mechanics.
There's another real possibility here, which is that bad mechanics act as
an aggrivating factor in conjunction with heavy workloads. This is
essentially the theory which Tom House (who I realize is not the most
popular pitching theorist around here) used in "The Pitching Edge". He
used a modifier to pitch counts based on a pitcher's mechanical
efficiency. IOW, a guy with bad mechanics might take only 85 or 90
pitches to get as tired (and reach the same injury potential) as a guy
with great mechanics would get in 100 pitches. This also goes along well
with the idea that pitches from the stretch are more dangerous than those
from the windup; the stretch is inherently less efficient mechanically
than the windup. I wonder if this could have important bearing on injury
rates among relievers.
The key here, though, is that mechanics are only a contributing factor to
injury. Perfect mechanics can't protect a pitcher if his manager lets him
throw 200 pitches per start. Similarly, a pitcher with poor mechanics can
be protected by putting him on a more stringent pitch count. This would
also be consistent with the idea that, when broken down into groups who
threw similar numbers of pitches at the same age, pitchers with bad
mechanics will tend to be most injured.
--
Raj (r...@alumni.caltech.edu)
Master of Meaningless Trivia (626) 585-0144
http://www.alumni.caltech.edu/~raj/
Mason M A
>given workload x at age y, n% of pitchers suffer a prolonged
>loss of effectiveness,
Has such data been gathered to support either PAP or the BBBA 3-start model?
If so, I didn't see it presented in either book this year. This is an
important objection to the PAP methodology--if it hasn't been back-tested, and
there hasn't been time to test it prospectively, then it really is of no proven
predictive use.
Mark
Samson
ke...@woolner.com wrote:
> In article <374705...@daedal.net>, j...@daedal.net says...
> > > Using ERA (or RA) preserves the situational context of the pitcher's
Ok, be it agreed (in 'lay man's' terms) that a pitcher's performance
is best evaluated by simply looking at the rate at which he
allows runs -- except for certain situations where runners are
inherited or bequeathed -- because, in general, runs are a direct
result of a pitcher's cumulative, sequential acts of ineptitude.
But what about that ERA/RA equivocation? There is still the question
of the number of runs that really *are* attributable to the pitcher,
even if he is the only pitcher in a real baseball game. ERA is
*supposed* to take care of this by absolving the pitcher of runs
that result from fielding errors. But then ERA lets the pitcher
off the hook for so many runs that are so remotely related to
errors that RA looks more valid. But then you're back to the
original problem. Is it fair just to assume that team fielding is
either essentially irrelevant or that it doesn't vary enough to
affect pitchers on different teams differently?
Couldn't opponents' RC/27, if not OPS, strike a balance between
between RA's overcharging and ERA's undercharging of runs to
pitchers? It would seem to be the simplest way of going about
it, as opposed to going back and re-reconstructing innings to
take out the silly UERs. To the extent that RC (or whatever
measure you like) is valid for predicting runs scored, and given
the fact that it already has errors taken out of it, it might be
useful for seeing just how screwed up ERA is (or isn't).
(I have a sinking feeling my logic is flawed, but I'm sure I'll
be rebutted if so...)
Anyway, I'd like to see 2B- and 3B-allowed figures given for
pitchers, or at least total bases. I don't see them in TB's or any
of STATS's online sources. I like the idea of OOPS. BA/OBP/SLG
lines have aesthetic appeal: they don't just tell how good a hitter
is, but _how a hitter is good_. I'd like to see how OOPS "mirrors"
OPS. Was Nolan Ryan more like the "anti-McGwire" or the "anti-
Henderson"? Is there an "anti-Kingman" profile? Is the OSLG/OOB
relationship to RA the same as SLG/OOB for a batter's run-creation?
Plus, OOPS is a great acronym. (Forget POPS. It's OOPS. It should
correlate with the number of times a pitcher thinks, "oops," per
PA.)
J. Edward Tuttle
necessarily better than RA/27 or ERA since OPS is a good measure for
batters.
Keith Woolner, in a well-reasoned argument (see the referenced post),
says not necessarily, mostly because batters don't control the
situations in which they hit, while pitchers, because of continuity, do.
I said you can construct a paradigm that gives situation-independence to
pitchers, so that OPS and POPS can be fairly applied to batters and
pitchers respectively. This approach removes the continuity aspect for
pitchers.
Keith says, yes, you can do that, but he doesn't want to. He says it's
better to keep the continuity, and therefore RA/27 or ERA is better for
pitchers than POPS.
I now say that's right, provided you want to keep the continuity. But
you pay a price for this retention, and in my opinion, the price is too
high. The presence of fielding errors/PBs messes things up, and
changing pitchers with runners on base messes things up even further.
By throwing away the continuity stuff, you can get a way to evaluate
pitchers just like you evaluate batters, and I think that's the way to
go.
J. Edward Tuttle
>
> Dan Schmidt writes:
>
> > James Tuttle writes:
>
> > | 3. The fact that the results of fielding errors should be
> > | measured in unearned bases, not unitary unearned runs.
>
> > That's a good point, but it's arguable whether one should change the
> > fundamental unit of pitcher measurement from runs to bases just to
> > address this issue.
>
> convert everything to runs or wins.
That's right; it takes four bases to score a run, so you could say a
base is a quarter of a run. And the conversion to wins is table-driven
and probabilistic.
I prefer the base as the fundamental unit because they're easy to count:
1 base, 2 bases, 3 bases, ... I like that better than 0.25 runs, 0.50
runs, 0.75 runs, ... And when doing this stuff in spreadsheets, I can
use integers for bases instead of two decimal places for runs.
J. Edward Tuttle
>
> Chris Dial writes:
>
> > I've gotten an explanation from David Grabiner before, but
> > my accepting his explanation is based solely on my respect
> > for his knowledge, so since I've never met him, occasionally
> > I doubt my faith, and ask: If OPS is the best q&d number for
> > hitters, then logic dictates that POPS (as named above)
> > *must* be the best q&d number for pitchers.
>
> situations, while hitters do not.
That's usually, but not always, the case. Fielding errors and PBs upset
it for pitchers, and so do pitching changes with runners on base. These
things happen often enough to warrant, IMO, the removal of this reliance
on the "continuity argument" for pitcher stats.
And removing the relance on continuity is easily accomplished, at little
or no cost. If an inning goes Out, Out, safe on error, single (runner
to third), Out, we can say (in reverse order, for emphasis) that the
pitcher faced five batters in these situations and with these results:
1. Two outs, runners at corners: got the third out.
2. Two outs, runner at first: gave up single, advancing runner to
third.
3. Two outs, bases empty: Batter safe on error; unearned by pitcher.
4. One out, bases empty: got an out.
5. No outs, bases empty: got an out.
> A linear model of offense, such as OPS or LWTS, assumes that the
> difference between a home run and a strikeout is independent of what
> the player does in his other situations.
You can apply the same model to defense.
> For a hitter, this is a good assumption, with the exception of lineup
> effects. A home run will lead to more runs if there are more runners
> on base, but a hitter can do very little to affect the number of
> runners on base for his home runs.
But you can normalize it, assuming a league-average baserunner
configuration for each PA.
> For a pitcher or a team, this assumption does not work.
[rest snipped]
But it can be made to work. A good analogy is to consider how Simpson's
Rule or the Trapezoid Rule works for finding areas, or how you
geometrically derive the Fundamental Theorem of Calculus. Regarding
intermediate points, when you do the expansion, you add something in one
term, then subtract it out in the next term, so that the overall total
only depends on the endpoints.
If we want to get some total area, we can break it up into A[1], A[2],
... and then add 'em up. The continuity approach for pitchers uses the
fact that A[2] is adjacent to A[1]. But you don't *have* to add 'em up
sequentially; you get the same result if you add them in random
sequence.
That's my point.
Ira Blum
"J. Edward Tuttle" wrote:
>
> Ira Blum wrote:
>
> > > All of the numbers in BP may be translated, but at least they have
> > > the same metrics as those of any normal sports fan. Translated ERA
> > > may be different than Raw ERA, but its still in the same metric.
>
> and Doug Drinen replied:
>
> > I'm not sure I understand your use of the word "metric." I think I
> > can answer your question, though.
>
> Ira is obviously an educated person. I used to think "metric" meant the
> metric system -- until I took some geometry course where the guy kept
> talking about "metrics." Then I had to look it up.
>
Thank you for the complement. I am an educated person. I just (last
Saturday)
received my Ph.D. in High Energy Physics from UT Dallas. (self plug).
> Here, I think Ira means "units of measure." No problema. When in Rome,
> talk like the Romans do.
>
I did, sort of. To follow what Doug said, the idea here was that SNVA
and WPA mean
wins over average, sort of like pitching wins. In both senses, they
miss the point. A .500 pitcher, (who would have a SNVA of 0 and a WPA
of 0) HAS value. But you can rewrite SNVA to SNW and SNL (support
neutral wins and losses) and that is a better unit of measuring a
pitcher than pitching wins over average, which penalizes .500 pitchers,
or pitching wins over replacement value, which is controversial at best
(where do you place replacement value, etc...) A pitcher who has a
SNW-L of 15.2-16.4 is much more valuable than one who is 5.1-5.6 That
is easy to tell even for a duffer. A pitcher who's SNW-L is 18.4-3.1 is
a Cy young Winner, one who's WPA is +12 may or may not be. (think 13-1
vs 22-10)
That's my whole argument in a nutshell.
Ira
Jeff Drummond
If we're talking about geeks, then I don't think we can overlook the
possibility that "prodigy" is referring to an online service provider.
Maybe James and Neyer shared an email account?
-Jeff j...@sgi.com
--
JEAN PAUL SARTRE for America Online
"In my journey to the end of the night, I must rely not only on the dialectical
paths of reason. I must have a good internet connection, one that eschews the
futile trappings of worldly ennui and provides only the most basic NetNews
capability. My ISP offers me this simple solace, and as my postings disappear
into the ether I am struck by the realization of their pointlessness. I may
not know if anyone sees them or not. It is of no consequence."
Michael David Jones
>> Chris Dial writes:
>> > I've gotten an explanation from David Grabiner before, but
>> > my accepting his explanation is based solely on my respect
>> > for his knowledge, so since I've never met him, occasionally
>> > I doubt my faith, and ask: If OPS is the best q&d number for
>> > hitters, then logic dictates that POPS (as named above)
>> > *must* be the best q&d number for pitchers.
>> The reason that this does not work is that pitchers create their own
>> situations, while hitters do not.
>That's usually, but not always, the case. Fielding errors and PBs upset
>it for pitchers, and so do pitching changes with runners on base. These
>things happen often enough to warrant, IMO, the removal of this reliance
>on the "continuity argument" for pitcher stats.
I don't see it, James. The difference between a PB and a WP, as well
as the difference between a fielding error and a double, is largely in
the mind of the official scorer. Yes, there are plays that are
plainly errors. There are also plays where an oufielder loses the ball
in the lights, or a ball takes a bad hop past an infielder, which are
equally "random", not of the pitchers' creation, but are never marked
with an 'E' and don't seem to ever come up in these discussions. The
problem of separating the pitcher's contribution to a situation from
the defense's remains as sticky as ever, and goes well beyond
fielding errors and passed balls.
...snip...
>> For a hitter, this is a good assumption, with the exception of lineup
>> effects. A home run will lead to more runs if there are more runners
>> on base, but a hitter can do very little to affect the number of
>> runners on base for his home runs.
>But you can normalize it, assuming a league-average baserunner
>configuration for each PA.
>> For a pitcher or a team, this assumption does not work.
>[rest snipped]
>But it can be made to work. A good analogy is to consider how Simpson's
>Rule or the Trapezoid Rule works for finding areas, or how you
>geometrically derive the Fundamental Theorem of Calculus. Regarding
>intermediate points, when you do the expansion, you add something in one
>term, then subtract it out in the next term, so that the overall total
>only depends on the endpoints.
>If we want to get some total area, we can break it up into A[1], A[2],
>... and then add 'em up. The continuity approach for pitchers uses the
>fact that A[2] is adjacent to A[1]. But you don't *have* to add 'em up
>sequentially; you get the same result if you add them in random
>sequence.
>That's my point.
I see that. It doesn't address the issue of clumping, though. A
pitcher's "events" are grouped inherently differently from batters'. A
batter who's having a "bad game" (perhaps because he didn't sleep well
the night before, or is fighting off a mild cold) will have 5 or 6
PA's affected by it, quite likely not all against the same pitcher. A
pitcher in the same situation may struggle through 25 or 30 PA. The
same is true for any event that affects an individual game (rain
delay, cold or wet weather, bad umpiring, etc.). This effect tends to
reduce the independence of pitchers' events much more than hitters'.
Mike Jones | jon...@rpi.edu
I see that only during the Reagan/Bush years has black entertainment
(which is most excellent, in my opinion) made any serious inroads.
- mike myke schwartz (my...@shell.portal.com)
- in alt.society.conservatism, July 1993
David Grabiner
> David Grabiner wrote:
> > A linear model of offense, such as OPS or LWTS, assumes that the
> > difference between a home run and a strikeout is independent of what
> > the player does in his other situations.
> You can apply the same model to defense.
> > For a hitter, this is a good assumption, with the exception of lineup
> > effects. A home run will lead to more runs if there are more runners
> > on base, but a hitter can do very little to affect the number of
> > runners on base for his home runs.
> But you can normalize it, assuming a league-average baserunner
> configuration for each PA.
That's effectively what LWTS does; the value of a home run is based on
the league-average value of home runs.
> > For a pitcher or a team, this assumption does not work.
> But it can be made to work. A good analogy is to consider how Simpson's
> Rule or the Trapezoid Rule works for finding areas, or how you
> geometrically derive the Fundamental Theorem of Calculus. Regarding
> intermediate points, when you do the expansion, you add something in one
> term, then subtract it out in the next term, so that the overall total
> only depends on the endpoints.
> If we want to get some total area, we can break it up into A[1], A[2],
> ... and then add 'em up. The continuity approach for pitchers uses the
> fact that A[2] is adjacent to A[1]. But you don't *have* to add 'em up
> sequentially; you get the same result if you add them in random
> sequence.
But if you do this, you will get inaccurate numbers. A pitcher who
allows very few baserunners will have more situationss than average in
which a home run is worth exactly one run, and thus will allow fewer
runs on his home runs than an average pitcher does.
What you would need to do to get an accurate figure is to use the
pitcher's own statistics to find the frequency of various base-out
situations. This is estimated by a model such as RC or OBP*SLG.
I haven't checked the numbers, but I would expect that pitchers' actual
runs allowed tend to exceed the linear prediction forom OPS for very
good and very bad pitchers. I do know that OPS (or even 1.2*OBP+SLG) is
not as good a predictor for team runs scored as OBP&SLG is, but the
difference is very small at the team level.
James Weisberg
>I prefer the base as the fundamental unit because they're easy to count:
>1 base, 2 bases, 3 bases, ...
James, a word of advice, you should really stop showing off;
it doesn't suit you. ;-)
David Grabiner
> David Grabiner wrote:
> > Dan Schmidt writes:
> > > James Tuttle writes:
> > > | 3. The fact that the results of fielding errors should be
> > > | measured in unearned bases, not unitary unearned runs.
>
> > > That's a good point, but it's arguable whether one should change the
> > > fundamental unit of pitcher measurement from runs to bases just to
> > > address this issue.
> > It's not necessary to treat a base as the fundamental unit; you can
> > convert everything to runs or wins.
> That's right; it takes four bases to score a run, so you could say a
> base is a quarter of a run. And the conversion to wins is table-driven
> and probabilistic.
However, when you make this correction, you lose precision; the value of
going from second to third base with two out is very small, while the
value of scoring from third base with two out is very large.
In its most complete form, LWTS gives only .22 runs for the difference
between a double and a triple, but .38 runs for the difference between a
triple and a homer, even though the difference between a triple and a
home is exactly one base while the difference between a double and a
triple may be more than one base if a runner on first cannot score.
PHSpiegel1
to see what his reaction will be to see a 97-post thread with the subject line
"BBBA question for Don Malcom."
- Peter (phspi...@aol.com)
J. Edward Tuttle
> > base is a quarter of a run.
>
> of going from second to third base with two out is very small, while
> the value of scoring from third base with two out is very large.
Yes, I understand that. Same thing applies to sac hits where you gain a
base, but at the cost of an out.
> In its most complete form, LWTS gives only .22 runs for the difference
> between a double and a triple, but .38 runs for the difference between
> a triple and a homer, even though the difference between a triple and
> a home is exactly one base while the difference between a double and a
> triple may be more than one base if a runner on first cannot score.
It's a tradeoff I made consciously. I attach value to the ability to
maintain counting statistics, and I'm willing to give up accuracy for
that.
There is, however, an offset. In the long run, players gain first base
w times, second base x times, third base y times, and they score z
times, where w > x > y > z.
My stranded-runner penalties catch runners at first s times, at second t
times, and at third u times, where s > t > u. (You can't be stranded
after scoring.)
So net bases works out to (w-s) + (x-t) + (y-u) + z.
With this set-up, the one run = four bases rule seems to work pretty
well.
J. Edward Tuttle
>
> James Tuttle writes:
> >David Grabiner wrote:
> >>
> >> [P]itchers create their own situations, while hitters do not.
> >>
> >That's usually, but not always, the case. Fielding errors and PBs
> >upset it for pitchers, and so do pitching changes with runners on
> >base. These things happen often enough to warrant, IMO, the removal
> >of this reliance on the "continuity argument" for pitcher stats.
>
> I don't see it, James. The difference between a PB and a WP, as well
> as the difference between a fielding error and a double, is largely in
> the mind of the official scorer.
Well, I know that, but I also know these guys are paid $75 per game to
make these decisions. To become a scorer, they even make you pass a
test.
I'll be the first to say that the quality of official scorekeeping can
stand significant upgrading, but I'm not holding my breath. Meanwhile,
we need to live with the decisions they make. Unless the league office
(or maybe Elias) changes decisions, they're non-negotiable -- a given.
> Yes, there are plays that are
> plainly errors. There are also plays where an oufielder loses the ball
> in the lights, or a ball takes a bad hop past an infielder, which are
> equally "random", not of the pitchers' creation, but are never marked
> with an 'E' and don't seem to ever come up in these discussions. The
> problem of separating the pitcher's contribution to a situation from
> the defense's remains as sticky as ever, and goes well beyond
> fielding errors and passed balls.
I agree with you; it's just that I don't question scorer's decisions.
You can't change 'em, so it's not worth arguing.
The other night some shortstop -- Renteria maybe -- made an incredible
stop of a sharp grounder headed toward left field. He knocked the ball
down and retrieved it in plenty of time to get the out at first. His
throw, however, pulled the first baseman off the bag and the
batter-runner was safe.
The batter was credited with a base hit. Had the throw been on target,
the guy would've been out by a mile. If this were an ordinary grounder,
there surely would've been an error on the throw. Had the shortstop not
gotten to the ball, it had "base hit" written all over it.
In the event, there was ample call to rule this a throwing error. But
because of the shortstop's incredible fielding getting to the ball, the
scorer obviously felt he didn't want to charge the error.
These dilemmas need to be professionally and consistently handled,
without regard to home team or visiting team, or who the pitcher,
batter, and fielder are, and I regret to say that MLB is not addressing
these issues.
That still doesn't change my answer though.
[snip]
> >If we want to get some total area, we can break it up into A[1],
> >A[2], ... and then add 'em up. The continuity approach for pitchers
> >uses the fact that A[2] is adjacent to A[1]. But you don't *have* to
> >add 'em up sequentially; you get the same result if you add them in
> >random sequence. That's my point.
>
> I see that. It doesn't address the issue of clumping, though.
Yes, it does. Suppose a pitcher gives up six hits -- all singles -- in
six innings of work. If they were all two-out singles -- one in each
inning -- then he faced 24 batters as follows:
- No outs, bases empty, got an out (six times)
- One out, bases empty, got an out (six times)
- Two outs, bases empty, gave up a single (six times)
- Two outs, runner on first, got an out (six times)
Now suppose the first six batters in the first inning all got singles,
everybody went station-to-station, and then he struck out the side.
After the first inning, he was perfect.
Three runs scored in the first due to clumping, and he faced twenty-four
batters as follows:
- No outs, bases empty, got an out (five times)
- No outs, bases empty, gave up a single (once)
- No outs, runner on first, gave up a single (once)
- No outs, runners on first and second, gave up a single (once)
- No outs, bases loaded, gave up a single (three times)
- No outs, bases loaded, got an out (once)
- One out, bases empty, got an out (five times)
- One out, bases loaded, got an out (once)
- Two outs, bases empty, got an out (five times)
- Two outs, bases loaded, got an out (once)
Total, twenty-four batters faced. Clumping is no problem.
> A pitcher's "events" are grouped inherently differently from batters'.
> A batter who's having a "bad game" (perhaps because he didn't sleep
> well the night before, or is fighting off a mild cold) will have 5 or
> 6 PA's affected by it, quite likely not all against the same pitcher.
> A pitcher in the same situation may struggle through 25 or 30 PA. The
> same is true for any event that affects an individual game (rain
> delay, cold or wet weather, bad umpiring, etc.). This effect tends to
> reduce the independence of pitchers' events much more than hitters'.
No, no, no, Mike! Your imagination is working overtime. Players are
out there every day, and they go through batting slumps and hot streaks
that last for several games. Starting pitchers only work once every
five days. Relievers are somewhere in the middle.
The only thing affecting this is the number of PAs or batters faced in a
season, along with the mix of position players (maybe 15) and pitchers
(maybe 10) on a team. Compare the number of batters faced by a starter
with the number of PAs by a regular. Compare the numbers for a reliever
and for a bench player.
Look -- you're raising some good points, but I think you've got it in
your head that the sequential nature of pitching matters more than it
really does. The fact is that the "pitching continuity" paradigm that
we're all accustomed to is not necessary for analysis purposes, and
indeed, probably does more harm than good.
Specifically, I point to the hoops people jump through to handle
inherited runners ... it's like Ptolemy's epicycles ... and the utter
sinfulness (if I may wax religious) of letting a pitcher off the hook
for everything that occurs after a two-out fielding error.
The poster formerly known as James now relinquishes the podium to his
esteemed colleague, the gentleman from Rochester ...
Michael David Jones
>> James Tuttle writes:
>> >David Grabiner wrote:
>> >> [P]itchers create their own situations, while hitters do not.
>> >That's usually, but not always, the case. Fielding errors and PBs
>> >upset it for pitchers, and so do pitching changes with runners on
>> >base. These things happen often enough to warrant, IMO, the removal
>> >of this reliance on the "continuity argument" for pitcher stats.
>> I don't see it, James. The difference between a PB and a WP, as well
>> as the difference between a fielding error and a double, is largely in
>> the mind of the official scorer.
>Well, I know that, but I also know these guys are paid $75 per game to
>make these decisions. To become a scorer, they even make you pass a
>test.
>I'll be the first to say that the quality of official scorekeeping can
>stand significant upgrading, but I'm not holding my breath. Meanwhile,
>we need to live with the decisions they make. Unless the league office
>(or maybe Elias) changes decisions, they're non-negotiable -- a given.
Why? Why not do what DA and ZR do: treat everything as "out" or "not
out".
...snip...
>[snip]
>> >If we want to get some total area, we can break it up into A[1],
>> >A[2], ... and then add 'em up. The continuity approach for pitchers
>> >uses the fact that A[2] is adjacent to A[1]. But you don't *have* to
>> >add 'em up sequentially; you get the same result if you add them in
>> >random sequence. That's my point.
>> I see that. It doesn't address the issue of clumping, though.
>Yes, it does. Suppose a pitcher gives up six hits -- all singles -- in
>six innings of work. If they were all two-out singles -- one in each
>inning -- then he faced 24 batters as follows:
>- No outs, bases empty, got an out (six times)
>- One out, bases empty, got an out (six times)
>- Two outs, bases empty, gave up a single (six times)
>- Two outs, runner on first, got an out (six times)
>Now suppose the first six batters in the first inning all got singles,
>everybody went station-to-station, and then he struck out the side.
>After the first inning, he was perfect.
That's not the clumping I'm talking about, but it's worth addressing.
When the second batter in that inning came to the plate, he had
nothing to do with producing that situation. The pitcher now has to
pitch from the stretch with a hole on the right side, and it's only
right to say that he has some of the responsbility for that.
...snip...
>> A pitcher's "events" are grouped inherently differently from batters'.
>> A batter who's having a "bad game" (perhaps because he didn't sleep
>> well the night before, or is fighting off a mild cold) will have 5 or
>> 6 PA's affected by it, quite likely not all against the same pitcher.
>> A pitcher in the same situation may struggle through 25 or 30 PA. The
>> same is true for any event that affects an individual game (rain
>> delay, cold or wet weather, bad umpiring, etc.). This effect tends to
>> reduce the independence of pitchers' events much more than hitters'.
>No, no, no, Mike! Your imagination is working overtime. Players are
>out there every day, and they go through batting slumps and hot streaks
>that last for several games. Starting pitchers only work once every
>five days. Relievers are somewhere in the middle.
They do? Players never have a bad night's sleep, wake up with a stiff
shoulder, play on a wet field? I think you're doing a bit of
handwaving here, James.
...snip...
>Specifically, I point to the hoops people jump through to handle
>inherited runners ... it's like Ptolemy's epicycles ... and the utter
>sinfulness (if I may wax religious) of letting a pitcher off the hook
>for everything that occurs after a two-out fielding error.
What hoops? The most complicated thing I've ever seen proposed is to
penalize a leaving pitcher (and credit the entering pitcher) with the
number of expected runs for the situation. Hardly Ptolemaic, and quite
a statement for someone who says *my* imagination is working overtime.
Mike Jones | jon...@rpi.edu
As part of the conversion, computer specialists rewrote 1,500
programs; a process that traditionally requires some debugging.
- USA Today, referring to the IRS switchover to a new computer system.