Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Hall of Merit > Discussion
Hall of Merit
— A Look at Baseball's All-Time Best

Monday, June 25, 2007

Kirk Gibson

Eligible in 2001.

John (You Can Call Me Grandma) Murphy Posted: June 25, 2007 at 05:26 PM | 62 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. John (You Can Call Me Grandma) Murphy Posted: June 25, 2007 at 05:30 PM (#2416720)
Not really a great player, but those postseason numbers are pretty impressive.
   2. OCF Posted: June 25, 2007 at 07:05 PM (#2416839)
So what do people think about his 1988 MVP? Here's the voting:

Gibson 272 with 13 1st place
Strawberry 236 with 7 1st place
McReynolds 162 with 4 1st place (McReynolds ahead of Strawberry?)
Van Slyke 160
W. Clark 135
Hershiser 111
Galarraga 105
   3. vortex of dissipation Posted: June 25, 2007 at 07:10 PM (#2416842)
So, was Gibson or Tim Salmon the best post WW2 player never to play in an All-Star Game?
   4. caprules Posted: June 25, 2007 at 07:12 PM (#2416844)
Are there any Gibson Pirates jerseys out there? What was the story there? Was there an injury in the outfield that spring for the Pirates, or was he picked up as a platoon player?
   5. DCW3 Posted: June 25, 2007 at 07:17 PM (#2416849)
So what do people think about his 1988 MVP? Here's the voting:

Clark seems like the best choice to me. Led the NL in VORP (even though he was only 11th in the majors) and RCAP, and played fine defense.
   6. Dr. Chaleeko Posted: June 25, 2007 at 07:38 PM (#2416865)
He threw the most valuable tantrum of 1988 when he got ripped about Orosco lining his hat with eyeblack during spring training. I remember reading about the incident time and again during and after the season, about how it "set the tone" for the team, blah blah blah.

Anyway, another tough MVP call in the NL
OPSleaders 1988
NAME       OPS
+  PA  GIDP SB/CS
--------------------------------
Strawberry 165  640   6   29/14
Clark      160  689   9    9
/1  
Gibson     149  632   8   31
/4  
Galarragga 149  661  12   13
/4   
Bonds      147  614   3   17
/11
Van Slyke  143  659   8   30
/9
Daniels    143  589  11   27
/6
McReynolds 142  600   6   21
/0
Bonilla    140  681   4    3
/5
Davis      140  543  11   35
/


Oh, and there's these guys:


Hershiser 23-8, 267 IP, 148 ERA+
Cone 20-3, 231 IP 145, ERA+

You can get rid Davis since he's redundant with Van Slyke who has more PT. Bonilla is an easy one to whack too. McReynolds fades, despite the 21/0, since he's got 20 fewer OPS+ points in 50-80 fewer PAs. Daniels, not enough PT.

Now it gets tough.

Van Slyke and Bonds is a toughie. Slick was a very good defender, so was Bonds. Andy's got the better steals percentage but a few more dPs. But he's got 40 PAs more. Van Slyke 104 RC or 6.25 per game. Bonds 97 RC but 6.51 per game. Slick 12 FRAA, Bonds 2 FRAA. It's so close.

But I think either of them knocks out Galarraga who has more DPs than both and who plays a less important position and had 0 FRAA.

Now Gibson. Gibson played LF, 6 FRAA, and everything else is as good or better than Bonds or Van Slyke, except 20 PAs he loses to AVS. I'm comfortable with him at this point as the leader in the clubhouse, so to speak.

So Gibson versus Clark. Gibson ceeds 57 fewer PAs with 11 fewer OPS+ points. Clark's DP rate is similar, and his SB rate is actually a little better than Gibson's. Gibson squeezes out a couple more runs on those steals. Clark is also a 9 FRAA, at a slightly less defensive position. When BP totals it up, they get 11.6 WARP1 for Clark and 10 for Gibson. That doesn't seem nutty to me.

So if it ain't Gibson, who is it?

Clark v. Strawberry: 5 OPS+ points versus 49 PAs. Straw has a lower dP rate, but his SB% is yucky. A big negative is that BP sees him as a -9 FRAA. Ouch. They've got him below Gibson at 9.2 WARP1. RC/27 is 6.91 for Straw versus 7.05 for Clark. The little stuff just keeps adding up for The Thrill.

But is it Clark? What about the pitchers?

Cone is clearly not quite Hershiser, a paler version. So it's Orel and Will. hershier's DERA was 3.17, or you could say a 142 DERA+. BP sees him as creating 10 wins of value, the same as Gibson, not enough to topple Clark. Or look at it this way, Clark's 7.05 R/27 is a 181 R/27+ versus the league's 3.88 R/G, well outpacing Hershiser's DERA+ (and his 1 FRAA and decent pitcher batting).

On the whole it seems to me that when you look closely, Kirk Gibson's tantrum impressed writers who didn't see a clear winner in the MVP sweepstakes and saw a good-but-not-eyepopping line of .290/27/76/31-4 as unrepresentative of his total contribution. Gibson (and Orel) were probably the second-best players in the NL, but Clark finished a distant fifth with Strawberry a close runner up (272 to 236).
   7. DCW3 Posted: June 25, 2007 at 08:55 PM (#2416935)
Led the NL in VORP (even though he was only 11th in the majors)

Oops, forgot to check pitchers--Hershiser was actually the NL VORP leader.
   8. David Concepcion de la Desviacion Estandar (Dan R) Posted: June 25, 2007 at 11:00 PM (#2417062)
I have Gibson deserving his MVP by a very comfortable margin (not counting pitchers). Here's how I see the top 10:

Glossary

BWAA: Batting wins above average
BRWAA: Baserunning wins above average (including SB/CS and James Click's non-SB baserunning runs)
DPWAA: Double play avoidance wins above average
FWAA: Fielding wins above average
Rep: Wins above average of a replacement player at the same position in the same playing time
WARP: Wins above replacement (BWAA + BRWAA + DPWAA + FWAA + Rep)

Player    Pos BWAA BRWAA DPWAA FWAA  Rep WARP
Gibson     LF  4.8   1.0   0.3  1.2 
-0.8  8.0
Larkin     SS  2.2   0.9   0.1  0.6 
-3.4  7.2
Clark      1B  6.1   0.5   0.1  0.2 
-0.1  7.1
Smith      SS  1.0   0.7   0.0  1.6 
-3.4  6.8
Butler     CF  4.0   0.4   0.2  0.4 
-1.3  6.5
Van Slyke  CF  4.5   0.3   0.3 
-0.1 -1.3  6.3
Strawberry RF  6.0  
-0.2   0.6 -1.3 -0.8  5.9
Daniels    LF  3.9   0.4   0.0  0.7 
-0.8  5.7
Davis      CF  3.2   0.6  
-0.3  0.8 -1.3  5.4
Bonds      LF  4.2   0.0   0.2  0.1 
-0.8  5.3 


Analysis

Gibson was only the third-best hitter in the league, but he brought everything else to the table that year. According to Chris Dial's Zone Rating-based data, he was an outstanding fielder in 1988 (Fielding WS and BP FRAA basically agree, albeit to a lesser extent). Gibson was also an outstanding baserunner above and beyond his 31/4 SB/CS, adding 5.1 non-SB baserunning runs according to James Click's EqBR stat, and did a good job of avoiding double plays. Add it all up, and he's an easy choice--8 WARP is about standard for a league MVP.

Clark was good on the basepaths, but not as good as Gibson. FRAA likes his fielding, but FRAA tends to go nuts about 1B defense, while Fielding WS and Zone Rating both see the Thrill's D as basically average. Gibson's defensive advantage over Clark (he played a more demanding position, and he fielded it better than Clark fielded his) and baserunning superiority easily outweigh the hitting difference in favor of Clark.

Strawberry was an abysmal fielder according to Zone Rating, and a poor one according to BP FRAA. And he got caught stealing a bunch too. He's well behind.

If you believe Nate Silver's research on replacement levels, then the Wizard and Larkin round out the top four. Ozzie had one of his better offensive seasons, wrought havoc on the basepaths, and was his usual self with the glove, while Larkin gave the Reds corner-quality offense and plus fielding from SS.

Brett Butler's 133 OPS+ doesn't make Dr. Chaleeko's chart, but it was virtually all OBP and was paired with terrific baserunning and plus defense from CF. He was the fifth-best player in the league.
   9. Chris Cobb Posted: June 26, 2007 at 12:47 AM (#2417294)
Posts 7 and 9 are why the "best player of the year" project will be interesting . . .
   10. Los Angeles Waterloo of Black Hawk Posted: June 26, 2007 at 01:02 AM (#2417328)
I'm far from unbiased, but I think Salmon has the edge over Gibby. Higher OPS+, a few more points of EqA, a bit more PA, a few more WARP1 and WARP3. Not sure about Win Shares. Salmon was an elite defensive RF at his peak, so I don't really know what the defensive difference is between them, if any. Tim was also a smart baserunner, it always seemed, so I don't know if Gibson gains a lot on that count.
   11. Los Angeles Waterloo of Black Hawk Posted: June 26, 2007 at 01:06 AM (#2417351)
Kirk Gibson's tantrum impressed writers who didn't see a clear winner in the MVP sweepstakes and saw a good-but-not-eyepopping line of .290/27/76/31-4 as unrepresentative of his total contribution

Using the "neutralize" tool at B-R, his 1988 would have carried the same value as a .323/29/98/36-? (with 136 runs scored) if he had done it in 2006 for the same team. The writers then obviously didn't think in those terms, but the context of park and era kind of trick the modern eye as to the value of the line he did put up.
   12. Misirlou was a Buddhist prodigy Posted: June 26, 2007 at 01:54 AM (#2417463)
Using the "neutralize" tool at B-R, his 1988 would have carried the same value as a .323/29/98/36-? (with 136 runs scored) if he had done it in 2006 for the same team.


But last year's NL MVP put up a ine of .313 58 149. the guy who everyone agreed got jobbed did .339 41 137. the third place guy did .313 45 136. The 4th place guy went .275 41 116 41/4 SB/CS, with stellar defense in CF. The 6th place guy hit 46 HR and drove in 95 as a leadoff hiter.

Gibby's translated stats put him in 6th place in BA, and out of the top 10 in HR and RBI.
   13. David Concepcion de la Desviacion Estandar (Dan R) Posted: June 26, 2007 at 01:54 AM (#2417464)
Salmon should have been the AL MVP in 1995. He had a massive season. Here's the same chart for the 1995 AL, straight-line-adjusted for season length and with an extra column to show WARP per 700 PA...the lower replacement levels are due to the DH adjustment. Also, I screwed up the Rep column in the previous chart, I put Rep per 162 games rather than Rep in the players' actual playing time, not that anyone noticed.

Player     Pos BWAA BRWAA DPWAA FWAA  Rep WARP WARP/700
Salmon      RF  5.7   0.5   0.2  0.8 
-1.8  8.9      8.7
Belle       LF  6.3   0.2  
-0.5  0.9 -1.8  8.5      8.4
Valentin    SS  3.4   0.3   0.3 
-0.3 -3.9  7.4      7.5
Thomas      1B  7.1   0.0   0.3 
-0.7 -0.6  7.1      6.9
Thome       3B  4.6   0.1   0.4  0.2 
-1.6  6.8      7.5
E
Martínez DH  7.0  -0.3   0.2  0.0  0.0  6.3      6.6
McGwire     1B  5.9   0.2   0.0 
-0.4 -0.7  6.2      9.2
Edmonds     CF  2.3   0.1   0.4  1.2 
-2.3  6.1      6.2
B
Williams CF  2.9   0.1   0.0  0.9 -2.4  5.7      5.9
Knoblauch   2B  3.7  
-0.1  -0.2  0.0 -2.4  5.3      5.6 
   14. David Concepcion de la Desviacion Estandar (Dan R) Posted: June 26, 2007 at 02:44 AM (#2417657)
argh, that blatantly didn't align, let's try again...

Player     Pos BWAA BRWAA DPWAA FWAA  Rep WARP WARP/700
Salmon      RF  5.7   0.5   0.2  0.8 
-1.8  8.9      8.7
Belle       LF  6.3   0.2  
-0.5  0.9 -1.8  8.5      8.4
Valentin    SS  3.4   0.3   0.3 
-0.3 -3.9  7.4      7.5
Thomas      1B  7.1   0.0   0.3 
-0.7 -0.6  7.1      6.9
Thome       3B  4.6   0.1   0.4  0.2 
-1.6  6.8      7.5
E
Martínez DH  7.0  -0.3   0.2  0.0  0.0  6.8      6.6
McGwire     1B  5.9   0.2   0.0 
-0.4 -0.7  6.3      9.2
Edmonds     CF  2.3   0.1   0.4  1.2 
-2.3  6.2      6.2
B
Williams CF  2.9   0.1   0.0  0.9 -2.4  6.1      5.9
Knoblauch   2B  3.7  
-0.1  -0.2  0.0 -2.4  5.7      5.6 
   15. Guapo Posted: June 26, 2007 at 03:03 AM (#2417720)
If I'm remembering correctly, Gibson was actually selected to a couple of All Star games, in 1985 by Sparky and in 1988 by Whitey Herzog. He chose to go hunting for three days instead.
   16. J. Lowenstein Apathy Club Posted: June 26, 2007 at 03:55 AM (#2417922)
Thesis : Kirk Gibson did less, with more talent, than any other star player of his generation. Including Doc and Straw.
   17. Dr. Chaleeko Posted: June 26, 2007 at 01:21 PM (#2418212)
Thesis : Kirk Gibson did less, with more talent, than any other star player of his generation. Including Doc and Straw.

Greg Jefferies fans object!!!!

Dan, your charts don't appear to include a column for Leadership Above Replacement (LAR), so I've adjusted them for you:
Player    Pos BWAA BRWAA DPWAA FWAA   LAR   Rep  WARP
Gibson     LF  4.8   1.0   0.3  1.2  10.0  
-0.8  18.0
Butler     CF  4.0   0.4   0.2  0.4   1.0  
-1.3   7.5
Van Slyke  CF  4.5   0.3   0.3 
-0.1   1.0  -1.3   7.3
Larkin     SS  2.2   0.9   0.1  0.6   0.0  
-3.4   7.2
Clark      1B  6.1   0.5   0.1  0.2   0.0  
-0.1   7.1
Smith      SS  1.0   0.7   0.0  1.6   0.0  
-3.4   6.8
Daniels    LF  3.9   0.4   0.0  0.7   0.0  
-0.8   5.7
Davis      CF  3.2   0.6  
-0.3  0.8   0.0  -1.3   5.4
Strawberry RF  6.0  
-0.2   0.6 -1.3  -5.0  -0.8   0.9
Bonds      LF  4.2   0.0   0.2  0.1 
-20.0  -0.8 -15.3 


That's a little more like it. Let's break Gibson's LAR down just a bit to see why it makes such a difference:
+ 5.5 above Rep for football mentality (he played RB at Michigan, don't you know!)
+ 1.3 above Rep for unshavedness
+ 1.2 above Rep for clutchiness
+ 1.1 above Rep for eyeblack tantrums
+ 0.6 above Rep for very broad shoulders (good for team carrying) and determined look on face
+ 0.3 above Rep for not throwing Mike Davis off bridge

I'm very surprised by how little he gets for the eyeblack tantrum. I thought that was THE big factor for him, but I've got to agree with the voters, it's football all the way---it's the overarching trait that bundles the others together into leadershippy goodness.
   18. JPWF13 Posted: June 26, 2007 at 02:16 PM (#2418262)
McReynolds 162 with 4 1st place (McReynolds ahead of Strawberry?)


Keith Hernandez told every reporter who came into the Mets clubhouse that McReynolds was better than Straw and was the team's "true" MVP- he did not do that out of any friendship/repect fro McReynolds either.

What I remember most about Gibson (aside from the 1988 post season) is that after being relentlessly hyped as the next Mickey Mantle, he finally turned in a good full season in 1984, and then told every reporter who'd listen that he'd given away hundreds of at bats- no one on base, 3+ run lead, he didn't care, he only hit when it counted- why would he say such thing? He was trying to say that he wasn't a "number hanger" that he was really BETTER than his .282 average, taht he could win a batting title if he wanted to, but he only put forth effort when it mattered to his team, yada yada yada...

I was pretty much eternally soured on him thereafter, and as a Met fan the 1988 MVP vote did not make happy.

However, according to BBREF he DID hit somewhat better with men on than with bases empty
and he did have a slightly higher OPS in tie games and 1 run games than in laughers (though his worst split was in late and close situations :-)

But then again Strawberry hit MUCH better with men on or RISP than he did with bases empty
(but Straw's worst split was also late and close - a common pattern I think- batters tend to face other teams better relievers in such situations- even clutch God Big Papi drops off in late and close situations)
   19. TomH Posted: June 26, 2007 at 02:35 PM (#2418272)
Davdi Ortiz splits, 2004-2006, from espn.com
situation......... AB HR RBI BB AVG OBA SLG . OPS
Close and Late 235 27 80 .40 .328 .427 .749 1.176

80 RBI in 235 AB. And that's not one of thsoe "with runners on" stats, either.

And of course that doesn't count the postseason, when in 2004, he became the only player in history with 3 game-ending postseason hits.

Since Papi joined the Red Sox in 2003 thru '06, he had 15 walk-off hits; the rest of the team combined had 19.
   20. DavidFoss Posted: June 26, 2007 at 03:40 PM (#2418340)
Win Shares gives the NL MVP to Will Clark (37 WS) in 1988 by a comfortable margin, but it has Gibson tied with Bonilla & McReynolds for second with 31 WS, Strawberry in 5th with 30.

Its one of those MVP votes that sticks out because the numbers don't look gaudy, but on closer look it doesn't look so bad. Not a lot of standout years that season.
   21. Daryn Posted: June 26, 2007 at 04:08 PM (#2418367)
If ever there was an MVP based on intangibles, it was Gibson's. And if ever there was an MVP that merited being awarded based on intangibles, it was Gibson's.

I always thought that about Terry Pendleton's MVP.
   22. Daryn Posted: June 26, 2007 at 04:09 PM (#2418371)
Actually, I always thought the first sentence applied. I thought Bonds merited the MVP.
   23. RoyalsRetro (AG#1F) Posted: June 26, 2007 at 04:26 PM (#2418395)
So what do people think about his 1988 MVP? Here's the voting:

Yikes, no one hit better than .313 that year? Only Straw hit more than 30 homers? Did they raise the mound that year?

I remember Gibby's one year with KC. I think he felt he had something to prove after getting let go by the Dodgers. He got off to a terrific start, and was among league leaders in several offensive categories in April. He then went on a 5-40 slump that brought his average down to the Mendoza line. I remember him being pretty awful in the field and he almost cost Bret Saberhagen a no-hitter that year. I also remember him hitting a ball over the entire right field seats onto the concourse.

Then we dealt him to Pittsburgh for Neal Heaton to make room for.....Kevin McReynolds. Blech.
   24. JPWF13 Posted: June 26, 2007 at 04:45 PM (#2418421)
Did they raise the mound that year?


No they expanded the strike zone.

In 1987 the NL went .261/.328/.404 and the AL went .265/.333/.425 it was the most offensive year in literally decades

They then expanded the strike zone- which pre season led to a great quote from BJ- "giving Danny Jackson a high strike is like giving a Grizzly bear the first swat".

In 1988 the NL went .248/.310/.363 and the AL went .259/.324/.391 (and Danny Jackson went 23-8 2.73 in 261 IP and hurt his arm...)

In hindsight 1988-1992 looks like an artificial lull in offensive levels- from 1984-1987 Baseball was clearly trending towards mroe offense and home runs- MLB put tey brakes on with a 1987/88 rule change.
   25. David Concepcion de la Desviacion Estandar (Dan R) Posted: June 26, 2007 at 06:22 PM (#2418511)
Win Shares has a strange Will Clark fetish. I'd be interested if a WS expert could explain what exactly makes it so high on him. Did those Giants outperform their Pythagorean records or component stats? Why didn't Kevin Mitchell benefit to the same degree, then?
   26. DavidFoss Posted: June 26, 2007 at 06:38 PM (#2418529)
Did those Giants outperform their Pythagorean records or component stats?

That was the first thing I checked. They underperformed by 3 games that year. In season durability is one issue (with WS's low baseline). Lack of positional adjustment and the small contribution to fielding probably helps Clark, too.
   27. JPWF13 Posted: June 26, 2007 at 06:53 PM (#2418537)
Win Shares has a strange Will Clark fetish.


Seems like it
The 1988 Giants won 86 Games
that's 258 win shares to go around
Team OPS+ was 108, ERA+ was 97
Offense gets 55%? (pitching 45, ignoring Dee for the moment)

The team scored 670 runs (637 runs created)
Will the Thrill had 113 runs created or 17.7% of the Giant's offense that year.
If the Giant's offense was assigned 142 winshares (55% of 258) and Will gets 17.7% then that's 25...
He did hit .316/.457/.620 with RISP and I think WS adjusts RC for that- but he also hit .251/.358/.398 when the game was tied...

seriously, unless BJ is going to divulge ALL his formulas his WS marks for Clark are pretty inexplicable
   28. RoyalsRetro (AG#1F) Posted: June 26, 2007 at 07:12 PM (#2418548)
Win Shares does favor "thrill" quite heavily.
   29. Chris Cobb Posted: June 26, 2007 at 07:36 PM (#2418573)
seriously, unless BJ is going to divulge ALL his formulas his WS marks for Clark are pretty inexplicable

??

All the formulas are in print in the win shares books -- your estimtes are off because you are not using the formulas correctly.

For batting win shares, a player gets a percentage of the team batting win shares equal to his share of team runs created _above the background runs_, not of total team runs created.

Following the win share book, the 1988 SF Giants won 83 games, for 249 win shares to go around.

Hitters were assigned 138.3 win shares, or 55.5% of total. So far, so good.

Now, Clark had 113 runs created out of 637 team runs created.

Background runs created = Outs X league avg. runs/out X park factor X .52

For Clark that is

31 = 433 X .14346 X .96 X .52

For the Giants as a team that is

316 = 4413 X .14346 X .96 X .52

Clark's bating win shares are thus

((113-31)/(637-316))*138.3 = 35.3

Win shares actually assigned Clark 34.3 bws for 1988, so some of the numbers I've used must be a little bit off, but it's clear that the printed results correspond to the printed formulas.

It's questionable that batting win shares are the best measure of Clark's offensive value for that season, but there's nothing inexplicable about where those numbers come from.
   30. Dr. Chaleeko Posted: June 26, 2007 at 07:59 PM (#2418601)
James specifically talks about Clark (esp the 1989 season) in one of the book's miniessays. The gist goes like this (working form memory, don't have it at my work):
-the park adjustment helps him a little
-avoiding DPs helps him a little
-good SB rates (even on small amounts of steals) helps him a little
-good defense helps him a little
-good with RISP helps him a little

Add up the littles and you get a lot.

That was my recollection of his explanation of Clark.
   31. JPWF13 Posted: June 26, 2007 at 08:17 PM (#2418621)
Background runs created = Outs X league avg. runs/out X park factor X .52


So essentially since the Giants created so few runs, subtracting "background runs"* gives Clark a disproportionately large of their offense (25.5%).

The Giants created 637 runs with 4413 outs (Will Clark created 113 with 433)
If Will Clark were replaced by a "background" hitter (ie: 31 runs in 433 outs)**
The Giants would have created only 555 runs and their pythag would be 75-87 rather than 86-76
11 wins (11 * 3 = 33)...***

*BJ doesn't like the phrase "replacement level"?

** 31 runs in 433 outs? Even in 1988 there was no team that would play a 1B that BAD. The worst 1B in the league that year was Moreland- 52 RC in 411 outs - equal to 55 in 433 outs (next worse was Stubbs with 57 per 433 outs) Both played in parks with similar park factors to SF.
Without the Thrill the Giants wouldn't have scored worse than 575 runs - a Pythag of 78-84 (8 wins short, 8*3 =24) I know James deliberately set his baseline low, but in this case it acts to really boost the Thrill's offensive winshares a bit too high

*** so if WS has a replacement level built in why did THT feel the need to calculate a Winshares above bench stat? Because James' "baseline" is way way too low...
   32. JPWF13 Posted: June 26, 2007 at 08:30 PM (#2418636)
Add up the littles and you get a lot.

That was my recollection of his explanation of Clark.


My favorite was his explanation for why Kevin Seitzer in 1987 was a better HITTER than Mark McGwire 1987. Why? Because Seitzer had a better offensive winning percentage ("OWP").
Seitzer hit .323/.399/.470 in a slight hitter's park
McGwire hit .289/.370/.618 in a pitcher's park - how could Seitzer have a higher OWP?

What was OWP? OWP was defined as what the winning percentage would be if you took an entire team full of one player combined with average defense and pitching.
However, OWP in practice didn't do that- James took a shortcut, rather than figure out park factors and adjust either his league baseline or the batter's production he simpy took the aggregate r/g figure for the player's team- so Seitzer got the benefit of KC's 118 ERA+ staff and KC's 93 OPS+ offense and McGwire got the disadvantage of Oakland's 95 ERA+ staff and Oakland's 107 ops+. (in other words the context for Mcgwire was not "average" it was inflated because Oakland had good hitting and poor pitching, Seitzer's context was absurdly low for the same reason).

OWP was a great idea, but James' execution was poor and he eventually abandoned it-
It's something that drives me (and others I've noticed) nuts, he'll have a great idea work on it, declare it finished, abandon it and then when anyone who tries to tweak his ideas/formulas - he claims it's all a waste of time, let's move on etc etc...
   33. Chris Cobb Posted: June 26, 2007 at 09:42 PM (#2418683)
JPWF13: Nice practical explication of the problem with the WS "background runs" baseline!
   34. Joey Numbaz (Scruff) Posted: June 27, 2007 at 01:36 AM (#2418998)
JPWF13 - what are you talking about in post 34?

OWP had nothing to do with the team's pitchers. Can you please provide a source (like a page number from an Abstract or something) for your accusation? Maybe I'm forgetting something he wrote?
   35. Cblau Posted: June 27, 2007 at 02:48 AM (#2419220)
According to Player Win Averages, the MVP probably should have been Eric Davis. Player Win Averages
   36. J. Lowenstein Apathy Club Posted: June 27, 2007 at 04:46 AM (#2419357)
Because James' "baseline" is way way too low...

It's only too low if you want it to do something it's not supposed to do. It works extremely well to do what it's supposed to do. My buddies at THT created WSAB because they want to use Win Shares to do something it wasn't designed to do.

The fundamental thing to remember about win shares is that if you want to use the concept of "replacement level" in *any* way, you should use something else. As I put it last week to a friend, win shares are about accounting, not economics.
   37. Dr. Chaleeko Posted: June 27, 2007 at 01:00 PM (#2419498)
Wow. Gary Varsho was the 71st best hitter in 1988 according to PWA. He got a lot of wins out of that .274/.280/.315 line.
   38. DanG Posted: June 27, 2007 at 01:43 PM (#2419541)
(he played RB at Michigan, don't you know!)

You're gonna get in trouble, Eric. Gibson played WR at Michigan St.!
   39. Mike Green Posted: June 27, 2007 at 02:01 PM (#2419562)
This will come up when it's time to discuss Will Clark, but it was not only Win Shares that liked Will Clark's defence during the period 1988-90. BP's metrics also did.

Taking a macro look at the SF run prevention ledger in 1989, it is easy to see why. The club led the league in DER at .725, and trailed the league in walks and strikeouts so had many balls in play. The outfield of Kevin Mitchell, Brett Butler and Candy Maldonado was unlikely to be anywhere more than a little above average overall. The infield of Riles, Uribe, Thompson and Clark probably contributed quite significantly to the overall ledger. There is room for debate about how much the first baseman actually contributes to the success of the inner defence in fielding range and receiving, but if any first baseman could be responsible for saving a significant number of runs in a season, Clark's 1989 would be one of the top likely candidates.
   40. JPWF13 Posted: June 27, 2007 at 02:08 PM (#2419567)
OWP had nothing to do with the team's pitchers. Can you please provide a source (like a page number from an Abstract or something) for your accusation? Maybe I'm forgetting something he wrote?


I have it at home
it was the 1988 Abstract- he had an article arguing that Seitzer was a better hitter than McGwire- because Seitzer's OWP was higher- if you look at an earlier abstract whrre he explains how he calculates OWP you realize the mistake he made - he developed RC then came up with RC/27 outs which he referred to as a ERA for batters
THEN he developed a pythag based upon that RC/27outs- but he did not use the league r/g figures, he used the r/g figures for that player's TEAM- which he even admitted was a short cut method of adjusting for parks

so Seitzer played for a team that gave up few runs and scored few, (let's say 4 and 4) so seitzers rc/27 was pythaged against a baseline of 4r/g
McGwire played for a team that scored 5 and gave up 5, so he was pythaged agaisnt a baseline of 5 runs/g (he didn't call it baseline- he called it "context")

so if Seitzer's RC/g was 7, his OWP would be .754 (7/4), if McGwire's RC/g was 8, his OWP would be lower at .719 (8/5)
   41. JPWF13 Posted: June 27, 2007 at 02:10 PM (#2419569)
OWP had nothing to do with the team's pitchers.


It wasn't SUPPOSED to, but the way he actually calculated it, it did, as well as one's hitting teammates as well.
   42. David Concepcion de la Desviacion Estandar (Dan R) Posted: June 27, 2007 at 03:40 PM (#2419661)
Clark's defense in 1989 was indeed stellar--Chris Dial's Zone Rating data has it at +10, putting it in a tie (with many others) for the 18th-best defensive season by a 1B since 1987. But here we're talking aboug 1988, which was only average by Zone Rating.

Note that WS wasn't actually that gaga about Clark's defense--he was consistently between 2.3 and 2.9 WS from 1988-91, slightly above average but not the stuff of which Gold Gloves are made. It's his batting WS that are batty. He had 41.1 batting WS in 1989, 5.7 more than his teammate (and league MVP) Kevin Mitchell. In fact, their offense was about equally valuable that year (at least according to my measure--Mitchell had 0.2 more offensive wins above average, but Clark had 35 more PA which makes up the difference relative to replacement level), and BP has Mitchell with 5 more BRAR than Clark. Since they were teammates, it can't be some funky team Pythagorean outperformance factor...what in God's name is going on here?

Clark was a nudge more valuable than Mitchell in 1989, but it was because of his fielding, not his hitting.
   43. Mike Green Posted: June 27, 2007 at 04:05 PM (#2419698)
Sorry, DanR. For 1988, Clark's defence may also have been above average, as BP has it. The team DER was significantly above average, with the even less impressive Kevin Mitchell at third. Zone Rating does not, as I understand it, attempt to apportion the success of the infield in turning ground balls into outs among the fielder and the first baseman. With Uribe and Mitchell on the left side of the diamond, that's a relevant factor. I seem to remember Clark making a fair number of swipe tags; I am sure that Steve Treder can supply names, dates and licence numbers with respect to all the events!
   44. Dr. Chaleeko Posted: June 27, 2007 at 04:26 PM (#2419726)
You're gonna get in trouble, Eric. Gibson played WR at Michigan St.!

Michigan/Michigan State, same thing. RB/WR, same thing. It's college football, it doesn't really matter and no one really cares.

[runs awayyyyyyy]

Thanks for the correction, Dan.
   45. David Concepcion de la Desviacion Estandar (Dan R) Posted: June 27, 2007 at 06:03 PM (#2419856)
BP FRAA includes scooping? News to me...I thought we had no way to measure scooping. (infield errors are part of Fielding WS for 1B, but only like 10%, and Fielding WS for 1B are in such a compressed range anyway that it doesn't matter).
   46. jimd Posted: June 27, 2007 at 07:08 PM (#2419961)
It wasn't SUPPOSED to, but the way he actually calculated it, it did, as well as one's hitting teammates as well.

James used the same technique in the first Historical Abstract. I remember noting at the time that it was flawed, though it was better than not attempting any park adjustment. Lacking park adjustments for all teams in all years, he adjusted the batting records dependent on team scoring context. However, that can't tell the difference between an average team playing in a 110 park, and a neutral park team which scored 10% more runs on offense while giving up 10% more runs on defense. It's better than doing nothing because the former is more common than the latter.
   47. Paul Wendt Posted: June 27, 2007 at 07:26 PM (#2419978)
WS wasn't actually that gaga about Clark's defense--he was consistently between 2.3 and 2.9 WS from 1988-91, slightly above average but not the stuff of which Gold Gloves are made. It's his batting WS that are batty. ...what in God's name is going on here?

batting well when it counts more than average.
in a word, clutchhitting.
   48. Mike Green Posted: June 27, 2007 at 07:26 PM (#2419979)
DanR, BP does not, as far as I know, publish any description of how FRAA is calculated. What I said is that BP has Clark as an above average defensive first baseman in 1988 (as he was consistently throughout his early career). That conceivably could be consistent with an average Zone Rating.

Infield errors are, of course, a weak proxy for a first baseman's receiving efficiency.
   49. Paul Wendt Posted: June 28, 2007 at 02:18 PM (#2420931)
Bill James on Will Clark 1989
The "mini-essay" is Win Shares page 213.
Eric's above description may be better than mine because he does not emphasize the clutchhitting effects here, merely lists them among many adjustments that add up to boosting Will Clark 1989 to the best batter-runner season of the 1980s. Of course some of the adjustments are included in every sabrmetric measure so they have no comparative weight.

The park factor BJ used is .9110. The two credits for clutchhitting amount to 11 runs, 8 for batting .389 with runners in scoring position and 3 for hitting 13 of 23 home runs with runners on base. Two other credits not included in every measure are for low GIDP and good (8 for 11) base stealing success; the mini-essay doesn't give run value for these two.
   50. JPWF13 Posted: June 28, 2007 at 02:30 PM (#2420940)
The park factor BJ used is .9110.


FWIW BBREF gives San Fran a 97 for that year (96 for '88 and 96 for '90)

Actually looking at SF's home/road numbers
I get raw single season park factors:
1988: .867
1989: .9221
1990: 1.015

Seems a bit odd... According to Ballparks.com there were no changes in dimensions or anything at that time... weather?
   51. David Concepcion de la Desviacion Estandar (Dan R) Posted: June 28, 2007 at 03:49 PM (#2421033)
Well, there ya go. .911 is really low, BB-ref has '89 SF at 97 (guess that's the three year average). GIDP and basestealing I account for, but sure, if you give him 11 runs for clutch hitting then it's a totally different equation--11 runs is what 1.3/1.4 wins in the 1989 NL run environment?
   52. Dr. Chaleeko Posted: June 28, 2007 at 08:37 PM (#2421492)
James is using a weighted five-year park factor (or am I thinking of park-run adjustment?), if I remember off the top of my head, that's weighted most heavily to the current year (obviously, I suppose).

I'm not crazy about the clutch stuff, myself, so Clark could be puffy. On the other hand, an interesting coincidence is that those 11 runs is about exactly the difference between Clark's WARP1 and Gibson's/Hershiser's!
   53. John (You Can Call Me Grandma) Murphy Posted: June 28, 2007 at 08:46 PM (#2421499)
I actually agree with both the clutch aspect of WS and their handling of park effects.
   54. Paul Wendt Posted: June 28, 2007 at 08:47 PM (#2421500)
OK. I scribbled that pretty quickly last night, posted it this morning.
Someone check for transcription error.

Chris Cobb #31 is right, I believe.
>>??
All the formulas are in print in the win shares books -- your estimates are off because you are not using the formulas correctly.<<

But I won't have time to check the detail explanation of the park factor myself.
   55. Los Angeles Waterloo of Black Hawk Posted: June 29, 2007 at 07:49 PM (#2422562)
OWP had nothing to do with the team's pitchers. Can you please provide a source (like a page number from an Abstract or something) for your accusation? Maybe I'm forgetting something he wrote?

JPFW13 referenced this above, but for a period there James made the "opponent" in the Pythagorean to calculate OWP the average runs scored per game for the player's team and opponents. So if a guy had 8.00 RC/27, and his team scored 4.50 and allowed 4.50, it would be (8^2)/(8^2+4.25^2).
   56. Devin has a deep burning passion for fuzzy socks Posted: July 01, 2007 at 02:59 AM (#2424960)
I'm not an expert in park factors, but looking at the Win Shares book, he converts the "park factor" to the "park run adjustment". Doing the math on that, the 89 Giants park run adjustment is .959545. That's the number that gets used when he's converting Runs Created into Batting Win Shares. How that compares to a BBREF Park Factor I could not tell you.
   57. Paul Wendt Posted: July 01, 2007 at 02:16 PM (#2425096)
The adjustment must account for (a) that the team plays ~1/2 of its games at home; (b) that the team would play 1/n of its games at home under a league-average-park schedule. For NL 1989, n=12.

Interleague play complicates the theory. So do the Divisions: NL teams have not played 1/n of its away games in each of the other parks since 1969, but most sabrmetricians use 1/n even today.
During the 12-team era (1969-93), by good fortune for sabrmetricians and analysts the two high-scoring ballparks Atlanta and Chicago were split between East and West. The cookie cutters were split Cincinnati in the West and SL-Pit-Phi in the East, which is better than zero-four.

I suppose that park factors in the Western Division were a bit less than 100 during the era, and Eastern ones a bit greater; if so, Western pitchers and Eastern batters are slightly overrated by the standard use of park factors. Maybe it adds up. With Dave Winfield on the board, a practical illustration would be the bias in measured "wins" during over his eight seasons in San Diego. Or since Winfield is a cinch, Will Clark pre-1993.
   58. Paul Wendt Posted: July 01, 2007 at 02:22 PM (#2425101)
"the standard use of park factors."
Each team played 45 games at the five other in-division parks (9 each) and 36 games at the six other-division parks (6 each), a detail ignored by accounting merely for 81 games at the eleven other parks (7.36 each).
   59. Spahn Insane Posted: July 07, 2007 at 02:29 PM (#2432004)
That's kind of unfair, Craig. Gibson was a football player in college and had a lot of catching up to do. Baseball has a way of neutralizing tools in a way that football and basketball don't.

Yes, and his injuries as a baseball player weren't nearly as self-inflicted as Doc's and Straw's.
   60. Cuban X Senators Posted: July 08, 2007 at 04:04 PM (#2433559)
I can't remember my source for this, but it's my understanding that there have been a couple of times when things have been allowed to get a little lax on folks getting their ballots in immediately after the season -- new crackdowns have come and then faded in the wake of MVP awards that seemed to be effected by late arriving ballots.

This probably is no longer such a huge issue since ballots are no longer sent out in mid-September to be mailed back, but are instantaneously zapped about the internet and quickly missed rather than assumed to be in the mail if they don't arrive.

But the two awards I heard had been swayed by late arriving ballots were Stargell's '79 and Gibby's '88.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
The Id of SugarBear Blanks
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 0.4287 seconds
49 querie(s) executed