Rating Defense or Does Rey's glove make up for his bat? (long-ish)
Chris Dial
through the asb-mets group has seen a Rey ROOLZ! thread. We split into two
teams, shirts and skins. No, it was statheads and morons. Unfortunately, I
had to be the captain of the moron side.
After many skirmishes and many moron casualties (fortunately, we have
infinite resources; just add WebTV), the battle came down to one thing: does
Rey's defense make up for his bat?
I took on this small thing by using a Total Player Ranking system. For each
player, I calculated his RC. After some discussion with the statheads'
peace negotiation committee, adjusted every player's RC to a season's worth
of outs. This turns out to be ~465. I then normalized every batter's RC to
RC/465, and adjusted for STATS 3-yr. park factor for runs. This is the RC a
player would contribute if he were Cal Ripken: played every inning of every
game. What's frightening is who the average player is: Eric Karros. Karros
played every inning, and hit .266 (lge avg) with an OBP of .329 (~lge avg).
It's a shame he isn't a shortstop. He actually played in enough innings to
count 165 games, and he made 480 outs, just above average. Thanks, Eric,
for your help.
A player's offensive value is PRC+ (park-adjusted, normalized to 465 runs
created).
Defense is a real bear to quantify. I used to score for STATS, so I have a
really good grasp of what they do and how well it works (in terms of
measuring balls in play). I think it is an excellent scoring system. How
STATS happens to sort the data and assign it isn't the best, but it is what
we have.
I do apologize to Dale Stephenson for not studying his work more, and I may
be making some errors he knows how to fix. Hopefully, he'll chime in with
some fixes.
I planned to normalize the defense as well. To do this, I have to establish
how many plays each position will get during the course of a season. And a
defensive season is defined by the Cal Ripken season: every inning of every
game. That is 162*8.75 = 1418. For each position, I determine the average
# of plays per game each defender playing in 130+ games gets. That average
multiplied by 162 is the number of plays each position is normalized to
(sometimes I fudge to make it a round number). That makes each position
with a different number. Here are the plays per season a Ripken can expect:
Pos AL NL
1B: 300 300 Note: the raw numbers here are 1.81 ppg and 1.82 ppg. Wow.
2B: 500 515
3B: 445 425
SS: 520 505
LF: 395 360
CF: 475 450
RF: 375 350
AL: 3010
NL: 2905
Pretty amazing that AL teams get just 100 more chances over 162 games.
For infielders, everything tends to even out. Since I normalize the chances,
lefty-righty staffs and groundball/flyball staffs don't affect the data too
much. It is possible that the players used to calc the seasons chances play
for LHP/RHP or GB/FB, but each was averaged with at least four starters.
Hard hit balls, tough chances etc tend to even out. I'd rather have the raw
data (plays in each zone for every player), but I takes what I gots.
Nonetheless, the plays I do have are certainly outs. At worst, ZR
undervalues defense (depending on pop-ups and line drives, and DPs).
With these season total of plays, I multiply by the fielder's ZR. That's
how many plays he would have made. I think this converts to runs easily,
and almost obviously: any play not made results in a baserunner.
Each out has a value: if it takes away a single, that's worth 0.47 (hit) +
0.27 (out). Each subsequent base is worth 0.31 runs (per Linear Wts)
SS/2B: ((plays made-2)*0.74)+(2*1.05)
1B/3B: (0.8*pm*0.74)+(0.18*pm*1.05)+(0.02*pm*1.36)
OF: [(0.76*pm*0.74)+(0.213*pm*1.05)+(0.026*pm*1.36)+assists*0.62]*pitf
pitching staff factor is calculated by: SLG+ = (total bases-(tb from
hr))/(ab-K-hr)
then set as a percentage of league average (less that team's SLG+)
This increases the value of plays made by OF who play in screwy parks or
behind bad pitchers. I am pretty sure it removes park effects mostly.
I compare the players to one another, rather than replacement value. I do
this mostly because I want to know who is better, not who needs to be
replaced, and partly because establishing defense for a replacement player
is damned near impossible. My experience with minor league official scorers
is that they suck. Not just in assigning errors, but even paying attention
to the freakin' game. And, yes, I would do a better job.
Basically, I can only really evaluate players with enough PT. I am in the
process of calculating the Defensive Runs Saved for 1996, so I can attempt
to establish the above plays per season. How much those numbers fluctuate
is not known to me.
And the answer is: Rey's DTPR: -21, worst in the league. His defense was
actually second to Royce Clayton, but his offense was a whopping -40 RC vs.
average.
After this post on the Mets' ng, the morons ostracized me. A man without a
country...
Chris Dial
Sean Lahman
> I took on this small thing by using a Total Player Ranking system.
Maybe you could pick a different name? That's what Pete Palmer calls
the output from his linear weights technique.
And though many people don't like his methods, Pete's attempt is to do
the same thing - rate players combined offense and defense.
--
Sean Lahman / se...@baseball1.com
The Baseball Archive - http://www.baseball1.com
Dan Szymborski
> Chris Dial wrote:
> > I took on this small thing by using a Total Player Ranking system.
>
> Maybe you could pick a different name? That's what Pete Palmer calls
> the output from his linear weights technique.
I'm fairly certain that Chris hasn't decided on a name for his system yet.
It seems fairly obvious to me that while the wording may not be ideal,
saying "a" Total Player Ranking system just meant that he knew it was
similar to Palmer's system. Using linear weights on defense and utilizing
zone rating, however, is a more interesting approach than the wacky
adjusted range factors tha TB uses.
> And though many people don't like his methods, Pete's attempt is to do
> the same thing - rate players combined offense and defense.
>
>
>
--
Dan Szymborski
Cze...@email.msn.com
"I would like to hear Elliot Carter's Fourth String Quartet, if only to
discover what a cranky prostate does to one's polyphony."
-- James Sellars
Ron Johnson
>As a Mets fan, I really enjoy watching Rey play defense. Anyone traveling
>through the asb-mets group has seen a Rey ROOLZ! thread. We split into two
>teams, shirts and skins. No, it was statheads and morons. Unfortunately, I
>had to be the captain of the moron side.
So how did we end up with Ken V. ? (I keep trying to trade him for
Barry M. Still no takers.)
(snippage)
>SS/2B: ((plays made-2)*0.74)+(2*1.05)
As I've commented before, I'd like to see a correction for DP
opportunities. ZR counts a DP as two plays made (reasonable)
but has no adjustment for the fact that the DP's in order
more frequently on some teams than others.
Still I haven't worked out how and the errors that this causes
aren't going to be huge.
>1B/3B: (0.8*pm*0.74)+(0.18*pm*1.05)+(0.02*pm*1.36)
>OF: [(0.76*pm*0.74)+(0.213*pm*1.05)+(0.026*pm*1.36)+assists*0.62]*pitf
>
>pitching staff factor is calculated by: SLG+ = (total bases-(tb from
>hr))/(ab-K-hr)
>then set as a percentage of league average (less that team's SLG+)
>
>This increases the value of plays made by OF who play in screwy parks or
>behind bad pitchers. I am pretty sure it removes park effects mostly.
Might be a problem in Fenway or in places with a sun field. IE in
parks where one of the fields is more difficult. Still, I think it's
a big step forward.
The real answer as you say is to get the data. Before she bowed out,
Sherri Nichols offered me (and a few others) the raw DA data to try
and get a handle on DA related park effects. Sadly it never happened.
(If this sounds like I'm complaining, I'm not. She was a new mom -
she had things to do. If you're lurking, we miss you.)
<'Nother snip>
(Rey comes out horrid.)
Of course after the style points for his perfect Telemark landing ...
No wait that's wrestling or something.
>After this post on the Mets' ng, the morons ostracized me. A man without a
>country...
What do you expect?
Best kind of stahead stuff. Right up to reporting results that you didn't
agree with before starting the study.
Dan, you did give him the secret handshake right?
--
RNJ
Chris Dial
>Chris Dial wrote:
>> I took on this small thing by using a Total Player Ranking system.
>
>Maybe you could pick a different name? That's what Pete Palmer calls
>the output from his linear weights technique.
I tried, unsuccessfully. How about Dial Ratings...is that ostentatious
enough? I am taking suggestions. My wife suggested "Huge Wastes of Time".
>
>And though many people don't like his methods, Pete's attempt is to do
>the same thing - rate players combined offense and defense.
I think that is the only "fair" way to do it. One of the general knocks on
any player value system is that player R saves more runs with his glove than
he costs with his bat. Every decent newsgroup has this argument. How about
the Angels' ng standard squabble over DiSarcina? The problem (to me) with
ZR is that fielders get shortchanged...that is save more runs than show up
in ZR. Probably not a lot more, but I am working on that as well.
How was Gary? He was the best defensive SS in the AL (+17 runs over
average). Unfortunately he was -34 runs with his bat. Basically as bad as
Rey...
It has been mentioned to me that _really_ good players skew the average,
making poor players look worse. AL SS has ARod and Nomar and Jeter, and Jay
Bell had a good season. But no, DiSar's defense doesn't make up for his
bat.
I don't think the Dial Ratings (better?) are perfect, and am open to
suggestions for improvement, but I'd like to keep it simple. Most of us can
do Tech-RCs, and can get ZRs. Once you extrapolate everyone to equal outs
and opportunities, you can size up who you'd rather have.
Chris Dial
Chris Dial
>In article <6s9sco$bp4$1...@supernews.com>, Chris Dial <acd...@intrex.com>
wrote:
>>As a Mets fan, I really enjoy watching Rey play defense. Anyone traveling
>>through the asb-mets group has seen a Rey ROOLZ! thread. We split into
two
>>teams, shirts and skins. No, it was statheads and morons. Unfortunately,
I
>>had to be the captain of the moron side.
>
>So how did we end up with Ken V. ? (I keep trying to trade him for
>Barry M. Still no takers.)
We've offered spannerjaxs, and Hank (3838?). I think that's a little more
fair...
>
>(snippage)
>
>
>>SS/2B: ((plays made-2)*0.74)+(2*1.05)
>
>As I've commented before, I'd like to see a correction for DP
>opportunities. ZR counts a DP as two plays made (reasonable)
>but has no adjustment for the fact that the DP's in order
>more frequently on some teams than others.
I haven't looked at this closely enough.
>
>Still I haven't worked out how and the errors that this causes
>aren't going to be huge.
>
>>1B/3B: (0.8*pm*0.74)+(0.18*pm*1.05)+(0.02*pm*1.36)
I think some adjustment should be made for 1B and infield throwing errors.
The Mariners' ng is often pissin' and moanin' about how Segui saves all
these throwing errors (and still Davis has 120). Playing SS for my 30+
team, I know what it's like to go from a 6'4" lifetime 1B to a 5'7"
short-armed former SS (that plays first because he's the manager). It's
like throwing it through the clown's mouth. Sure, my situation is
exagerrated, but I know the good 1B saves *tons* of errors. Every scoop is
a throwing error, and there are several chances per game per 1B.
>>OF: [(0.76*pm*0.74)+(0.213*pm*1.05)+(0.026*pm*1.36)+assists*0.62]*pitf
>>
>>pitching staff factor is calculated by: SLG+ = (total bases-(tb from
>>hr))/(ab-K-hr)
>>then set as a percentage of league average (less that team's SLG+)
>>
>>This increases the value of plays made by OF who play in screwy parks or
>>behind bad pitchers. I am pretty sure it removes park effects mostly.
>
>Might be a problem in Fenway or in places with a sun field. IE in
>parks where one of the fields is more difficult. Still, I think it's
>a big step forward.
Even in Fenway, if the pitchers surrender a bunch of 2B, it will show up in
their pitf. The Metrodome has that bad roof, too.
>
>The real answer as you say is to get the data. Before she bowed out,
>Sherri Nichols offered me (and a few others) the raw DA data to try
>and get a handle on DA related park effects. Sadly it never happened.
Some of you know her. I'll be happy to go through the data, I very familiar
with the Project Scoresheet system, and it would really help with looking at
other things.
><'Nother snip>
>
>(Rey comes out horrid.)
>
>Of course after the style points for his perfect Telemark landing ...
>
>No wait that's wrestling or something.
>
>>After this post on the Mets' ng, the morons ostracized me. A man without
a
>>country...
>
>What do you expect?
>
>Best kind of stahead stuff. Right up to reporting results that you didn't
>agree with before starting the study.
I was stunned Rey still came up so bad. I did the same thing cursing
McMichael the previous year...
>
>Dan, you did give him the secret handshake right?
I'm still planning that trip to Baltimore (to pick up my asb-bo's prediction
award, cough cough)...
Still on the burner: catchers. That is some tough defense to analyze...and
the question is: does Rodriguez' defense make up for his bat vs. Piazza?
Chris Dial
Chris Dial
exaggerated, but I know the good 1B saves *tons* of errors. Every scoop is
NawrockiT
: I think some adjustment should be made for 1B and infield throwing errors.
: The Mariners' ng is often pissin' and moanin' about how Segui saves all
: these throwing errors (and still Davis has 120). Playing SS for my 30+
: team, I know what it's like to go from a 6'4" lifetime 1B to a 5'7"
: short-armed former SS (that plays first because he's the manager). It's
: like throwing it through the clown's mouth. Sure, my situation is
: exaggerated, but I know the good 1B saves *tons* of errors. Every scoop is
: a throwing error, and there are several chances per game per 1B.
What complicates matters is that these errors are never charged to the 1B, so
you can't even look at his errors to see whether he can catch or not. Tino
Martinez is terrible at catching throws that bounce into the dirt, but of
course he's never held responsible for it.
Tino does have ten errors this year, but he also should get credit for a lot of
Brosius' 22 errors.
Tom Nawrocki
Chris Dial
<199809011412...@ladder01.news.aol.com>...
>Chris Dial wrote:
>
>: I think some adjustment should be made for 1B and infield throwing
errors.
>: The Mariners' ng is often pissin' and moanin' about how Segui saves all
>: these throwing errors (and still Davis has 120). Playing SS for my 30+
>: team, I know what it's like to go from a 6'4" lifetime 1B to a 5'7"
>: short-armed former SS (that plays first because he's the manager). It's
>: like throwing it through the clown's mouth. Sure, my situation is
>: exaggerated, but I know the good 1B saves *tons* of errors. Every scoop
is
>: a throwing error, and there are several chances per game per 1B.
>
>What complicates matters is that these errors are never charged to the 1B,
so
>you can't even look at his errors to see whether he can catch or not. Tino
>Martinez is terrible at catching throws that bounce into the dirt, but of
>course he's never held responsible for it.
Sorry, Tom, I may have misled you where I was going with that.
>
>Tino does have ten errors this year, but he also should get credit for a
lot of
>Brosius' 22 errors.
And he can. Again, it won't be a perfect system, but instead of looking at
scoops (those will tend to even out over 160 games), look at infield
throwing errors. Ryne Sandberg didn't have a throwing error for like 10
seasons. It wasn't just because he made good throws; having a good first
baseman helped tremendously. Last season, Seattle had 68 throwing errors.
Some were OF throws, but probably few, so there were probably 55 throwing
errors, with Paul Sorrento at first. In spite of Davis' arm, how many
throwing errors did Cora and ARod make this season versus last season?
STATS has these numbers. How they are distributed is the only problem with
better quantification. Since the most throwing errors for a team is 69
(Boston - Vaughn! and Sand Diego - Joyner), and the low is 34, there should
be an available breakdown. We may be talking about 10 plays a year on
average, but you have a swing of 30 baserunners/outs worth nearly 22 runs.
Just trying to piece it together.
Chris Dial
Samson
> better quantification. Since the most throwing errors for a team is 69
> (Boston - Vaughn!...
That's just Wakefield throwing him all those knucklers...
Dale J. Stephenson
[I like the concept. I have meant to getting around to creating a ZR-based
DR, since DA data has dried up and it beats using Total Baseball...]
[snip]
> I planned to normalize the defense as well. To do this, I have to establish
> how many plays each position will get during the course of a season. And a
> defensive season is defined by the Cal Ripken season: every inning of every
> game. That is 162*8.75 = 1418. For each position, I determine the average
> # of plays per game each defender playing in 130+ games gets. That average
> multiplied by 162 is the number of plays each position is normalized to
> (sometimes I fudge to make it a round number). That makes each position
> with a different number. Here are the plays per season a Ripken can expect:
> Pos AL NL
> 1B: 300 300 Note: the raw numbers here are 1.81 ppg and 1.82 ppg. Wow.
> 2B: 500 515
> 3B: 445 425
> SS: 520 505
> LF: 395 360
> CF: 475 450
> RF: 375 350
>
> AL: 3010
> NL: 2905
>
> Pretty amazing that AL teams get just 100 more chances over 162 games.
>
What is considered a "play" here? ZR opportunities, or an estimate
of chances started? I'm guessing ZR opportunities, since otherwise using
a ZR multiple doesn't work. My scoreboards are packed away right now,
but IIRC the lists in back just have the league average, not the total
opportunities and outs. Am I remembering incorrectly?
> For infielders, everything tends to even out. Since I normalize the chances,
> lefty-righty staffs and groundball/flyball staffs don't affect the data too
> much. It is possible that the players used to calc the seasons chances play
> for LHP/RHP or GB/FB, but each was averaged with at least four starters.
> Hard hit balls, tough chances etc tend to even out. I'd rather have the raw
> data (plays in each zone for every player), but I takes what I gots.
> Nonetheless, the plays I do have are certainly outs. At worst, ZR
> undervalues defense (depending on pop-ups and line drives, and DPs).
>
Do DPs still contaminate the ZR opportunities?
> With these season total of plays, I multiply by the fielder's ZR. That's
> how many plays he would have made. I think this converts to runs easily,
> and almost obviously: any play not made results in a baserunner.
>
> Each out has a value: if it takes away a single, that's worth 0.47 (hit) +
> 0.27 (out). Each subsequent base is worth 0.31 runs (per Linear Wts)
>
OK, I follow this so far...
> SS/2B: ((plays made-2)*0.74)+(2*1.05)
The doubles are fixed for ss/2b? Or is there a pm in there?
> 1B/3B: (0.8*pm*0.74)+(0.18*pm*1.05)+(0.02*pm*1.36)
Per DA data, giving up multiple bases at first and third is not
a function of DA -- the "guarding the line" effect. Lacking specific
data about doubles and triples through first allowed, though, it's
a good approximation.
> OF: [(0.76*pm*0.74)+(0.213*pm*1.05)+(0.026*pm*1.36)+assists*0.62]*pitf
>
> pitching staff factor is calculated by: SLG+ = (total bases-(tb from
> hr))/(ab-K-hr)
> then set as a percentage of league average (less that team's SLG+)
>
> This increases the value of plays made by OF who play in screwy parks or
> behind bad pitchers. I am pretty sure it removes park effects mostly.
>
The doubles/triples allowed versus DA is a big variable in the DA reports.
Again, lacking the specific 2b/3b allowed data this is a reasonable
estimate, but it introduces a huge possible area of error. FWIW, the only
outfielder in the DA reports who was good enough at preventing doubles/triples
to negate his poor range was Raul Mondesi.
I don't think assists should be here, at least without folding in
extra bases taken from the "holding runner" section of the scoreboard.
The runners taking extra bases on the fielder have to be considered
along with the assists he gets nailing some of them.
I'm a bit leery of the pitching factor. SLG+ is a function of fielding
as well as pitching. If you have an outfield of poor-range outfielders
who are positioned too shallowly, the resulting bad performance of the
pitchers will make the lot of poor-range outfielders look average.
Also, I would bet groundball pitchers have a SLG+ substantially superior
to that of flyball pitchers. A groundball staff (especially an
excellent groundball staff -- Atlanta?) should have a way low SLG+, and
a resulting degradation of their outfielder's performance, through no
fault of their own. What we need is real park effects....
I'm of the unsupported opinion that when you take out walks, strikeouts,
homers, flyball/groundball tendencies, and the running game, that fielding
is much more important for determining safe/out than pitching. I
believe strongly in park effects, but I'm unconvinced that there really
is a large "bad pitcher" effect on DA/ZR-like measures.
> I compare the players to one another, rather than replacement value. I do
> this mostly because I want to know who is better, not who needs to be
> replaced, and partly because establishing defense for a replacement player
> is damned near impossible. My experience with minor league official scorers
> is that they suck. Not just in assigning errors, but even paying attention
> to the freakin' game. And, yes, I would do a better job.
>
I think average is the correct baseline defensively. Although there may
be a rookie effect on fielding (see the DRs of most shortstops, and guys
like Griffey), a "replacement fielder" includes guys like Rafael Belliard
too.
--
Dale J. Stephenson * dst...@sirius.com * past his prime
"I know nothing, Colonel Turner, nothing."
-- Sgt. Schuerholz
Ron Johnson
Dale J. Stephenson <dst...@sirius.com> wrote:
>Do DPs still contaminate the ZR opportunities?
Yeah. Counts as two plays made (reasonable) but there's no adjustments
for how frequently the DP is in order. Better still would be to have
DPs treated as DA did.
As Chris has commented, the data is there.
--
RNJ
Ron Johnson
Dale J. Stephenson <dst...@sirius.com> wrote:
>"Chris Dial" <acd...@intrex.com> wrote:
>
>I'm a bit leery of the pitching factor. SLG+ is a function of fielding
>as well as pitching. If you have an outfield of poor-range outfielders
>who are positioned too shallowly, the resulting bad performance of the
>pitchers will make the lot of poor-range outfielders look average.
>
>Also, I would bet groundball pitchers have a SLG+ substantially superior
>to that of flyball pitchers.
Something more to check. I've started a study on pitching/defense
retationship. Turned out to be way more work than I thought.
I can tell you that there is a tiny correlation between the GB/FB
out rate of the pitcher and the rate that the defense converts
opportunities into outs. Or to put it in English, groundball pitchers
tend to make the defense look better. But the effect is minor as best I
can tell.
>A groundball staff (especially an excellent groundball staff -- Atlanta?)
>should have a way low SLG+, and a resulting degradation of their
>outfielder's performance, through no fault of their own. What we need
>is real park effects....
I'm with you here.
>
>I'm of the unsupported opinion that when you take out walks, strikeouts,
>homers, flyball/groundball tendencies, and the running game, that fielding
>is much more important for determining safe/out than pitching. I
>believe strongly in park effects, but I'm unconvinced that there really
>is a large "bad pitcher" effect on DA/ZR-like measures.
That's not what I've found. (cautionary note. I'm digging up more data)
I've found wildly differing rates for pitchers on the same staffs.
I don't have error data though so I'm in the process of working out
estimated errors - (raj you've muttered something about this. Details?)
maybe this will smooth things out.
Right now what I have is that there's no meaningful correlation between the
team a pitcher pitches for and the rate that balls put into play are
converted into outs. Surprised me.
--
RNJ
Chris Dial
>In article <dsteph-0309...@ppp-asok06--070.sirius.net>,
>Dale J. Stephenson <dst...@sirius.com> wrote:
>>"Chris Dial" <acd...@intrex.com> wrote:
>>
>>I'm a bit leery of the pitching factor. SLG+ is a function of fielding
>>as well as pitching. If you have an outfield of poor-range outfielders
>>who are positioned too shallowly, the resulting bad performance of the
>>pitchers will make the lot of poor-range outfielders look average.
>>
>>Also, I would bet groundball pitchers have a SLG+ substantially superior
>>to that of flyball pitchers.
Nope. Certainly not at the team level.
>
>Something more to check. I've started a study on pitching/defense
>retationship. Turned out to be way more work than I thought.
Here's my list, sorted by gb/fb rate:
Tm SLG+ gb/fb
Cin 1.001 1.122
PhN 1.095 1.157
LA 0.938 1.200
ChN 1.013 1.207
Fla 0.988 1.280
SF 0.986 1.298
Mon 0.964 1.352
Atl 0.928 1.405
StL 0.977 1.459
NYM 0.969 1.545
SD 1.046 1.551
Col 1.099 1.582
Hou 0.953 1.608
Pit 1.044 1.778
The average SLG+ for teams with *low* gb/fb rates was 0.998. Oddly, the
distribution was exactly the same in the AL (0.998 ave SLG+ for low gb/fb)
>
>I can tell you that there is a tiny correlation between the GB/FB
>out rate of the pitcher and the rate that the defense converts
>opportunities into outs. Or to put it in English, groundball pitchers
>tend to make the defense look better. But the effect is minor as best I
>can tell.
>
>>A groundball staff (especially an excellent groundball staff -- Atlanta?)
>>should have a way low SLG+, and a resulting degradation of their
>>outfielder's performance, through no fault of their own. What we need
>>is real park effects....
>
>I'm with you here.
I'm not so sure...
>>
>>I'm of the unsupported opinion that when you take out walks, strikeouts,
>>homers, flyball/groundball tendencies, and the running game, that fielding
>>is much more important for determining safe/out than pitching. I
>>believe strongly in park effects, but I'm unconvinced that there really
>>is a large "bad pitcher" effect on DA/ZR-like measures.
>
>That's not what I've found. (cautionary note. I'm digging up more data)
>I've found wildly differing rates for pitchers on the same staffs.
>I don't have error data though so I'm in the process of working out
>estimated errors - (raj you've muttered something about this. Details?)
>maybe this will smooth things out.
>
>Right now what I have is that there's no meaningful correlation between the
>team a pitcher pitches for and the rate that balls put into play are
>converted into outs. Surprised me.
I looking forward to more defense data...
Chris Dial
Chris Dial
It is ZR opportunities. I don't have all the Scoreboards, so I'm not sure
how they are listed. STATS seems to change that format yearly. I got my #
of plays by taking all players at a given position that played 130+
defensive games (Innings/8.75) and averaging their individual chances per
game. I then multiplied by 162 games.
>
>> For infielders, everything tends to even out. Since I normalize the
chances,
>> lefty-righty staffs and groundball/flyball staffs don't affect the data
too
>> much. It is possible that the players used to calc the seasons chances
play
>> for LHP/RHP or GB/FB, but each was averaged with at least four starters.
>> Hard hit balls, tough chances etc tend to even out. I'd rather have the
raw
>> data (plays in each zone for every player), but I takes what I gots.
>> Nonetheless, the plays I do have are certainly outs. At worst, ZR
>> undervalues defense (depending on pop-ups and line drives, and DPs).
>>
>Do DPs still contaminate the ZR opportunities?
I don't know if "contaminate" is the word I'd use. But, yes, they are
included.
>
>> With these season total of plays, I multiply by the fielder's ZR. That's
>> how many plays he would have made. I think this converts to runs easily,
>> and almost obviously: any play not made results in a baserunner.
>>
>> Each out has a value: if it takes away a single, that's worth 0.47 (hit)
+
>> 0.27 (out). Each subsequent base is worth 0.31 runs (per Linear Wts)
>>
>OK, I follow this so far...
>
>> SS/2B: ((plays made-2)*0.74)+(2*1.05)
>
>The doubles are fixed for ss/2b? Or is there a pm in there?
It is presently a fixed number. It is fixed with just masking tape, so we
can take it out with better information. I'd like a pm, but I don't have
enough DA or ZR raw data.
>
>> 1B/3B: (0.8*pm*0.74)+(0.18*pm*1.05)+(0.02*pm*1.36)
>
>Per DA data, giving up multiple bases at first and third is not
>a function of DA -- the "guarding the line" effect.
I'm not sure what you mean here.
>Lacking specific
>data about doubles and triples through first allowed, though, it's
>a good approximation.
I struggled with these figures. Finally I worked out that these are
basically the distribution of hits not homers.
>
>> OF: [(0.76*pm*0.74)+(0.213*pm*1.05)+(0.026*pm*1.36)+assists*0.62]*pitf
>>
>> pitching staff factor is calculated by: SLG+ = (total bases-(tb from
>> hr))/(ab-K-hr)
>> then set as a percentage of league average (less that team's SLG+)
>>
>> This increases the value of plays made by OF who play in screwy parks or
>> behind bad pitchers. I am pretty sure it removes park effects mostly.
>>
>The doubles/triples allowed versus DA is a big variable in the DA reports.
>Again, lacking the specific 2b/3b allowed data this is a reasonable
>estimate, but it introduces a huge possible area of error. FWIW, the only
>outfielder in the DA reports who was good enough at preventing
doubles/triples
>to negate his poor range was Raul Mondesi.
>
>I don't think assists should be here, at least without folding in
>extra bases taken from the "holding runner" section of the scoreboard.
>The runners taking extra bases on the fielder have to be considered
>along with the assists he gets nailing some of them.
Well, every assist is definitely an out, and a baserunner removed. They
definitely have to count. The holding runner, to me, is far too affected by
the hit itself. I'll have to think about it some more. When I look at the
Scoreboard leaders from 1996 (1997 SB), the RF list is headed by 6 turf
fielders, Sosa and Mondesi. That certainly screams park effect to me. I'm
not sure how it will work.
>
>I'm a bit leery of the pitching factor. SLG+ is a function of fielding
>as well as pitching. If you have an outfield of poor-range outfielders
>who are positioned too shallowly, the resulting bad performance of the
>pitchers will make the lot of poor-range outfielders look average.
That's true. But it's hard to tell, which is which. I'll check the pitf
and see how it matches up with HR allowed.
>
>Also, I would bet groundball pitchers have a SLG+ substantially superior
>to that of flyball pitchers. A groundball staff (especially an
>excellent groundball staff -- Atlanta?) should have a way low SLG+, and
>a resulting degradation of their outfielder's performance, through no
>fault of their own. What we need is real park effects....
I posted the NL pitf earlier, and the pitf will be biased against the good
staff. But I think the Braves' pitchers are good *independent* of who is
behind them. Here they are again:
Atl 0.928
ChN 1.013
Cin 1.001
Col 1.099
Fla 0.988
Hou 0.953
LA 0.938
Mon 0.964
NYM 0.969
PhN 1.095
Pit 1.044
StL 0.977
SD 1.046
SF 0.986
Ana 0.975
Bal 0.953
Bos 1.036
ChA 0.988
Cle 1.007
Det 0.965
KC 0.965
Mil 0.951
Min 1.035
NYY 0.984
Oak 1.106
Sea 1.021
Tex 1.027
Tor 0.991
>
>I'm of the unsupported opinion that when you take out walks, strikeouts,
>homers, flyball/groundball tendencies, and the running game, that fielding
>is much more important for determining safe/out than pitching. I
>believe strongly in park effects, but I'm unconvinced that there really
>is a large "bad pitcher" effect on DA/ZR-like measures.
Well, I am open to suggestions on how to "cipher" it. It was tough working
that out (and convincing myself it made enough sense to fly here). My
database will allow me to manipulate the numbers considerably, without too
much work.
>
>> I compare the players to one another, rather than replacement value. I
do
>> this mostly because I want to know who is better, not who needs to be
>> replaced, and partly because establishing defense for a replacement
player
>> is damned near impossible. My experience with minor league official
scorers
>> is that they suck. Not just in assigning errors, but even paying
attention
>> to the freakin' game. And, yes, I would do a better job.
>>
>I think average is the correct baseline defensively. Although there may
>be a rookie effect on fielding (see the DRs of most shortstops, and guys
>like Griffey), a "replacement fielder" includes guys like Rafael Belliard
>too.
Good point.
Thanks for the input.
Chris Dial
Dale J. Stephenson
(Ron Johnson) wrote:
> In article <dsteph-0309...@ppp-asok06--070.sirius.net>,
> Dale J. Stephenson <dst...@sirius.com> wrote:
> >In article <6s9sco$bp4$1...@supernews.com>,
> >"Chris Dial" <acd...@intrex.com> wrote:
> >
> >I'm a bit leery of the pitching factor. SLG+ is a function of fielding
> >as well as pitching. If you have an outfield of poor-range outfielders
> >who are positioned too shallowly, the resulting bad performance of the
> >pitchers will make the lot of poor-range outfielders look average.
> >
> >Also, I would bet groundball pitchers have a SLG+ substantially superior
> >to that of flyball pitchers.
>
> Something more to check. I've started a study on pitching/defense
> retationship. Turned out to be way more work than I thought.
>
> I can tell you that there is a tiny correlation between the GB/FB
> out rate of the pitcher and the rate that the defense converts
> opportunities into outs. Or to put it in English, groundball pitchers
> tend to make the defense look better. But the effect is minor as best I
> can tell.
>
In balls that were tracked by DA, for instance, the DAs of infielders
are significantly higher than that of outfielders. IOW, even if
defense is *completely* independent of pitching effects, a groundball
pitcher will have a better DA against than a flyball pitcher, because
the people who are handling his chances are better at converting them
into outs. (Or more accurately, it's easier to handle a grounder to
first than a flyball to right.) It's not quite that simple, of course.
DA reports didn't have *infield* flies logged, and those should be the
higher percentage plays of all.
It sounds like you're working with actual data, rather than just
correlating defensive efficiency with GB/FB. The latter says nothing
about affecting the chances at individual positions, just distributing
the chances among higher percentage positions. For the same reason, I'd
guess the defensive efficiency is better among righthanded pitchers (face
more lefties hitting the ball to first/second) than lefthanded pitchers
(face more righties hitting the ball to short/third).
> >A groundball staff (especially an excellent groundball staff -- Atlanta?)
> >should have a way low SLG+, and a resulting degradation of their
> >outfielder's performance, through no fault of their own. What we need
> >is real park effects....
>
> I'm with you here.
> >
> >I'm of the unsupported opinion that when you take out walks, strikeouts,
> >homers, flyball/groundball tendencies, and the running game, that fielding
> >is much more important for determining safe/out than pitching. I
> >believe strongly in park effects, but I'm unconvinced that there really
> >is a large "bad pitcher" effect on DA/ZR-like measures.
>
> That's not what I've found. (cautionary note. I'm digging up more data)
> I've found wildly differing rates for pitchers on the same staffs.
> I don't have error data though so I'm in the process of working out
> estimated errors - (raj you've muttered something about this. Details?)
> maybe this will smooth things out.
>
Is this defensive efficiency, or are you working with real ZR/DA-at-position
behind a pitcher data? I expect that pitchers distribute the ball
differently between popups, infield lineouts, groundball to X, and flyball
to Y. All of those are converted at somewhat different rates. What I
haven't seen strong evidence for is that ZR/DA at a position is strongly
influenced by the pitching staff -- but I've lacked a good way to test
this, so if you've got a way, I'm *very* interested in the results.
Anecdotal points:
DA for infielders showed a fairly strong correlation for infielders who
changed teams. Not true, IIRC, for outfielders. DA subjectively seems
to be more consistent than starting pitching (which, granted, is not
saying a great deal).
In 1990-1991, the pitching staff was fairly stable, but the infield
had a dramatic improvement at every position (in terms of personnel's
past DA). The pitching staff improved in similarly dramatic fashion,
excepting Leibrandt and flyballer John Smoltz. Holdover Jeff Blauser,
on a *rate* basis, was unchanged in DR (he did dramatically improve
in later years).
If you take away the low walk rate, the good strikeout rate, and the
few homers (plus the many chances he handles himself), plus the
normally higher DA (and lower SLG) of a high gb ratio, there isn't
a lot of Greg Maddux's remarkable years to explain with "good defense"
and/or "easy chances".
> Right now what I have is that there's no meaningful correlation between the
> team a pitcher pitches for and the rate that balls put into play are
> converted into outs. Surprised me.
>
> --
> RNJ
Dale J. Stephenson
> Dale J. Stephenson wrote in message ...
> >In article <6s9sco$bp4$1...@supernews.com>, "Chris Dial" <acd...@intrex.com>
> wrote:
[defensive chances snipped]
> >What is considered a "play" here? ZR opportunities, or an estimate
> >of chances started? I'm guessing ZR opportunities, since otherwise using
> >a ZR multiple doesn't work. My scoreboards are packed away right now,
> >but IIRC the lists in back just have the league average, not the total
> >opportunities and outs. Am I remembering incorrectly?
>
>
> It is ZR opportunities. I don't have all the Scoreboards, so I'm not sure
> how they are listed. STATS seems to change that format yearly. I got my #
> of plays by taking all players at a given position that played 130+
> defensive games (Innings/8.75) and averaging their individual chances per
> game. I then multiplied by 162 games.
>
OK, this seems good. I found a single scoreboard that hadn't been
packed away (the 1990 scoreboard), and it doesn't even list the averages
of the leagues. Pfft. But it does have chances and innings for individual
players, so your procedure should give a very reasonable estimate.
I also notice that P > Outs for outfielders. How does an outfielder
get a putout without it being considered a ZR chance? (Or does this
change for later books -- I know there's been some changes with
the out-of-zone plays).
> >
> >> For infielders, everything tends to even out. Since I normalize the
> chances,
> >> lefty-righty staffs and groundball/flyball staffs don't affect the data
> too
> >> much. It is possible that the players used to calc the seasons chances
> play
> >> for LHP/RHP or GB/FB, but each was averaged with at least four starters.
> >> Hard hit balls, tough chances etc tend to even out. I'd rather have the
> raw
> >> data (plays in each zone for every player), but I takes what I gots.
> >> Nonetheless, the plays I do have are certainly outs. At worst, ZR
> >> undervalues defense (depending on pop-ups and line drives, and DPs).
> >>
> >Do DPs still contaminate the ZR opportunities?
>
>
> I don't know if "contaminate" is the word I'd use. But, yes, they are
> included.
>
I'm not in favor of eliminating DP evaluation, but I *do* think they
should be considered separately from straight ZR/DA type out conversion.
You need the DP-ops in order to make it fair. OTOH, STATS also gives
DP conversion for pivot men, which I'd like to roll in as well. I'm
not sure how to split the DP credit between initiator and pivot man.
Any ideas?
> >
> >> With these season total of plays, I multiply by the fielder's ZR. That's
> >> how many plays he would have made. I think this converts to runs easily,
> >> and almost obviously: any play not made results in a baserunner.
> >>
> >> Each out has a value: if it takes away a single, that's worth 0.47 (hit)
> +
> >> 0.27 (out). Each subsequent base is worth 0.31 runs (per Linear Wts)
> >>
> >OK, I follow this so far...
> >
> >> SS/2B: ((plays made-2)*0.74)+(2*1.05)
> >
> >The doubles are fixed for ss/2b? Or is there a pm in there?
>
>
> It is presently a fixed number. It is fixed with just masking tape, so we
> can take it out with better information. I'd like a pm, but I don't have
> enough DA or ZR raw data.
>
I have the DA reports for 88-96. Here's the league number of doubles/triples
for 93-95 in the NL.
2b@2b 3b@2b 2b@ss 3b@ss
93 50 6 38 2
94 41 6 42 3
95 42 8 55 1
I go ahead and adjust 2b/3b allowed by player, but it doesn't amount
to much. The career NEB (Net Extra Base prevented) leaders at
second and short are Ryne Sandberg (+9 in 7 seasons) and Royce Clayton
(+9 in 5 seasons). As you know that's in the order of half a run per
season -- only worth counting at all because the data happens to be
there.
There's more impact to make on the negative side, of course. Kevin
Elster managed -13 in 5 seasons, while Chuck Knoblauch managed
-21 (!) in 6 seasons. Of course, it's *still* not a lot of runs,
and when a second baseman manages to give up that many doubles you've
got to wonder if he's been penalized for Kirby Puckett playing too
deep.
Of course, you don't penalize anyone, since it's a constant. That's
not true at first/third, where the doubles/triples are more common
*and* the fielder is likely to have more influence on the result of
the play.
> >
> >> 1B/3B: (0.8*pm*0.74)+(0.18*pm*1.05)+(0.02*pm*1.36)
> >
> >Per DA data, giving up multiple bases at first and third is not
> >a function of DA -- the "guarding the line" effect.
>
> I'm not sure what you mean here.
>
It's pretty simple. A first baseman can stand very close to the first
base bag, giving up a hole in part of his area for hits to go through
(sometimes he *has* to do this, due to a runner being held). But this
way he can prevent the ball from being hit right down the line, which
is more likely to be a double or triple by the time the outfielder gets
to it. You've seen the prevent-the-double defense employed in ball
games -- first and third at the lines, outfielders playing deep.
A fielder can naturally play closer to the line, preventing extra base
hits at the possible expense of a lower DA/ZR. Your formula assumes
that the extra bases allowed are directly proportional to the hits
allowed. I don't think that's true:
A snippet from the career DR leaders:
->FIRST BASE
*Range*
Most Hits Stolen 88-96:
132) Mark Grace
77) John Olerud
67) Mark McGwire
63) Rafael Palmiero
52) Tino Martinez
Least Hits Stolen 88-96:
-67) Pedro Guerrero
-53) J.T. Snow
-47) Frank Thomas
-44) Fred McGriff
-38) Paul Sorrento
Single Season Best:
36) Mark Grace (1992)
Single Season Worst:
-40) Fred McGriff (1992)
*Preventing Extra Bases*
Most Extra Bases Saved 88-96:
30) Will Clark
Jeff Bagwell
17) Glenn Davis
16) John Olerud
15) Mo Vaughn
Eric Karros
Least Extra Bases Saved 88-96:
-31) Tino Martinez
-27) Andres Galarraga
-22) Frank Thomas
Rico Brogna
-21) David Segui
Single Season Best:
13) Jeff Bagwell (1996)
Single Season Worst:
-14) Jeff Bagwell (1991)
Andres Galarraga (1994)
-->THIRD BASE
*Range*
Most Stolen Hits 88-96:
154) Terry Pendleton
89) Robin Ventura
82) Matt Williams
73) Tim Wallach
69) Kelly Gruber
Least Stolen Hits 88-96:
-75) Dean Palmer
-56) Todd Zeile
-46) Gary Sheffield
-43) Sean Berry
-41) Howard Johnson
Best Season:
45) Chris Sabo (1988)
Worst Season:
-39) Dean Palmer (1992)
*Knocking It Down*
Most Extra Bases Prevented 88-96:
50) Ken Caminiti
49) Robin Ventura
38) Matt Williams
27) Tim Wallach
22) Wade Boggs
Least Extra Bases Prevented 88-96:
-25) Scott Leius
-24) Dean Palmer
-22) Craig Worthington
-21) Luis Salazar
-19) Howard Johnson
Jim Presley
Best Season:
14) Tim Naehring (1996)
Worst Season:
-14) Scott Leius (1995)
Yes, there's some correlation here. But Mark Grace had 132 NHS while
logging only *2* NEB. Meanwhile, Rico Brogna managed to get -22 NEB
in two seasons, with basically average DA (-3 HS). At third, while
Ventura, Williams and Wallach managed to accumulate plenty of NEB
and NHS, Terry Pendleton had -11 NEB to go with his outstanding 154
NHS. Terry did a lot of things right, but preventing extra base
hits wasn't one of them.
Now, making hits into outs is a lot more important than preventing extra-base
hits, so I don't think your system is fatally flawed at all. But the
skills are independent enough that superimposing an extra-base weight
on ZR proficiency distrubs me.
> >Lacking specific
> >data about doubles and triples through first allowed, though, it's
> >a good approximation.
>
>
> I struggled with these figures. Finally I worked out that these are
> basically the distribution of hits not homers.
>
Sure, on average. But a first and third baseman can have an influence
on what kind of hits get through their area. (Though I'll grant with
the smaller ZR ranges, they probably *aren't* hit through their area.)
> >
> >> OF: [(0.76*pm*0.74)+(0.213*pm*1.05)+(0.026*pm*1.36)+assists*0.62]*pitf
> >>
> >> pitching staff factor is calculated by: SLG+ = (total bases-(tb from
> >> hr))/(ab-K-hr)
> >> then set as a percentage of league average (less that team's SLG+)
> >>
> >> This increases the value of plays made by OF who play in screwy parks or
> >> behind bad pitchers. I am pretty sure it removes park effects mostly.
> >>
> >The doubles/triples allowed versus DA is a big variable in the DA reports.
> >Again, lacking the specific 2b/3b allowed data this is a reasonable
> >estimate, but it introduces a huge possible area of error. FWIW, the only
> >outfielder in the DA reports who was good enough at preventing
> doubles/triples
> >to negate his poor range was Raul Mondesi.
> >
> >I don't think assists should be here, at least without folding in
> >extra bases taken from the "holding runner" section of the scoreboard.
> >The runners taking extra bases on the fielder have to be considered
> >along with the assists he gets nailing some of them.
>
>
> Well, every assist is definitely an out, and a baserunner removed. They
> definitely have to count. The holding runner, to me, is far too affected by
> the hit itself.
Every caught stealing is an out and a baserunner removed -- but without
knowing how many people are running, catcher assists don't do you much
good. Outfield isn't *that* bad, but since STATS kindly presents us
with chances for runners to advance and the percentage that *do* advance,
we should be able to combine this with the assists to see the effect on
the running game -- are all those assists because people are taking lots
of chances, or because of the great arm?
Assists aren't a function of ZR at all, so it really belongs in a separate
line.
> I'll have to think about it some more. When I look at the
> Scoreboard leaders from 1996 (1997 SB), the RF list is headed by 6 turf
> fielders, Sosa and Mondesi. That certainly screams park effect to me. I'm
> not sure how it will work.
>
Interestingly, the 1990 scoreboard also screams park effect, but with
the opposite reasoning: "You'll note from the chart that eight of the nine
leaders [in hold percentage, three from each field] are from American League
teams. This probably isn't because American League outfielders are
superior. More likely, it's because AL parks tend to be smaller than their
National league counterparts. American League parks are also primarily
grass fields, and on grass an outfielder can charge the ball much more
easily. (Seven of the nine leaders played their home games on grass.)
NL outfielders, playing on bigger fields that usually have an artificial
surface, are forced to play deeper. That gives runner a better opportunity
to advance."
At any rate, I'll bet the park effect on hold percentage is absolutely
dwarfed by park effect on ZR/DA. It's a concern, but I don't think it's
enough of one to throw out the data.
BTW, here's the NHS vs NEB of Raul Mondesi:
94 Mondesi,Raul DA 0.530 NHS -20 NEB 7 DR -12.21
95 Mondesi,Raul DA 0.603 NHS 2 NEB 14 DR 5.69
96 Mondesi,Raul DA 0.580 NHS -2 NEB 40 DR 10.70
** Mondesi,Raul DA 0.573 NHS -20 NEB 61 DR 4
He's an extreme case, but the extra base multiplier *really* won't
work on guys like this.
Hey, I love the Braves' rotation too. Suppose for a moment that they're
equally good regardless of who is behind them. Klesko or Andruw Jones --
it doesn't matter. Then there's no point in evaluating outfield defense
at all. It doesn't matter. How much difference does this adjustment
make? How many runs positive would the Atlanta outfield be if it were
comprised of the three players with the *best* outfield ZR at their
position? How many runs negative would the Colorado outfield be if it
were comprised of the three players with the *worst* outfield ZR at
their positions?
Another problem, which I'm sure you're aware of, is asymetric outfields.
A generalized outfield adjustment is never going to do the leftfielder
in Fenway much good.
>
> >
> >I'm of the unsupported opinion that when you take out walks, strikeouts,
> >homers, flyball/groundball tendencies, and the running game, that fielding
> >is much more important for determining safe/out than pitching. I
> >believe strongly in park effects, but I'm unconvinced that there really
> >is a large "bad pitcher" effect on DA/ZR-like measures.
>
>
> Well, I am open to suggestions on how to "cipher" it. It was tough working
> that out (and convincing myself it made enough sense to fly here). My
> database will allow me to manipulate the numbers considerably, without too
> much work.
>
What we need is a few years of complete play-by-play data for every game,
so we can compare DA/ZRs behind different pitchers, look at home/away
conversion, break down conversion rates in different zones, look at infielder
lineout effects, count the popups for pitchers, etc. etc. Sadly, we don't
have it, and can't get it without some thousands to throw at STATS.
Lacking that, we should start with an unadjusted (for park or pitching)
DR, and then fiddle endlessly with things like pitf for the DR/A.
It's a reasonable approach, but I think the pitf effect is unproven
enough that we should have the "raw" DR as well.
David Marc Nieporent
NawrockiT <nawr...@aol.com> claimed:
>Chris Dial wrote:
>: I think some adjustment should be made for 1B and infield throwing errors.
>: The Mariners' ng is often pissin' and moanin' about how Segui saves all
>: these throwing errors (and still Davis has 120). Playing SS for my 30+
>: team, I know what it's like to go from a 6'4" lifetime 1B to a 5'7"
>: short-armed former SS (that plays first because he's the manager). It's
>: like throwing it through the clown's mouth. Sure, my situation is
>: exaggerated, but I know the good 1B saves *tons* of errors. Every scoop is
>: a throwing error, and there are several chances per game per 1B.
>What complicates matters is that these errors are never charged to the
>1B, so you can't even look at his errors to see whether he can catch or
>not. Tino Martinez is terrible at catching throws that bounce into the
>dirt, but of course he's never held responsible for it.
>Tino does have ten errors this year, but he also should get credit for a
>lot of Brosius' 22 errors.
Obviously this is a factor of 1b defense, but it's got to be overstated.
The difference in errors between the throwing infielders on different
teams just isn't that large.
(The only argument I can see is that a bad scooper deters infielders from
even *trying* tough throws.)
--
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
Chris Dial
I believe that this is because of plays made out of zone. This is almost
completely a function of shifted OFs. I would expect there to be a bigger
effect for CF, but I haven't looked.
Many things I do are tweaked based on my personal experience/observation.
DP-ops seem to be pretty random (# per team), and I am not confident in
STATS ability to deliver the proper breakdowns on these numbers. The
seasonal # of DP ops seems fairly constant (225 per AL team in 96, 97) and
the std dev is about 9 chances. How significant that is, when I don't think
there is a correlation between these Ops season to season. I'll have to
check the ZRs of the MI for the teams with higher DP-ops.
{snip 2b/3b stuff}
Thanks for the xbh data. That certainly makes 2 a pretty good starting
point, but it looks like I'll increase it to 3.5 (3.64 in AL and 3.35 in NL)
>Of course, you don't penalize anyone, since it's a constant. That's
>not true at first/third, where the doubles/triples are more common
>*and* the fielder is likely to have more influence on the result of
>the play.
>
>> >
>> >> 1B/3B: (0.8*pm*0.74)+(0.18*pm*1.05)+(0.02*pm*1.36)
>> >
>> >Per DA data, giving up multiple bases at first and third is not
>> >a function of DA -- the "guarding the line" effect.
>>
>> I'm not sure what you mean here.
>>
>It's pretty simple. A first baseman can stand very close to the first
>base bag, giving up a hole in part of his area for hits to go through
>(sometimes he *has* to do this, due to a runner being held). But this
>way he can prevent the ball from being hit right down the line, which
>is more likely to be a double or triple by the time the outfielder gets
>to it. You've seen the prevent-the-double defense employed in ball
>games -- first and third at the lines, outfielders playing deep.
>
>A fielder can naturally play closer to the line, preventing extra base
>hits at the possible expense of a lower DA/ZR. Your formula assumes
>that the extra bases allowed are directly proportional to the hits
>allowed. I don't think that's true:
>
<snip DR leaders>
>
>Yes, there's some correlation here. But Mark Grace had 132 NHS while
>logging only *2* NEB. Meanwhile, Rico Brogna managed to get -22 NEB
>in two seasons, with basically average DA (-3 HS). At third, while
>Ventura, Williams and Wallach managed to accumulate plenty of NEB
>and NHS, Terry Pendleton had -11 NEB to go with his outstanding 154
>NHS. Terry did a lot of things right, but preventing extra base
>hits wasn't one of them.
>
>Now, making hits into outs is a lot more important than preventing
extra-base
>hits, so I don't think your system is fatally flawed at all. But the
>skills are independent enough that superimposing an extra-base weight
>on ZR proficiency distrubs me.
I'm not sure that they are all skills like that. Positioning of infielders
isn't always their decision, especially at the corners. And from watching
him, Pendleton's particular style of play was to be a big "backhander", and
shade the hole, while Ventura, Williams, and Wallach were play close to the
line. That type of positioning is mostly habitual, and for this system I
don't see it as very practical to establish each players particular habit,
esp. when I have to depend on STATS giving up data.
Score board did a "study" on whether or not guarding the line worked (1993?)
and their data showed, IIRC, that guarding the line usually gave up more
singles than it prevented doubles; that is it wasn't a strong strategy.
I'll have to re-read that.
>
Personal observation and experience still leads me to believe CF and LF
assists are rarely based on the fielder's arm. With only 30 OF assists per
team, runners trying to tag up and score are usually the outs, with several
batters trying to stretch singles into to doubles, and then there are guys
getting doubled off.
>
>Assists aren't a function of ZR at all, so it really belongs in a separate
>line.
It sort of is. I guess I should multiply the SLG+ and then add the assists.
I also think the hold% is widely misled, because a fielder will create his
own chances. If a LF (we'll call him Gregg Jefferies) is slow to come in on
a sinking liner, he holds a runner, that thought he should have caught the
damned thing. Jefferies takes a ZR hit, but gains a hold.
>
>> I'll have to think about it some more. When I look at the
>> Scoreboard leaders from 1996 (1997 SB), the RF list is headed by 6 turf
>> fielders, Sosa and Mondesi. That certainly screams park effect to me.
I'm
>> not sure how it will work.
>>
>Interestingly, the 1990 scoreboard also screams park effect, but with
>the opposite reasoning: "You'll note from the chart that eight of the nine
>leaders [in hold percentage, three from each field] are from American
League
>teams. This probably isn't because American League outfielders are
>superior. More likely, it's because AL parks tend to be smaller than their
>National league counterparts. American League parks are also primarily
>grass fields, and on grass an outfielder can charge the ball much more
>easily. (Seven of the nine leaders played their home games on grass.)
>NL outfielders, playing on bigger fields that usually have an artificial
>surface, are forced to play deeper. That gives runner a better opportunity
>to advance."
Well, that's bizarre. Nonetheless, I don't think the hold stat has been
around and evaluated enough to include yet.
>
>At any rate, I'll bet the park effect on hold percentage is absolutely
>dwarfed by park effect on ZR/DA. It's a concern, but I don't think it's
>enough of one to throw out the data.
>
>BTW, here's the NHS vs NEB of Raul Mondesi:
>
>94 Mondesi,Raul DA 0.530 NHS -20 NEB 7 DR -12.21
>95 Mondesi,Raul DA 0.603 NHS 2 NEB 14 DR 5.69
>96 Mondesi,Raul DA 0.580 NHS -2 NEB 40 DR 10.70
>** Mondesi,Raul DA 0.573 NHS -20 NEB 61 DR 4
>
>He's an extreme case, but the extra base multiplier *really* won't
>work on guys like this.
It may not help him as much, but with an expected error of 5 runs or so, can
it appreciably affect his status?
>
>> >
>> >I'm a bit leery of the pitching factor. SLG+ is a function of fielding
>> >as well as pitching. If you have an outfield of poor-range outfielders
>> >who are positioned too shallowly, the resulting bad performance of the
>> >pitchers will make the lot of poor-range outfielders look average.
>>
>> That's true. But it's hard to tell, which is which. I'll check the pitf
>> and see how it matches up with HR allowed.
I posted these elsewhere.
>>
>> >
>> >Also, I would bet groundball pitchers have a SLG+ substantially superior
>> >to that of flyball pitchers. A groundball staff (especially an
>> >excellent groundball staff -- Atlanta?) should have a way low SLG+, and
>> >a resulting degradation of their outfielder's performance, through no
>> >fault of their own. What we need is real park effects....
To summarize, just the opposite occurred. >>
>Hey, I love the Braves' rotation too. Suppose for a moment that they're
>equally good regardless of who is behind them. Klesko or Andruw Jones --
>it doesn't matter. Then there's no point in evaluating outfield defense
>at all. It doesn't matter. How much difference does this adjustment
>make? How many runs positive would the Atlanta outfield be if it were
>comprised of the three players with the *best* outfield ZR at their
>position? How many runs negative would the Colorado outfield be if it
>were comprised of the three players with the *worst* outfield ZR at
>their positions?
I like this idea. I'll run the numbers.
>
>Another problem, which I'm sure you're aware of, is asymetric outfields.
>A generalized outfield adjustment is never going to do the leftfielder
>in Fenway much good.
A "generalized" OF adjustment? If the Bosox staff gives up 2b off the wall,
it will show up in the SLG+.
>>
>> >
>> >I'm of the unsupported opinion that when you take out walks, strikeouts,
>> >homers, flyball/groundball tendencies, and the running game, that
fielding
>> >is much more important for determining safe/out than pitching. I
>> >believe strongly in park effects, but I'm unconvinced that there really
>> >is a large "bad pitcher" effect on DA/ZR-like measures.
>>
>>
>> Well, I am open to suggestions on how to "cipher" it. It was tough
working
>> that out (and convincing myself it made enough sense to fly here). My
>> database will allow me to manipulate the numbers considerably, without
too
>> much work.
>>
>What we need is a few years of complete play-by-play data for every game,
>so we can compare DA/ZRs behind different pitchers, look at home/away
>conversion, break down conversion rates in different zones, look at
infielder
>lineout effects, count the popups for pitchers, etc. etc. Sadly, we don't
>have it, and can't get it without some thousands to throw at STATS.
>Lacking that, we should start with an unadjusted (for park or pitching)
>DR, and then fiddle endlessly with things like pitf for the DR/A.
>
>It's a reasonable approach, but I think the pitf effect is unproven
>enough that we should have the "raw" DR as well.
Unproven enough? Hell, I just made it up. ;-). I'll work on some of these
things and get back with it.
Chris Dial