Interesting article
Jason
http://www.timesonline.com/sp1214b.html
As unlikely as it is, what do you think about Martin, Schmidt, Hermansen and
Nunez for Griffey?
Jason
ESEM4
Jason
for Nunez? I was just quoting the article...
--Jason--
ESEM4 wrote in message <19991215102554...@ng-fb1.aol.com>...
>Nunez????
Keith J. DeRenzo
Keith J. DeRenzo wrote in message
>I'd deal Martin, Young and Schmidt, and settle for Cordero or
>Aven as a stopgap 1B, but I doubt that any deal between Seattle and
>the Pirates would unfold like that.
I should clarify that I'd do this *if* I thought that there was a
reasonable chance to resign Junior.
Keith
Keith J. DeRenzo
Jason wrote in message <837437$m...@r02n01.cac.psu.edu>...
Actually, the article said that the offer *could* include players
*such as* Martin, Schmidt, Hermansen and Nunez.
As much as I would love to see Griffey as a Buc, I would hesitate to
give up Hermansen only becuase it's likely that the Pirates would lose
Junior as a free agent in 2001.
Trading Hermansen would be worth Griffey in terms of talent, but you
can't give up a likely future star for a one-year rental unless you're
a bona fide contender and/or have a truckload of great prospects to
make up for the guy you're dealing away.
Now, I'm fairly convinced that the Cordero signing is a precursor to
Martin being dealt somewhere, but I think that any deal that takes on
Griffey's current salary would have to include Kevin Young and/or Pat
Meares. I'd deal Martin, Young and Schmidt, and settle for Cordero or
Aven as a stopgap 1B, but I doubt that any deal between Seattle and
the Pirates would unfold like that.
Of course, Kevin McC., has given some *extremely* optimistic quotes to
the press about when he thinks a quality revenue sharing agreement
might be reached. Who knows what the Bucs will be able to spend in
2000 and 2001. . .
Side note: The same article says that the Bucs and O's are discussing
Schmidt or Cordova for BJ Surhoff. Hopefully one of two things
happened here:
1) The writer misspelled Calvin Pickering.
2) The Cordero signing put an end to this potential tragedy. Yeesh.
Keith
Jason
Keith J. DeRenzo wrote in message
>*such as* Martin, Schmidt, Hermansen and Nunez.
Well, obviously any potential deal is hypothetical at this point, unless of
course Cam has some secret deal with the Beaver County Times.
>As much as I would love to see Griffey as a Buc, I would hesitate to
>give up Hermansen only becuase it's likely that the Pirates would lose
>Junior as a free agent in 2001.
And even if they were able to pull a multi-year deal with Griffey out of
their - uh - behinds, it would seal the fates of Kendall and Giles.
>Trading Hermansen would be worth Griffey in terms of talent, but you
>can't give up a likely future star for a one-year rental unless you're
>a bona fide contender and/or have a truckload of great prospects to
>make up for the guy you're dealing away.
>
>Now, I'm fairly convinced that the Cordero signing is a precursor to
>Martin being dealt somewhere, but I think that any deal that takes on
>Griffey's current salary would have to include Kevin Young and/or Pat
>Meares. I'd deal Martin, Young and Schmidt, and settle for Cordero or
>Aven as a stopgap 1B, but I doubt that any deal between Seattle and
>the Pirates would unfold like that.
Is there any reason to believe either can play first base? Has either done
it before proffesionally? I'm not the biggest KY fan but the thought of
Cordero at first gives me shivers.
>Of course, Kevin McC., has given some *extremely* optimistic quotes to
>the press about when he thinks a quality revenue sharing agreement
>might be reached. Who knows what the Bucs will be able to spend in
>2000 and 2001. . .
I'd be satisfied if they spend enough to keep the *good* players they
already have.
--Jason--
Love Rhino
>
> Is there any reason to believe either can play first base? Has either
done
> it before proffesionally? I'm not the biggest KY fan but the thought
of
> Cordero at first gives me shivers.
Cordero played about half of 1998 at 1B for the White Sox. From what i
saw (White Sox second favorite team) and from what i've heard/read
Cordero is a surprisingly mediocre fielder at first for a converted
middle infielder. I remember hearing about him having special problems
in picking up throws in the dirt...
Voros will remember this better since he watches more Sox games than
i...
-ken
p.s. Of course, he doesn't hit anywhere near enough for more than spot
duty at 1B...
Voros
> Cordero played about half of 1998 at 1B for the White Sox. From what i
> saw (White Sox second favorite team) and from what i've heard/read
> Cordero is a surprisingly mediocre fielder at first for a converted
> middle infielder. I remember hearing about him having special problems
> in picking up throws in the dirt...
> Voros will remember this better since he watches more Sox games than
> i...
I do. Cordero is much like Michael Barrett in that what appears like
versatility on the surface is actually complete incompetence at any
defensive position he's asigned to. I believe it is related to his not
particularly caring about his fielding.
The real problem with Cordero was his presence keeping Jeff Abbott out of
the regular lineup until late 1998. If Abbott would have got himself more
firmly established, he might have survived Robespierre Manuel's reign of
terror (Manuel switched starters at four different positions a month into
the season: LF, CF, 1B and 3B. He eventually went back to the 1B and 3B he
started with.)
> p.s. Of course, he doesn't hit anywhere near enough for more than spot
> duty at 1B...
He doesn't hit anywhere near enough to play anywhere but the midlle
infield. He doesn't field at all. An absolutely stupid trade.
p.s. On MLB's official website they have up to date depth charts for all
the teams. In the Pirates section, Cordero was listed as the starting
thirdbaseman. With the acquisition of Aven right before Cordero, I truly
believe this to be the general plans for him. Awful.
--
Voros McCracken
vo...@daruma.co.jp
http://www.enteract.com/~mccracke/dips
Wickdeer
>From: Voros vo...@daruma.co.jp
>Date: Wed, 15 December 1999 11:39 PM EST
>Message-id: <839qen$cgp$1...@eve.enteract.com>
>
SNIP
>
>p.s. On MLB's official website they have up to date depth charts for all
>the teams. In the Pirates section, Cordero was listed as the starting
>thirdbaseman. With the acquisition of Aven right before Cordero, I truly
>believe this to be the general plans for him. Awful.
>
I suspect what it comes down to is a spring training competition between
Hermanson and Ramirez. Winner gets to stay in the majors while the loser goes
to AAA with Cordero taking their spot.
I must say that the prospect of a Martin/Giles/Cordero outfield fills my heart
with defensive dread.
Wick Deer
Erik Simpson
: I suspect what it comes down to is a spring training competition between
: Hermanson and Ramirez. Winner gets to stay in the majors while the loser goes
: to AAA with Cordero taking their spot.
You could well be right, assuming Martin is gone by then. If so, it's a
monumentally stupid way to make such a big decision. I'm confused by the
tendency, both in management statements and in many posts on this group,
to equate the minor league performances of Hermansen and Ramirez. I
really have no problem with Hermansen waiting another year, or part of
one; the evidence I've seen suggests that Aven is a better corner
outfielder right now, and Cordero might be one, too. (Not that I like the
Cordero signing, mind.) Ramirez, on the other hand, has been a genuine
monster in the minors, proving clearly that he should be given a *real*
chance to be a full-time starter in the majors. Letting a small sample of
spring training performance determine the difference between them, rather
than years of evidence in the minors, would be a huge mistake, IMO.
Erik
Chris Conley
Erik Simpson <esim...@dept.english.upenn.edu> wrote:
>Wickdeer (wick...@aol.com) wrote:
>
>: I suspect what it comes down to is a spring training competition
>: between Hermanson and Ramirez. Winner gets to stay in the majors
>: while the loser goes to AAA with Cordero taking their spot.
>You could well be right, assuming Martin is gone by then. If so,
>it's a monumentally stupid way to make such a big decision.
While that's certainly true, it's also a monumentally stupid way
to frame the decisions to be made.
What if they're _both_ ready? Do the Pirates send Hermansen or
(*shudder*) Ramirez back to the minors simply because they "lost"
the head-to-head contest in spring training? Why should there
be no chance for both to win starting jobs?
Frankly, I don't see much of an excuse for sending either one back
to the minors. Ramirez is obviously not going to learn much from
thrashing AAA pitching once again; he needs to be installed at 3B
and left there for the foreseeable future.
But Hermansen's already gone through two AAA campaigns at this
point, and he was reasonably good given that he's *still* young
for the league. Is a third year at the same level going to help
him more than struggling in the majors would?
I do agree that there's a difference in readiness between the two:
Ramirez thrashed AAA and should be able to handle moving on, while
Hermansen's suspect plate discipline means that he may struggle to
make the jump. But I think they both should be given every
opportunity to play full time in the majors NOW rather than be
forced to fight with each other for the right to fight with the
likes of Cordero and Aven for PT.
- Chris
Erik Simpson
: In article <83bbng$ljv$2...@netnews.upenn.edu>,
: Erik Simpson <esim...@dept.english.upenn.edu> wrote:
: >Wickdeer (wick...@aol.com) wrote:
: >
: >: I suspect what it comes down to is a spring training competition
: >: between Hermanson and Ramirez. Winner gets to stay in the majors
: >: while the loser goes to AAA with Cordero taking their spot.
: >You could well be right, assuming Martin is gone by then. If so,
: >it's a monumentally stupid way to make such a big decision.
: While that's certainly true, it's also a monumentally stupid way
: to frame the decisions to be made.
: What if they're _both_ ready? Do the Pirates send Hermansen or
: (*shudder*) Ramirez back to the minors simply because they "lost"
: the head-to-head contest in spring training? Why should there
: be no chance for both to win starting jobs?
Well, from a statheaded perspective, I would answer that Hermansen simply
hasn't shown that he's likely to be a decent major leaguer next year. His
MLE's suck rocks. If Giles couldn't play center, then I would be willing
to accept a somewhat subpar rookie season from Hermansen. But Giles *can*
play center, and the Pirates have at least one better option available for
corner OF.
: Frankly, I don't see much of an excuse for sending either one back
: to the minors. Ramirez is obviously not going to learn much from
: thrashing AAA pitching once again; he needs to be installed at 3B
: and left there for the foreseeable future.
I agree completely, as I said.
: But Hermansen's already gone through two AAA campaigns at this
: point, and he was reasonably good given that he's *still* young
: for the league. Is a third year at the same level going to help
: him more than struggling in the majors would?
I have no idea. He's had one really bad year at AAA and one pretty bad
year. (This is in terms of absolute performance, not adjusting for age.
I hope it's clear that I'm not knocking him as a player in the longer
term; I'm thrilled we have him, just not sure he needs to be a major
league regular yet.) Mightn't it be good for him to improve a bit more,
actually hit AAA pitching *well* for a bit, and then come up later in the
season? I don't know the answer, and as far as I know, we don't have
enough data to give a definitive analytical answer. Or even a tentative
one, to my knowledge. Given those uncertainties, I'm having trouble
finding a problem with failing to give him a starting job right away when,
as far as I can tell, Aven is likely to be a significantly better player
next year.
: I do agree that there's a difference in readiness between the two:
: Ramirez thrashed AAA and should be able to handle moving on, while
: Hermansen's suspect plate discipline means that he may struggle to
: make the jump. But I think they both should be given every
: opportunity to play full time in the majors NOW rather than be
: forced to fight with each other for the right to fight with the
: likes of Cordero and Aven for PT.
Well, I agree with the conclusion because I don't think Ramirez shouldn't
be fighting *anyone* for PT at this point. For me, though, the same
reasoning that leads me to that conclusion (Ramirez's minor league
performance shows me that he's ready to be a valuable contributor right
now, and that he has nothing to gain from mashing AAA pitching again) also
leads me to wonder whether the Hermansen and the Pirates are served best,
in the short and long term, by handing him a job at this point.
Erik
Voros
> : What if they're _both_ ready? Do the Pirates send Hermansen or
> : (*shudder*) Ramirez back to the minors simply because they "lost"
> : the head-to-head contest in spring training? Why should there
> : be no chance for both to win starting jobs?
> Well, from a statheaded perspective, I would answer that Hermansen simply
> hasn't shown that he's likely to be a decent major leaguer next year. His
> MLE's suck rocks.
Well this is a place for discussion, so I suppose this is as good a place
as any to address my problems with MLEs (as useful as I think they are).
MLEs are essentially factors applied to various offensive statistics (.82
plus park and league effects). Anyway the way it works is that Hits (for
example) get multiplied by this factor to get the new hit total and then
added to the AB-H total to get the new AB total. So a player who had 30
hits in 100 At bats in the minors would project to hit .260 in 94.6 at
bats in the majors (if the factor was .82, i.e. neutral leagues, neutral
parks).
Now, my argument with MLEs is that they set the y-intercept to zero for
this linear function OR in english whether the player hit .200 or .350 his
AVG gets adjusted by the same degree. So that in the above example a
player who only had 25 hits in those 100 at bats would project to hit
.215.
Now this works okay for Home Runs and relatively okay for Walks and
Strikeouts, but for batting averages the process breaks down on the
perimeters. One reason is that batting average is a statistic composed of
several other statistics: 1B, 2B, 3B and HR on the hits end and ground
outs, fly outs, strike outs, reached on errors and hits on the other end.
The relationship between each of these statistics and player ability
varies. For example Home Runs is much more related to player ability than
reached on errors.
The fact is that an adjustment to a players batting average needs to vary
in degrees by how high the player's batting average is. A .240 hitter in
AAA should really only downgrade to about .230 in the majors while a .370
hitter might get knocked all the way down to .300 or .290.
We can get more accurate if we break batting average down into components
we're relatively sure about our accuracy and others where there's quite a
bit of fluctuation.
Which leads to my favorite stat in the whole wide world: Batting Average
When Not Striking Out or Homering ("BAWNSOOH"). BAWNSOOH is a stat which
can safely be said to be volatile. Some of the best BAWNSOOH seasons in
the 90's have been posted by Paul O'Neill. Some of the worst BAWNSOOH
seasons in the 90's have been posted by Paul O'Neill. The stat has a wide
degree of fluctuation from year to year for many players (e.g. Al Martin
1997 and Al Martin 1998). If a player posts a very high BAWNSOOH one year,
we need to make sure we don't give it too much weight. Over several years
the stat begins to mean something concrete. For anyone into statistics, a
linear regression of player BAWNSOOHs in back to back seasons from
1946-1998 yields a correlation of only about .39, whereas things like HRs,
BBs and SOs are in the .80 range.
Now if the stat is unreliable at the major league level, one can
hypothesize that it's ability to transfer from the minor to major league
level is also suspect.
To bring this back to Chad Hermansen, major league average BAWNSOOHs
currently reside around .310 for non pitchers. There are subtle ways some
players are better at BAWNSOOH than others. Speed helps some as does
hitting left handed and being young. Chad Hermansen's BAWNSOOH in AAA
this year was .296, which would be below average in the majors. To knock
Hermansen's overall batting average down into the .235 range MLEs would
like to do, you'd have to knock his new major league BAWNSOOH down into
the .270s. I will stats now that young player's with Hermansen's speed
almost never post BAWNSOOHs in the .270s over a full season. As such,
Hermansen's AVG has been way overadjusted by MLEs, whereas something like
Jeremy Giambi's MLE from 1998 was way underadjusted.
The point is that I feel that the MLEs (while generally helpful)
weaknesses are unfairly concentrated in the case of Chad Hermansen. I
suspect Hermansen can hit at least .250 next year in the majors with some
power. The projection I posted a few months back was:
.258/.314/.476
Which I'll stand by for now. It's not as good as Aven's and about the same
as Cordero's. So the question is, "Is the difference in Aven's production
over Hermansen significant enough to keep Hermansen in AAA for a third
year especially considering the Pirates don't figure to contend next
year." I personally don't think so. All I'd ask from Hermansen in ST is
good effort and he gets the job. He has speed and power and his plate
discipline appears to be improving (although still not real good) and the
Pirates aren't pushing any all stars out of the way to let him play.
> If Giles couldn't play center, then I would be willing
> to accept a somewhat subpar rookie season from Hermansen. But Giles *can*
> play center, and the Pirates have at least one better option available for
> corner OF.
But how much better is a matter for debate. An extra 6 runs or so next
year wouldn't be justification enough for me to have Hermansen in AAA.
Erik Simpson
: Erik Simpson <esim...@dept.english.upenn.edu> wrote:
: > : What if they're _both_ ready? Do the Pirates send Hermansen or
: > : (*shudder*) Ramirez back to the minors simply because they "lost"
: > : the head-to-head contest in spring training? Why should there
: > : be no chance for both to win starting jobs?
: > Well, from a statheaded perspective, I would answer that Hermansen simply
: > hasn't shown that he's likely to be a decent major leaguer next year. His
: > MLE's suck rocks.
: Well this is a place for discussion, so I suppose this is as good a place
: as any to address my problems with MLEs (as useful as I think they are).
And thank you for doing so! My goodness, Voros, are you going to publish
this? I've spent a lot of time reading and writing about minor league
translations for years now, and this strikes me as the most interesting
theoretical contribution that's come along in a long time. If you're
right--which I don't doubt--you should be able to put a really cool study
together pretty easily.
[snip explanation of MLE's as STATS and others--including me--have done
them]
: The fact is that an adjustment to a players batting average needs to vary
: in degrees by how high the player's batting average is. A .240 hitter in
: AAA should really only downgrade to about .230 in the majors while a .370
: hitter might get knocked all the way down to .300 or .290.
Gotcha. Is there a quick-and-dirty way to deal with this? For instance,
say a given translation has a factor of .8 for batting average done the
traditional way. Could you take a base value of, say, .200, apply a more
extreme factor, and approximate the effect? Let's call the new factor .4
and see what happens (doing math in my head--please correct mistakes):
Raw BA BA*.8 .200+(BA-.200)*.4
.200 .160 .200
.250 .200 .220
.300 .240 .240
.350 .280 .260
.400 .320 .300
Obviously, that's just a first guess at the kind of number you'd want to
plug in, but that's the kind of clustering effect you're looking for,
right?
[snip intro to BAWNSOOHs]
: Now if the stat is unreliable at the major league level, one can
: hypothesize that it's ability to transfer from the minor to major league
: level is also suspect.
Sure. I follow your logic, and I think it's one of the coolest things
in this area I've seen in a long time. Fantastic! If indeed your
projection is more solid than one based on traditional MLE's, then I, too,
think that the difference between Hermansen and Aven isn't large enough to
justify keeping Chad down again. The Cordero signing gets uglier by the
second, doesn't it?
Erik
Dan Szymborski
says...
> Erik Simpson <esim...@dept.english.upenn.edu> wrote:
> > : What if they're _both_ ready? Do the Pirates send Hermansen or
> > : (*shudder*) Ramirez back to the minors simply because they "lost"
> > : the head-to-head contest in spring training? Why should there
> > : be no chance for both to win starting jobs?
>
> > Well, from a statheaded perspective, I would answer that Hermansen simply
> > hasn't shown that he's likely to be a decent major leaguer next year. His
> > MLE's suck rocks.
[...] (interesting idea snipped)
The next thing we'd have to do is to see if it works. You could
design the study fairly easily. Just find coupled sets of
players with similar BA/OBP/SLG MLEs and get them into high and
low BAWNSOOH groups.
First off, you're going to run into the big problem. Namely,
that while you may feel that BA needs to be adjusted for MLEs
differently, the fact of the matter is that when they're
adjusted the way they currently are, the batting averages
generated by the MLE process *do* correlated extremely well with
future BA. Also, I'm still puzzled by the reaching on errors
thing as errors aren't contained in BA.
I've actually had an ongoing experiment (which I've worked on
for about a year and a half, but don't have enough data right
now to complete) to try to pinpoint types of players with future
value under/overvalued by MLEs (and hopefully major league
stats). Since I've already stimulated the interest of a
publication, I don't want to go further until I can prove it.
I don't have enough data right now to do that, so I'm not much
help there (yet), but I can do a quickie check (a very quickie
check). Chris Dial did an MLE study about a year and a half ago
(which the moron still hasn't written up yet) and since his
study showed MLEs to work for BA/OBP/SLG for these 20 players
already, I'm now going to see what I can do with the data.
I'm going to use some *very* simplistic methods as I don't
believe I have enough data (or good enough data) to use some of
the nifty tools that you can whip out for certain things.
The 20 players involved were taken from the 95/96 season. I
believe that Chris required that there were 250 at-bats in both
seasons (one in the minors and then one in the majors). The
players are Rich Aurilia, Tony Clark, Ron Coomer, Johnny Damon,
Carlos Delgado, Jermaine Dye, Willie Greene, Butch Huskey, Derek
Jeter, Jason Kendall, Matt Lawton, Rey Ordonez, Joe Randa, Rich
Renteria, Alex Rodriguez, F.P. Santangelo, Michael Tucker, Ernie
Young, Jermaine Allensworth, and Darren Bragg.
Obviously, anytime I refer to standard error, I'm talking about
from the MLE year to the MLB year. For the record, I get
standard errors of 24 points for BA, 32 points of OBP, and 52
points for SLG. These are very typical for players in the
majors, too, and is a pretty good example of why anyone only
looking at one year (unless more is unavailable) should be shot
behind the barn.
Substituting BA for BAWNSOOHR (you *really* need a better
acronym so I'm using Voros Batting Average) makes it a little
harder, so I'm going to split the group in half.
The high VBA group consisted of Bragg, Delgado, Jeter, Damon,
Tucker, Rodriguez, Allensworth, Renteria, Kendall, and Aurilia.
Their composite VBA was .343. For the MLEs, the standard error
was 29 points of BA, 33 points of OBP, and 58 points of SLG.
The low VBA group consisted of Clark, Dye, Coomer, Young,
Huskey, Lawton, Randa, Santangelo, Greene, and Ordonez. Their
composite VBA was .291. For the MLEs, the standard error was 19
points of BA, 30 points of OBP, and 46 points of SLG.
For the record, the high VBA group as a whole outperformed their
MLE OPS in the majors the next season by 13 points.
The low VBA group as a whole outperformed their MLE OPS in the
majors the next season by 46 points.
The American League went up 24 points of OPS from 1995 to 1996.
The National League went down a point. 5 of the high VBA group
were American League players. 5 of the low VBA group were
American League players.
Wow, the fun of small sample sizes. I set forth crunching data
that I knew would be inconclusive and came up with inconclusive
results, thus wasting about an hour of Excel crunching in order
to write a really long post that everyone will be half asleep by
the 2nd paragraph.
In other words, keep working on it, VM. Get more data and a
catchy name. How about McCraverage?
> > If Giles couldn't play center, then I would be willing
> > to accept a somewhat subpar rookie season from Hermansen. But Giles *can*
> > play center, and the Pirates have at least one better option available for
> > corner OF.
>
> But how much better is a matter for debate. An extra 6 runs or so next
> year wouldn't be justification enough for me to have Hermansen in AAA.
>
>
--
Dan Szymborski
Cze...@msn.com
Voros
says...
>: Well this is a place for discussion, so I suppose this is as good a place
>: as any to address my problems with MLEs (as useful as I think they are).
>
>And thank you for doing so! My goodness, Voros, are you going to publish
>this?
My publishing company's in the red so I might have trouble. :)
>I've spent a lot of time reading and writing about minor league
>translations for years now, and this strikes me as the most interesting
>theoretical contribution that's come along in a long time.
Thanks.
>If you're right--which I don't doubt--you should be able to put a really
>cool study together pretty easily.
It isn't that easy. I'm in the process of combining what I have for age
factors and comparing minor and major league data in consecutive seasons.
Thwarting my efforts has been the lack of usable park data for minor league
parks. Right now I'm downloading and compiling all the home/road splits from
1999. Pain in the ass.
>
>[snip explanation of MLE's as STATS and others--including me--have done
>them]
>
>: The fact is that an adjustment to a players batting average needs to vary
>: in degrees by how high the player's batting average is. A .240 hitter in
>: AAA should really only downgrade to about .230 in the majors while a .370
>: hitter might get knocked all the way down to .300 or .290.
>
>Gotcha. Is there a quick-and-dirty way to deal with this? For instance,
>say a given translation has a factor of .8 for batting average done the
>traditional way. Could you take a base value of, say, .200, apply a more
>extreme factor, and approximate the effect? Let's call the new factor .4
>and see what happens (doing math in my head--please correct mistakes):
>
>Raw BA BA*.8 .200+(BA-.200)*.4
>.200 .160 .200
>.250 .200 .220
>.300 .240 .240
>.350 .280 .260
>.400 .320 .300
>
>Obviously, that's just a first guess at the kind of number you'd want to
>plug in, but that's the kind of clustering effect you're looking for,
>right?
In a sense. The problem is that batting average consists of components that
need to be clustered (BAWNSOOH) and those that don't (HR and SO). For
example, Russ Branyan's .208 is a bit different than Keith Johns'.
>
>[snip intro to BAWNSOOHs]
>
>: Now if the stat is unreliable at the major league level, one can
>: hypothesize that it's ability to transfer from the minor to major league
>: level is also suspect.
>
>Sure. I follow your logic, and I think it's one of the coolest things
>in this area I've seen in a long time. Fantastic!
What is most interesting is the pitching equivalent of BAWNSOOH. Whereas
understanding BAWNSOOH for hitters can be important, understanding it's
equivalent for pitching is VITAL. I've posted several times in the past few
months on this subject in rec.sport.baseball.analysis (under Defense
Independent Pitching Stats). To summarize, the percentage of batted balls in
play that are hits for a pitcher during a season has virtually NO
relationship to pitching ability. As a result any stats that can be affected
by defense (all of them except for BFP, BB, SO, HBP, HR) in any way have a
very low level of correlation to any type of pitching ability. To see how
this could affect your observations on pitchers, look at the:
(H-HR)/(BFP-BB-HBP-SO)
totals for Aaron Sele and Kevin Millwood in 1999. Imagine what happens to
your observations when you suddenly set that figure to equal for both.
Anyway I digress (only because that's what I've been working on). I'm in the
process of trying to set up my own MLEs, but it's a difficult task,
especially when it doesn't have your undivided attention.
For the time being I simply use the standard MLEs for my projections. But
since the projections already take into account the instability of BAWNSOOH
and uses multiple year data, it does a nice job of filtering out the MLEs
weaknesses.
>If indeed your projection is more solid than one based on traditional MLE's,
>then I, too, think that the difference between Hermansen and Aven isn't
>large enough to justify keeping Chad down again. The Cordero signing gets
>uglier by the second, doesn't it?
Yes it does. In addition, STATS projections for Hermansen are not too far
from mine with the bulk of the difference being in Batting Average. A good
rule of thumb for minor leaguers is to not get too wrapped up in their
Batting Averages. By looking at their HR, BB, SO and SB totals and their Age,
you get a very good idea of the various talents of the player.
--
Voros McCracken
vo...@daruma.co.jp
http://www.enteract.com/~mccracke/dips
"As always, victory finds a hundred fathers
but defeat is an Orphan"
-Count Galeazzo Ciano
Voros
Organization:
Dan Szymborski <Cze...@mindspring.com> wrote:
>> > Well, from a statheaded perspective, I would answer that Hermansen simply
>> > hasn't shown that he's likely to be a decent major leaguer next year. His
>> > MLE's suck rocks.
> [...] (interesting idea snipped)
> The next thing we'd have to do is to see if it works. You could
> design the study fairly easily. Just find coupled sets of
> players with similar BA/OBP/SLG MLEs and get them into high and
> low BAWNSOOH groups.
The problem is that the low BAWNSOOH groups might have a much greater
problem getting their playing time in the majors. Nonetheless a good idea.
> First off, you're going to run into the big problem. Namely,
> that while you may feel that BA needs to be adjusted for MLEs
> differently, the fact of the matter is that when they're
> adjusted the way they currently are, the batting averages
> generated by the MLE process *do* correlated extremely well with
> future BA.
Actually, they don't. (Oh my god straight contradiction) They correlate
about as well as major league batting averages. But that does not mean
they correlate well as major league averages really don't. Or to be more
specific BAWNSOOH (VBA, Mr. McCracken's long slow descent into insanity,
whatever you want to call it) correlates not very well for major league
numbers. To wit, I came up with the following for major leaguers in back
to back seasons with 300+ AB since 1946:
$BB=(BB/(AB+BB)), $SO=(SO/AB), $HR=(HR/(AB-SO)) and $H ((H-HR)/(AB-HR-SO)
or BAWNSOOH, VBA, etc.
Here are the linear regression correlations for the four:
$BB=.773
$SO=.863
$HR=.798
$H =.423
There were 6,754 comparisons. This qualifies as statistically significant.
We see that the STAT in question correlates very poorly in comparison to
$BB, $SO and $HR.
The theory is of course that if $H doesn't correlate well with major
league numbers it probably won't do much better with MLEs and major league
numbers.
> Also, I'm still puzzled by the reaching on errors
> thing as errors aren't contained in BA.
I probably was unclear. I tried to set up all the hit scenarios and all
the At Bat scenarios (of which reaching on an error is one).
<intro to MLE discussion snipped>
> The 20 players involved were taken from the 95/96 season. I
> believe that Chris required that there were 250 at-bats in both
> seasons (one in the minors and then one in the majors). The
> players are Rich Aurilia, Tony Clark, Ron Coomer, Johnny Damon,
> Carlos Delgado, Jermaine Dye, Willie Greene, Butch Huskey, Derek
> Jeter, Jason Kendall, Matt Lawton, Rey Ordonez, Joe Randa, Rich
> Renteria, Alex Rodriguez, F.P. Santangelo, Michael Tucker, Ernie
> Young, Jermaine Allensworth, and Darren Bragg.
> Obviously, anytime I refer to standard error, I'm talking about
> from the MLE year to the MLB year. For the record, I get
> standard errors of 24 points for BA, 32 points of OBP, and 52
> points for SLG. These are very typical for players in the
> majors, too, and is a pretty good example of why anyone only
> looking at one year (unless more is unavailable) should be shot
> behind the barn.
Ahh but when you talk about stats from one year, some (HR, BB and SO) are
surprisingly reliable as shown above.
> Substituting BA for BAWNSOOHR (you *really* need a better
> acronym so I'm using Voros Batting Average)
VBA. I like it, though is this microsoft's acronym for that monster they
call visual basic? BAWNSOOH was dubbed by Doug Drinen who has more clout
than me so I went with it.
> makes it a little
> harder, so I'm going to split the group in half.
> The high VBA group consisted of Bragg, Delgado, Jeter, Damon,
> Tucker, Rodriguez, Allensworth, Renteria, Kendall, and Aurilia.
> Their composite VBA was .343. For the MLEs, the standard error
> was 29 points of BA, 33 points of OBP, and 58 points of SLG.
> The low VBA group consisted of Clark, Dye, Coomer, Young,
> Huskey, Lawton, Randa, Santangelo, Greene, and Ordonez. Their
> composite VBA was .291. For the MLEs, the standard error was 19
> points of BA, 30 points of OBP, and 46 points of SLG.
> For the record, the high VBA group as a whole outperformed their
> MLE OPS in the majors the next season by 13 points.
> The low VBA group as a whole outperformed their MLE OPS in the
> majors the next season by 46 points.
> The American League went up 24 points of OPS from 1995 to 1996.
> The National League went down a point. 5 of the high VBA group
> were American League players. 5 of the low VBA group were
> American League players.
> Wow, the fun of small sample sizes. I set forth crunching data
> that I knew would be inconclusive and came up with inconclusive
> results, thus wasting about an hour of Excel crunching in order
> to write a really long post that everyone will be half asleep by
> the 2nd paragraph.
I followed it and although it's inconclusive, it's inconclusive in favor
of my theory so if a lot more data is added giving the same level of
results it would eventually start to indicate something.
> In other words, keep working on it, VM. Get more data and a
> catchy name. How about McCraverage?
Interestingly enough, Quinton's VBA is a very high .344 (one of the
reason's to be afraid of him) so maybe using my last name might not be a
bad idea as I can always say I named it after him.
For now, I'll stick with VBA or maybe the Perlish $H.
Dan Szymborski
says...
> ews.mindspring.com>:
> Organization:
>
> Dan Szymborski <Cze...@mindspring.com> wrote:
> >> > Well, from a statheaded perspective, I would answer that Hermansen simply
> >> > hasn't shown that he's likely to be a decent major leaguer next year. His
> >> > MLE's suck rocks.
>
> > [...] (interesting idea snipped)
>
> > The next thing we'd have to do is to see if it works. You could
> > design the study fairly easily. Just find coupled sets of
> > players with similar BA/OBP/SLG MLEs and get them into high and
> > low BAWNSOOH groups.
>
> The problem is that the low BAWNSOOH groups might have a much greater
> problem getting their playing time in the majors. Nonetheless a good idea.
>
> > First off, you're going to run into the big problem. Namely,
> > that while you may feel that BA needs to be adjusted for MLEs
> > differently, the fact of the matter is that when they're
> > adjusted the way they currently are, the batting averages
> > generated by the MLE process *do* correlated extremely well with
> > future BA.
>
> Actually, they don't. (Oh my god straight contradiction) They correlate
> about as well as major league batting averages.
Well, I worded that badly.
> But that does not mean
> they correlate well as major league averages really don't. Or to be more
> specific BAWNSOOH (VBA, Mr. McCracken's long slow descent into insanity,
> whatever you want to call it) correlates not very well for major league
> numbers. To wit, I came up with the following for major leaguers in back
> to back seasons with 300+ AB since 1946:
Keep up the work. WHen you get something concrete, do tell!
--
Dan Szymborski
Cze...@msn.com