Jim Albright / the japanese insider
Sadaharu Oh and Cooperstown, Part II
III. Statistical Analysis
A. My projection
You
have now entered the section of the discussion of Oh some will dismiss as pure
fantasy. If you are one of the folks who do not believe it is possible to
project what a player would do in the major leagues from his performance in
another league, you may want to skip this section entirely. We will use projections because they place
the accomplishments for a player from a non-major league situation into a
readily understood context, namely major league performance. Once we enter such a readily understood
context, it is easier to get a reasonable fix on the quality of the player.
I’m
going to describe the techniques used to arrive at all the career numbers for
the Oh projection in a fair amount of detail
If you don’t want to be bothered with all that math, feel free to skim
the text until you get to the final projection and the evaluation of what those
numbers mean. That projection will look
like a career batting line, with at
bats, runs, hits, doubles, triples, home runs, RBI, walks, batting average, on
base percentage, and slugging percentage.
The
biggest single factor in reaching the estimate is the difference between facing
Central League pitching in Central League parks versus facing major league
pitching in major league parks. Fortunately, there were 66 players who played
in the Central League between 1960 and 1980 who also played in the majors.
These players provide the basis for determining the size of this factor,
because the differences in their home run rates between the majors and the
Central League is almost exclusively the result of this factor. I didn’t see a
strong bias toward good home run parks in the majors nor in Japan among these
players. It is true that if one weighted the at bats by age, the players would
unanimously be older in Japan. Fortunately, home run hitting is what Bill James
has termed an "old player’s" skill, one which seems to decay more
slowly in the players who have this skill and will have the greatest effect on
the calculation. However, I also endeavored to use a method which would
minimize the age issue in order to provide the most accurate measurement
possible, since I am not aware of any studies which would give a reasonable
estimate of the size of the effects of aging on home run hitting. Also, I
wanted to ensure that the quality of players was as close to identical as
possible. An example of a situation I wanted to avoid was comparing Willie Mays
as a part-time 19 year old in the Negro Leagues to the accomplished Willie Mays
of the major leagues. I have seen writers place Willie’s totals in both
leagues, and make no effort to account for the vast difference between a bit
player’s effect on the Negro League total and a Hall of Famer’s effect on the
major league total. It is clear that such an approach can seriously skew the
data.
My
solution to getting the best possible comparison between the Central League and
major league data was to do the following was first to determine in which
league the player had the fewest career at bats. I then entered that leagues at
bats and home runs in the columns assigned to it. I then got the same number of
at bats in the other league, starting with the nearest season in time, and
prorating a the season home run total when I had to use part of a season to
reach the at bat total I was seeking. Two examples should help to explain how
this works.
Sam Perlozzo his no
homers in 26 major league at bats in 1977 and 1979. These were the lower
totals, and were entered in the major league columns. He want to Japan in 1980
and hit 15 homers in 473 at bats. I prorated the Japanese figure to 26 at bats,
which made his home run total 0.8. Roy White played several seasons in Japan,
but 1980 is the only one which qualifies for our study. That season, Roy hit 29
homers in 469 at bats. His nearest season in the majors was 1979, in which he
hit 3 homers in 205 at bats. His next nearest season was 1978, when he hit 8
homers in 346 at bats. However, I only needed 264 at bats from 1978 to reach
469, so I prorated the 1978 home run figure to 264 at bats, which gives Roy a
major league home run figure of 9.1 in 469 at bats. The man with the largest
individual effect on the results is Willie Kirkland, who had 2323 matched at
bats with 100.2 homers in the majors and 126 in the Central League. These are
the largest totals by any individual in each category . The overall totals are
23,817 matched at bats, 575.0 major league home runs, and 1071.9 Central League
homers. Thus, we will multiply Central League homers by 575.0/ 1071.9 or 0.536
to account for this difference.
This
technique was repeated for hits, doubles, triples and walks. The adjustment figures yielded by this
approach were:
Hits: 0.904 2B: 0.829 3B 2.149 HR: 0.536
BB: 1.148
Now that we have a
major league estimate to work from, we
can address the issue of Oh’s career. Injuries aren’t a factor in Oh’s
case, so we don’t have to concern ourselves with how to handle them. The issues
regarding how much playing time Oh should receive are subjective in nature. I
set several guidelines to follow. First, there would be no fictitious seasons
unless he needed 10 or less homers to meet to meet a significant milestone
(500, 600, 700, 715, or 750 homers, for example ). Since Oh turned out not to
meet that criteria, no fictitious seasons were created. Another guideline was
that Oh’s first solid or better season in Japan would be the one we used as his
last year in the minors in our determining when he came to the majors (in real
life, he came to the Central League right out of high school, which would not
have happened in the majors). The reasoning behind this guideline lies in two
facts: first, the major leagues like to have a player have success at the
highest level of the minors before coming to the majors, and second, the fact
Japanese baseball has long been somewhere in quality between the highest level
of the minors and the major leagues. This guideline excluded the 1959 and 1960
seasons (1959: .161 average, 7 HR in 94 games, 1960: .270 average, 17 HR in 130
games). A third guideline was that so long as Oh was a productive hitter for a
first baseman, he would have his totals adjusted upward to reflect the longer
major league season. This took care of the 1963 through 1979 seasons as far as
I was concerned, because even taking away 10% of his batting average as would
have happened in real life and going to the major league home run figure he was
averaging between .258 and .300 with 21-34 homers and a lot of walks. If there
was a question about his productivity, I felt the two best other options were
either to give him no playing time at all or to give him his actual playing
time. My reasoning behind this was choosing any other amount of playing time
was even more subjective than those three choices (none, actual, or adjusted to
major league length), and thus less desirable. If I really couldn’t decide
between two of those choices, I probably would have selected an average of the
two. I was able to reach a decision within those three choices for the two
difficult seasons, however.
The
three hard decisions in terms of how much playing time to give Oh, as far as I
am concerned, are 1961, 1962 and 1980. Under my initial guidelines, Oh should
have his rookie season in 1961, and even though he slid back from his 1960
marks to a .253 average with 13 homers and 64 walks in 127 games. Until I ran
the batting average numbers, I thought that it should be his rookie season,
partly because it was an expansion year. When those marks are converted to major
league levels, they are not impressive for a first baseman (.228 and 7 homers
in his actual playing time). The 1962 season makes a much better rookie year,
and although it converts to a .246 average with 24 homers, that’s a solid year
for a 22 year old rookie seen as a future star, especially when he throws 72
walks into the mix. That season, when seen as a rookie year, deserved to be
expanded to the full major league schedule, in my opinion.
The other hard case is 1980. The situation here is a 40 year
old star who has played well every year for the past 18 seasons, but then starts
to be slowed by advancing age. His major league equivalents for his actual
performance (including keeping his number of games at the actual level) are a
.214 average with 16 homers and a 72 walks for a .323 on base percentage. I
think he’s entitled to one off year before being forced out of the game, so
it’s easy to dismiss the option of giving him no playing time. I think the most
likely way this would play out is that at the start, Oh’s manager keeps sending
him out there to play, hoping he’ll come out of his "slump".
Eventually, he’ll try giving his aging veteran a day or two off in hopes that
is the answer. If it worked, you can be sure the manager would do the same thing
every time Oh slumped again. This would keep Oh from being an everyday player.
If it didn’t, the manager would probably realize Oh had gotten old and would
want to see what his options were in the organization to replace Oh, maybe even
next year. Either way, a proud man like Oh would see the handwriting on the
wall and announce his retirement at the end of the season. This would help
ensure him a good amount of playing time, because there would be fans in the
stands every day thereafter who would want to see Oh play one last time. This
would create pressure on the manager to play Oh. Any way you look at the
situation, the best of my three favored choices is to give Oh his actual
playing time, at least in my opinion. He might play more than that, but it is
at least as likely he’d play less. When you’re making an estimate, that sure
ought to help you pick that number.
The
walks figure we will actually use is 1.000, without any upward adjustment for
playing time for this one piece of data.
Oh already has what would be third in career number of walks, and the
adjustment figure given above multiplied by the factor for longer seasons would
give him 39% more walks. This seems too
high, and may be seen as giving some additional credence to the argument Oh
received four strikes per at bat, whether it was because he was Sadaharu Oh or
because he was a Giant, or some combination of those two. This approach is conservative, but still
recognizes that Oh in fact had superb command of the strike zone. I effectively eliminated the extra walks
called for by the walk factor. However,
I concluded it wasn’t accurate to eliminate the season length adjustment in the
same fashion. I will detail what I chose
to do about that issue when I talk about the season length adjustment. Similarly,
we chose to use Oh’s actual career stolen base figure of 84 both because stolen
bases are of little import in assessing Oh’s career and because the players
playing in both leagues were predominantly slow sluggers. Since Oh arguably is in that classification,
perhaps that isn’t a serious concern, but the unimportance of stolen bases to
an examination of Oh’s career value frankly did not justify the work necessary
to come up with a conversion factor.
As
indicated above, I chose 1962 as the first season of Oh’s major league career.,
with the exceptions of Oh’s final season of 1980 (as discussed previously, I
used a 1.000 season factor that year) and the slightly shorter 1972 major
league season due to a strike, the seasonal adjustment factor is the major
league 162 games divided by the length of the NPB season that year for each
team. I multiplied this factor times at
bats and the statistics I have major league conversions for (hits, doubles,
triples, home runs) except for walks. If
I had used just this factor (and not the factor I came up with for walks to
account for the differences between the leagues) on walks, he would have had
455 added to his total. It doesn’t feel
right to have all those plate appearances disappear, so what I did is made them
all at bats and gave him his career batting average times those at bats, his
rate of doubles per at bat times those 455 AB, and so on. His production is slightly diminished because
he gains 330 outs in place of those walks, but I think it is better than giving
him nothing. Also, it makes his
production per plate appearance less, which means I’ve made the projection more
conservative. This helps ensure we do
not seriously overestimate Oh’s production in the major leagues, and I find
that desirable.
The
last step to coming up with a “park” adjustment. If I had runs scored in Yomiuri home games
versus runs scored in road games, I’d divide the runs scored in road games by
the runs in home games, adjust if the number of home and road games weren’t the
same, and take the square root and multiply it times hits, doubles, triples,
home runs and walks. Since I don’t have
that data, I substituted the league average of runs per game divided by the
runs per game in Yomiuri games, and then ran the square root and apply the
adjustment to the same statistics. It’s
the closest number I could come up with to replace the data I wanted. Over Oh's career, it’s reasonable, though I
suspect the fact Yomiuri had two of the very best hitters ever in NPB together
for so many years probably works to influence the result in the direction of
saying the Giants played in a park that encouraged scoring that really has
nothing to do with the effect of the park.
The run factor I’m using was multiplied by each season’s plate
appearances, added together, and the sum divided by his career plate
appearances to yield the park adjustment factor. The square root, which is the one applied to
the various statistics mentioned above, is 0.979 in this case.
While I had to use
season by season data to deal with playing time issues in order to make my projections, the
adjustment factors are designed for Oh’s entire career, not individual
seasons. Therefore, we will not use the
single season projections as part of my formal presentation regarding Oh’s
worthiness for the HOF. Instead, we will
restrict ourselves here to working with the career totals estimated for Oh, as
these totals are within the intended bounds of the adjustment figures.
At Bats and all other factors will rise by the
season length factor, but the net result for most factors will be that Oh’s
totals will actually drop, especially after his first three seasons are dropped
on the grounds he wouldn’t have reached the majors until 1962. The abbreviation of his career will somewhat
counteract the drops dictated by the adjustment factors in the percentage stats
(average, OB pct, and slugging pct). He
will get about 8% more hits in the seasons after 1962, but in over 20% more at
bats in those same seasons. He will hit
less than 64% as many homers in 1962-1980 as he did in real life. The adjustment factors demonstrate that the
circumstances Oh faced were not of major league caliber.
I wanted to fill in two other
categories, runs and RBI. Here, my
relatively recently acquired skills in doing regressions were quite
useful. Given that I had already
projected Oh’s major league batting marks, I could use a regression of major
league hitters to come up with run and RBI figures. I chose to limit my set of players to guys
after 1919 (the game was very different before the home run became such a large
element of the game), and also to guys who played enough to have a chance at
making the Hall of Fame. The only
position players not to have at least 5000 plate appearances to make the Hall
played several seasons in the Negro Leagues.
For runs, I dropped home runs
from the regression, as by definition, guys score a run when they hit one of
those. I calculated how many times guys
started on base at first (hits – doubles – triples – home runs), second
(doubles) and third (triples) plus stolen bases. Triples and stolen bases are strongly related
to speed, so those categories not only involve the chance an average player
could score from the base reached (and of course, players can steal third or
home) but also whether the player in question might have the speed to score a
little more often from any given base.
The least predictive p value of those four categories was in the 10-30
level. The r-squared value is over
.927, which is to say this regression tracks the number of career runs quite
well. And it becomes even more accurate
when the home runs are added back in. The
standard error is just under 60.9 runs.
The coefficients are 0.245093 for times on first, 0.479712 for doubles, 1.522654
for triples, 0.453462 for steals and -34.7247 for the intercept. Just for clarity, the formula is:
0.245093 * (hits- doubles-
triples- home runs) + 0.479712 * doubles + 1.522654 * triples + 0.453462 *
steals = runs – home runs
Using the projection results
and Oh’s 84 career steals, the result when the home runs are added in is 1789.
For RBI, I started with the
thought of doing a regression of singles, doubles, triples and home runs. However, the regression had a coefficient of
less than -1.5. Now, I could accept that
because of speed, a guy might hit leadoff and have less RBI opportunities. I could buy a negative coefficient between 0
and -0.5 without much fuss. I might have
accepted between -0.5 and -1.0. However,
this is less than -1.5, and for a hit that drives in every runner on base. I preferred to do the regression without
triples. The least predictive p
value is expressed in terms of 10-35 power.
The r squared value is over .95, which is excellent, and the standard
error is just over 67.7 runs. The
coefficients are 0.196225 for singles (hits – doubles- triples – home runs),
0.693741 for doubles, and 1.96489 for home runs, with an intercept of
17.31794. For clarity, the formula is:
0.196228 * (hits – doubles –
triples – home runs) + 0.693741 * doubles + 1.96489 * home runs +
17.31794.
Plugging the values from Oh’s
projection into this regression formula, the result is 1723 RBI.
Putting all of the above
together, Oh’s career line is most impressive:
|
AB |
R |
H |
2B |
3B |
HR |
TB |
RBI |
BB |
avg |
obp |
slg |
|
||||
|
10394 |
1789 |
2845 |
373 |
39 |
548 |
4942 |
1723 |
2189 |
0.274 |
0.400 |
0.475 |
|
||||
I find it most interesting that this projection closely
resembles a) his actual performance in exhibitions against major leaguers, and
b) the anecdotal assessments major leaguers made of him. I guess those who deny the value of such
projections will claim that I was merely lucky.
I expect such a reaction from that group, because otherwise, they'd have
to concede that there is some real validity to the projections. I’m sure you can guess my conclusion, but I’m
willing to let each individual reader consider the evidence and reach his or
her own conclusions.
One way we will use the projection in the formal case
examining Oh’s worthiness for a plaque in Cooperstown is to look at how players
with various totals in certain categories fared in terms of induction into the Hall of Fame. I’m
excluding active players as their career totals are not yet known, and Pete
Rose and the guys commonly regarded as having used PEDs (Barry Bonds, McGwire,
Sosa, Palmiero, Alex Rodriguez, and Sheffield) really can’t be regarded as
instructive in terms of Oh’s qualifications, as the current exclusion of these
guys has very little to do with an assessment of their statistical achievements
at face value. I regard Jim Thome, Derek
Jeter and Chipper Jones as highly likely to make the Hall of Fame when they get
their chance, but Bobby Abreu and Johnny Damon as longshots at best. I suspect David Ortiz will make it eventually
despite the bias against the DH, but I’m not sure of that, and I strongly doubt
he’ll do it quickly after the writers get their first chance to vote on his
case. I am aware Ortiz is listed in the Mitchell
report as having tested positive, but he’s the one guy the commissioner’s
office has publicly exonerated, he hasn’t tested positive on any other occasion
I’m aware of, and the public perception, accurate or not, seems to me to be
that he was not using PEDs.
There are eleven players
within 416 plate appearances of Oh’s projected total , and except for Bonds,
Jeter and Alex Rodriguez, all are in the Hall.
I’d say that means all nine of the guys who aren’t considered PED users
have made the Hall or will:
Player |
PA |
Cal
Ripken |
12883 |
Eddie
Murray |
12817 |
Stan
Musial |
12718 |
Barry
Bonds |
12606 |
Derek
Jeter |
12602 |
SADAHARU
OH |
12583 |
Craig
Biggio |
12504 |
Willie
Mays |
12496 |
Dave
Winfield |
12358 |
Robin
Yount |
12249 |
Alex
Rodriguez |
12207 |
Paul
Molitor |
12167 |
In home runs, there are 18 guys with between 500
and 600 home runs. McGwire, Palmiero, Manny
Ramirez and Sheffield are all linked to PEDs.
Thirteen of the other 14 are in, and the one who isn’t is David Ortiz,
who I regard as about a 50% chance of making it.
Player |
HR |
Frank
Robinson |
586 |
Mark
McGwire |
583 |
Harmon
Killebrew |
573 |
Rafael
Palmeiro |
569 |
Reggie
Jackson |
563 |
Manny
Ramirez |
555 |
Mike
Schmidt |
548 |
SADAHARU
OH |
548 |
David
Ortiz |
541 |
Mickey
Mantle |
536 |
Jimmie
Foxx |
534 |
Frank
Thomas |
521 |
Willie
McCovey |
521 |
Ted
Williams |
521 |
Ernie
Banks |
512 |
Eddie
Mathews |
512 |
Mel
Ott |
511 |
Gary
Sheffield |
509 |
Eddie
Murray |
504 |
In walks, the only one of the
top six who is not in the Hall is the PED associated Barry Bonds:
Player |
BB |
Barry
Bonds |
2558 |
R
Henderson |
2190 |
SADAHARU
OH |
2189 |
Babe
Ruth |
2062 |
Ted
Williams |
2021 |
Joe
Morgan |
1865 |
Carl
Yastrzemski |
1845 |
In total bases, there are
seven major leaguers within 100 of Oh’s projected total. All are in the Hall:
Player |
TB |
Mel
Ott |
5041 |
Jimmie
Foxx |
4956 |
SADAHARU
OH |
4942 |
Derek
Jeter |
4921 |
Ted
Williams |
4884 |
Honus
Wagner |
4870 |
Paul
Molitor |
4854 |
Al
Kaline |
4852 |
In times on base, there are
five guys within 200 times on base of Oh’s projection. Four are already in, and I expect Jeter to quickly join them:
Player |
TOBwe |
SADAHARU
OH |
5013 |
Babe
Ruth |
5004 |
Tris
Speaker |
4998 |
Willie
Mays |
4960 |
Derek
Jeter |
4911 |
Eddie
Collins |
4891 |
There are twelve guys with
career on base percentages between .390 and .410 and at least 10000 plate
appearances. Sheffield is connected to
PEDs, Abreu is a longshot, but Thome and Chipper Jones are likely to make it
quickly. Moreover, all of them have over
1600 less plate appearances than the Oh projection, and only Thome and Jones
have better OBPs than the projection, and only by two and one points
respectively:
Player |
OBP |
PA |
Paul
Waner |
0.404 |
10766 |
Charlie
Gehringer |
0.404 |
10245 |
Jim
Thome |
0.402 |
10313 |
Chipper
Jones |
0.401 |
10614 |
Rickey
Henderson |
0.401 |
13346 |
SADAHARU
OH |
0.400 |
12583 |
Luke
Appling |
0.399 |
10254 |
Bobby
Abreu |
0.395 |
10081 |
Cap
Anson |
0.394 |
11331 |
Gary
Sheffield |
0.393 |
10947 |
Rod
Carew |
0.393 |
10550 |
Joe
Morgan |
0.392 |
11329 |
Honus
Wagner |
0.391 |
11749 |
In runs scored, there are
eleven players within 140 runs of Oh’s projection. We have Palmiero and his PED link, Jeter, who
should make it soon after the writers get to vote on him, and Johnny Damon, who
is a longshot but also has fewer runs:
Player |
R |
Derek
Jeter |
1923 |
Craig
Biggio |
1844 |
Frank
Robinson |
1829 |
Carl
Yastrzemski |
1816 |
SADAHARU
OH |
1789 |
Paul
Molitor |
1782 |
Mickey
Mantle |
1676 |
Dave
Winfield |
1669 |
Johnny
Damon |
1668 |
Rafael
Palmeiro |
1663 |
Ken
Griffey |
1662 |
Joe
Morgan |
1650 |
So there’s seven categories in which players not
associated with PEDs nor Pete Rose who did at least as well as Oh are almost
unanimously in or will be. No matter
what one thinks of these comparisons, if one accepts the projection as being
reasonably accurate, it is clear Oh is quite worthy of induction into
Cooperstown.
Bill
McNeil did a similar projection of Oh’s stats for his book King of Swat His projection was based on 550 at bats, and
I will put my projection in the same
terms.
AB H 2B 3B HR avg slg
Albright 550 151 20 2 29 .274 .475
As you can see, they are
rather similar. I have exchanged several
emails with Mr. McNeil, and he has graciously indicated to me (and granted
permission for me to share with you) that he feels that my projection of Oh is
superior to his, essentially because his system was devised in the context of
evaluating all Japanese players, while my approach was much more focused
on Oh’s circumstances. In either case,
we both project Oh to be worthy of the HOF.
In fact, Mr. McNeil’s book Other Stars has Oh as the third best first
baseman of all time, behind Gehrig and Foxx.
IV.
A Calculation of Oh’s MLB WAA and WAR
I’ve got one last way I’m going to try and put Oh into
proper context. I’m going to do my best
to calculate the MLB WAR for Oh from the information I have, starting with the
projection . That particular metric
seems to be the one that’s dominant today, and I’m mainly basing this on the
version used by baseball-reference.com.
The first step is to estimate batting runs created. I did this by adding total bases to walks,
and multiplying that total by 0.32 , adding that to hits multiplied by 0.26,
and multiplying the quantity (at bats minus hits) by 0.10. It’s essentially a linear weights equation of
0.32 times walks plus 0.48 times singles plus 0.80 times doubles plus 1.12
times triples plus 1.44 times homers minus 0.10 times outs. The result is 1982.3 runs. Next, I needed to figure out the number of
games (using 25.5 outs per game due to double plays, caught stealing and
baserunning errors) Oh used. This is
simply his (at bats – hits)/ 25.5, and the result is 296.0. Since this is a major league evaluation, I had
to decide what run figure to use. My
first rule is that players in a league which don’t use the DH in league games,
like Oh’s Central League, would be placed in the National League once the
American League began using the DH in 1973.
For the seasons before that, I used whichever League had the most runs
per game, as this would yield the most conservative result. I think there might have been one exception
to the assertion that the National League figure was used for each season Oh is
projected to play in the majors. I
figured out the career average run total by weighting it in two ways: one by multiplying by Oh’s projected plate
appearances, and the other by multiplying by Oh’s projected outs, calculated as
at bats minus hits. I added up all the
weighted numbers and divided the plate appearances calculation by Oh’s
projected plate appearances, and divided the weighted projected outs total by
Oh’s projected (at bats minus hits). I
chose the higher of the two results to be used for the average runs per game of
25.5 outs made by Oh, and also to calculate the number of runs per win. In this case, it was the plate appearances
mark of 3.94 runs per game per team. For
batting runs above average, we take the 1982.3 batting runs created and
subtract (3.94 runs per game times 296 games), which comes out to 814.5 batting
runs above average. If there’s a small
rounding difference, it’s because I didn’t do rounding on the spreadsheet I
used to calculate this figure, but in reporting the figures I used, I needed to
report rounded figures.
Next, the baseball reference WAA has runs above or below
average in baserunning and double plays.
They’re two separate categories in their calculation, but I was able to
run a regression of the sum of these two figures by major leaguers with at
least 5000 plate appearances after 1954 (when we first have gidp data) that
seemed to yield better results than doing them separately. I used
three categories in the regression:
triples divided by (hits minus home runs), triples and stolen
bases. The least predictive p value of
these three figures was less than 0.00049, the r-squared result was over .698, and the standard error below
17.5. The coefficients for the
categories were 108.4542 for the triples divided by (hits minus home runs),
0.288497 for the triples, and 0.134134 for the stolen bases, with an intercept
of -30.9337. For clarity, the formula
is:
108.4542 * (3B/ (H – HR)) +
0.288497 * 3B + 0.134134 * SB -30.9337
Using the figures from Oh’s
projection and his 84 career stolen bases, we get a figure of -6.3.
The next figure in the baseball-reference WAA calculation
was the hardest one to resolve. It was
the fielding runs above or below average for the player’s position. With the limited data available for Oh and
other NPB players, it was hard to come up with a decent calculation for this
category. It wasn’t until in desperation
I tried using gold gloves by position that I got even a modestly successful
result. The coeffieints were:
11.73626 per gold glove at
catcher
11.26132 per gold glove at
first base
7.581625 per gold glove at
second base
18.98308 per gold glove at
third base
14.67945 per gold glove at
shortstop and
11.45049 per gold glove in
the outfield
with an intercept of -7.66891
The least predictive p value
was at first base, at 0.000107, the
r-squared value was just over .265, and the standard error just under
56.7. It’s pretty rough, but it is
better than assuming everyone is average, which was about the only alternative
I saw. I’m plugging in NPB Diamond
Gloves for MLB Gold Gloves. Oh won 9
Diamond Gloves in his last nine seasons, the only ones awarded during his
career. They all came at first, so the
calculation puts him at a very favorable mark for a first baseman of +93.7.
The last category is the position adjustment, which
Baseball-Reference.com gives as -9 runs per 150 games times 9 innings. Oh did have two games in the outfield, but in
a season where he played 130 games, and played 130 at first. I chose to disregard these two apparently cameo
appearances in the outfield. He didn’t
play in the field in 32 games in which he appeared, and 16 of those were before
1962, when I have him starting his major league career. That means he missed 16 games, and I’ll bump
it up by the league adjustment factor and round up that figure to 20
games. He has more than 4.2 plate
appearances per defensive game, so I’m going to assume each defensive game was
nine innings. Since I’m assuming each
game is 9 innings, I can just use -9 runs per 150 games. Deducting 20 games from his projection, I
have 2975/150 * (-9) for a result of -178.5 runs for the positional adjustment.
To get our runs above average, we take the 814.5 batting
runs above average and subtract the 6.5 runs below average for the combination
of baserunning and grounding into double plays then add the 93.7 runs above
average fielding at first base and subtract 178.5 runs for the first base
position adjustment and get 723.4 runs above average. To turn this into wins above average, we need
to divide this figure by the runs per win figure. The easiest way to calculate the runs per win
is a calculation used by Tom Tango, which is 1.5 times the average runs per
game of 3.94 we already calculated plus 3, or 8.92 runs per win. Dividing the 723.4 runs above average by that 8.92 runs per win figure yields 81.1
wins above average. In order to go from
WAA to WAR, we first need to add the runs
above replacement to the 723.4 runs above average figure. We calculate the runs above replacement at
20.5 runs per 600 plate appearances, which we’ll approximate by at bats plus
walks. We have 12583 plate appearances
divided by 600 times 20.5, or 429.9 runs.
That means he has 1153.3 runs above replacement, and at 8.92 runs per
win, he has 129.3 WAR
Those WAA and WAR marks are
spectacular, as I will make clear in a moment.
Here are the position players with over 70 WAA:
Player |
WAA/pos |
Babe Ruth |
125.6 |
Barry Bonds |
123.6 |
Willie Mays |
110.4 |
Ty Cobb |
101.7 |
R Hornsby |
97.5 |
Ted
Williams |
94.2 |
Hank Aaron |
92.4 |
Honus
Wagner |
91.9 |
Tris
Speaker |
88.2 |
Stan Musial |
81.4 |
Sadaharu
OH |
81.1 |
Mickey
Mantle |
79.0 |
Eddie
Collins |
78.9 |
Lou Gehrig |
78.2 |
Alex
Rodriguez |
75.9 |
Mike
Schmidt |
73.3 |
Now the guys with over 100
WAR:
Player |
WAR/pos |
Babe Ruth |
163.1 |
Barry Bonds |
162.5 |
Willie Mays |
155.9 |
Ty Cobb |
151.1 |
Hank Aaron |
142.5 |
Tris
Speaker |
133.8 |
Honus
Wagner |
130.9 |
Sadaharu
OH |
129.3 |
Stan Musial |
128.2 |
Rogers
Hornsby |
126.9 |
Eddie
Collins |
124.0 |
Ted
Williams |
123.2 |
Alex
Rodriguez |
117.7 |
Lou Gehrig |
112.3 |
R Henderson |
110.7 |
Mickey Mantle |
109.6 |
Mel Ott |
107.6 |
Frank
Robinson |
107.4 |
Nap Lajoie |
107.4 |
Mike
Schmidt |
106.6 |
Joe Morgan |
100.4 |
Oh’s estimated WAA and WAR
are placing him among the elite players ever to play the game.
Let’s look at first basemen
with over 50 WAA:
Player |
WAA/pos |
Sadaharu
OH |
81.1 |
Lou Gehrig |
78.2 |
Albert
Pujols* |
64.9 |
Jimmie Foxx |
62.7 |
Cap Anson |
55.4 |
Dan
Brouthers |
54.9 |
Roger
Connor |
54.2 |
Jeff
Bagwell |
51.9 |
Pujols has the asterisk (*)
because he’s active.
Now the first basemen with
over 75 WAR:
Player |
WAR/pos |
Sadaharu
OH |
129.3 |
Lou Gehrig |
112.3 |
Albert
Pujols* |
99.6 |
Jimmie Foxx |
96.3 |
Cap Anson |
94.0 |
Roger
Connor |
84.2 |
Jeff
Bagwell |
79.8 |
D Brouthers |
79.5 |
Believe it or not, I’m
comfortable with ranking Oh over Gehrig and every other first baseman on a career
basis, as Pujols is about 1500 plate appearances below Oh’s projection at this
point, Anson 1200 below, and Gehrig and Foxx over 2600 each below. Oh was a notch below them per game, but he
played so well that giving him that much more playing time allows him to pass
them all. Oh probably belongs below
Gehrig on peak, as he only inches ahead in WAA on significantly more playing
time. I’m not sure there is a consensus
on even whether we should try to combine peak and career to determine who the
greatest is. Even if there is, there
certainly is no consensus on how it should be done. That, plus the fact that the Oh projection is
an estimate of his major league value, albeit one I contend is reasonable but
on the conservative side, means it’s hard to definitively say Oh is the
greatest first baseman of all time. I do
think he belongs in the discussion for that honor, however. I am comfortable in saying he’s at least the
second best first baseman of all time, as he’s got enough room to cover those
issues and stay ahead of Foxx and Pujols
A. Pitchers in
Japan are the real stars/ Was Nagashima greater?
One issue which was
raised is that pitchers have been the bigger stars in Japan, and therefore
shouldn’t one of them precede Oh into Cooperstown. The point about pitchers being the bigger
stars is arguable, but in terms of who should go to Cooperstown, it is
irrelevant. If we ever get around to
honoring players solely for their play in Japan, we should start with the very
best in that group and work our way down.
No one can rival Oh as the greatest player in Japan.
Some English-speaking
writers have written that the Japanese public regards Nagashima as the greatest
player in Japanese baseball history. I
cannot say whether or not this accurately reflects the sentiments of the
Japanese public. It has been suggested
that Nagashima is regarded as the best “Japanese” player, Oh being excluded
because his father is Chinese. Another
suggestion, courtesy of Josh Reyer, is that while Oh’s statistics were regarded
as better, Nagashima was seen as more “clutch” because he came up with more
memorable hits, homers, and defensive plays.
Fred Ivor-Campbell indicates that this perception lasted most if not all
of Nagashima’s career, but that Oh emerged from Nagashima’s shadow when the
latter retired. That seems reasonable,
because much of Oh’s advantage in career numbers came once Nagashima
retired. The fact Oh could sustain that
high level of performance for many more years would make his superiority as a
player clear. Certainly, Oh’s statistics
are far superior. Oh and Nagashima were
teammates for 15 years, and Oh has 645 more games, 1156 more at bats, 697 more
runs, 315 more hits, 424 more homers, 648 more RBI, 1421 more walks, and 4 more
MVP awards. Any reasonable
interpretation of the record clearly shows Oh to be the greater player. It does seem to be true that Nagashima was
more popular. Oh’s Chinese heritage may
or may not be one factor. It also seems
that Nagashima was much more outgoing and willing to show his emotions on the
field, while Oh rarely showed emotion and was generally reserved. Fans have always preferred outgoing guys who
show their emotion to reserved guys who don’t, and the Japanese fans seem to
have the same preference.
B. The “National” Hall of Fame
The last argument
against Oh we will address is the argument that Cooperstown is the National Hall of Fame and is limited to
those who have contributed to the game in North America. First of all, no one in the debate has yet cited
anything beyond the name of the institution as proof there is any formal
restriction on who the Hall of Fame may honor.
Second, even if such a restriction exists, it certainly can be changed
about as easily and rapidly as the
sudden decision to allow Negro Leaguers to be honored on an equal basis with
the major leaguers. Third, the Hall
should honor all the best players in the game, no matter where they played
or who they played against, because they all have helped to make it the great
game it is. Fourth, the game is becoming
increasingly international in scope. In
2002, nearly a quarter of the major leaguers were born outside the 50 states. Seventeen different countries were represented
in the majors, and a total of 31 in the minors.
About half of all minor leaguers were born outside the 50 states. We now have major league all-stars from the
Orient, and undoubtedly we will have many more.
Ichiro is almost certain to be voted into the Hall of Fame when he
becomes eligible. We even allow those outside
North America to vote for the major league all-star teams. Under such circumstances, the “National”
argument seems to me to be hopelessly parochial and possibly even
self-defeating. It certainly looks
hypocritical to promote diversity on one hand while denying the game’s highest
honor to foreigners who have been subjected to a de facto bar nearly as
sacrosanct as the color line was before Jackie Robinson. Even honoring the players in Japanese
baseball history who are worthy of Cooperstown seems to be inadequate
compensation for siphoning off at least some of Japan’s elite players. Maybe the Japanese wouldn’t have come even if
they were given a realistic opportunity to do so, but to deny them plaques in
Cooperstown solely on such speculative reasoning is plainly ridiculous.
Furthermore, Oh has
had a tremendous influence on Japanese baseball as its greatest player, as its
goodwill ambassador, and as a successful manager. He came into contact with many major
leaguers, and his career has touched present day major league managers like Jim
Tracy, Davey Johnson, Charlie Manuel and Bobby Valentine. Isn’t it likely Ichiro learned something
from Oh, whether as a youngster or as an opponent of Oh’s teams, or
some other way? Oh’s influence upon
major league baseball may be small today, but that influence will almost surely
grow with the increased influx of Japanese players. Also, listen to Steve Garvey: “ I learned a lot . . . from Sadaharu
Oh. I spent some time with him in spring
training in 1971, and again in ’75 and ’79.
He always talked about the use of his legs as the single biggest asset
to his power . . . . You’ve got to use
your whole body to hit the ball effectively, not just your arms. That’s the difference between a power hitter
and a slap hitter.”
The “National”
argument is at best a dinosaur doomed to extinction by the strong existing
trend toward international growth in the game. Eventually, I believe MLB will
have a permanent presence in Japan in some form, and at that point, it will
need to please its Japanese fans. When
that occurs, the “National” argument will surely fall. It may hold sway until that time, but it is
only staving off its eventual losing fate.
V.
Conclusion
There is a large body
of evidence available on Oh’s worthiness for the Hall of Fame, and it strongly
indicates that he is a very worthy candidate.
All the arguments against his candidacy either do not hold water or are
simply overwhelmed by the mountain of evidence in his favor. For all the reasons set forth in this review
of the evidence, he richly deserves a
plaque in Cooperstown, and it is likely that some day there will be such a
plaque. However, there is one more thing
we should consider: Oh is 77 now, and
although he is in apparent good health despite having been treated for cancer
several years ago, the time in which he can personally enjoy the honor he so
eminently deserves is limited. Those of
you who are convinced by the evidence presented by this analysis should do what
you can to see that Oh is honored while he can still enjoy it. It is the right thing to do.