Rob Neyer is, of course, my very good friend of many years standing.  My son Reuben is getting married in about 15 hours (that was Friday evening.  He was married on Saturday.)  The day Reuben was born, Rob was the first person to hold him, other than my wife and myself and the medical professionals.  He wrote a marvelous book in 2017, Power Ball, which won the Casey Award given to the best baseball book of the year.   The book is 99.9999% true, but there is one sentence in there which I strongly disagree with, so of course I am about to write a 20-page article here to demonstrate that that one sentence. . . .OK, it is two sentences. . . to demonstrate that those two sentences from page 26 should not be believed. 

          Contrary to popular opinion, batting average was not valued so highly by the market.  Most baseball executives did know an "empty" batting average when they saw one. 

            Contrary to Neyer’s opinion, batting average completely dominated the market, and most baseball executives into the mid-1990s didn’t have the foggiest notion of the difference between an empty batting average and a productive hitter.   And you couldn’t explain it to them, because they didn’t understand the supporting concepts. 

Context places Neyer’s characterization as somewhere between 1950 and 2002, with talk on one end about Ralph Kiner and, on the other end, the description "before Moneyball."  But I got to wondering. . .is that really true?   DID people in the pre-sabermetric era have a clear understanding of the difference between an empty batting average and a truly productive hitter?

A lot of what I have believed over the years has turned out on investigation to be not true; perhaps the same should be expected of what my friends believe.  In any case, I woke up on Tuesday morning, October 6, with a fully-formed idea for a study racing through my head.   That happens to me two or three times a year; I do not know whether my subconscious actually forms these studies while I am sleeping or whether I have formed them before but just not been aware of it, but in any case once every few months that happens to me. 

I thought of a straightforward way to test, if not this theory perfectly, at least a closely related concept.  Before I get to that. . . .I think that I may have invented or at least popularized the expression "an empty batting average".   I could be wrong; you might study it and find that the phrase was in common use before me, or, more likely, that it was occasionally used before me.  But I think I created that one.  Doesn’t matter.

            Anyway, here is the approach.  Suppose that we take all players in the era 1950 to 1975 who have either 15 Win Shares or 2.5 WAR in a season.   15 Win Shares and 2.5 WAR (Baseball Reference WAR) are about the same thing; there are not a lot of players who have one but not the other, and also, they represent about the bottom of the barrel for players drawing meaningful support in MVP voting, which is what I am going to be studying here.   Take all players with 15 Win Shares or 2.5 WAR in the last quarter-century of the pre-sabermetric era. 

            Then we look up, for each player (a) his batting average, and (b) his secondary average.  Then we can sort the players into three groups:

1)     Higher batting average than secondary average (the empty average group),

2)     Essentially the same batting average as secondary average, and

3)     Higher secondary average than batting average, which will call the low-average productive group.

I will note that you could also do the study by contrasting HITS with (RUNS + RBI).  A player who has a significantly higher batting average than secondary average will have a more hits than runs + RBI at least 80% of the time, and vice versa.   If you did the study that way, you’d get essentially identical results to doing it this way. 

Going back to the phrase under dispute:

Contrary to popular opinion, batting average was not valued so highly by the market.  Most baseball executives did know an "empty" batting average when they saw one. 

That contains a flexible word there, valued "so" highly.  Well. . .how highly was it valued?   Neyer could be saying that people think that the evaluation of a player rested 90% on batting average, 10% on everything else, but really, it was only 80% on batting average, so, you see, it was not valued so highly as people think.  

Language is never precise.   The evaluation of players pre-Moneyball was not 90% based on batting average, or 80%.  It was about 45%.   It was about 45% based on batting average, 30% on RBI, 15% on fielding, and 10% on everything else.  What I mean by that is that, if you had comprehensive salary data from before 1975 and you wanted to predict salaries based on statistical performance, you would have to use batting average as about 45% of the method, and RBI as about 30%, with little space for anything else from the stat room.  

Also, we should note that there could be a difference between the way baseball executives evaluated players and how MVP voters evaluated players.   We’re actually studying MVP voters here, not baseball executives.  It could be true that baseball executives in that era understood the difference between productive and unproductive hitters, but MVP voters didn’t.  I doubt it, but. . .could be, and in any case Rob doesn’t give us any source for his "most baseball executives did know that. . " claim.  I am certain, knowing Rob, that he had some basis for saying that, since Rob likes doing baseball research more than he likes almost anything else in life, but I don’t know what the basis was, because if he explained that, it was in a footnote, and I don’t do footnotes.   

 I rounded up all the necessary data for the 1950 and 1951 seasons, which took me about 9 hours of work.   Getting a little bit more efficient, I realized that I could assemble and process the data at a rate of about four teams an hour, which would mean that to complete the study by the process first imagined would take me a mere 150 hours.  Also, it would give me absurdly more data than was necessary to do a persuasive study of the issue. 

So I didn’t do that.   I cut the study group back from 1950 to 1975 (original plan) to 1950 to 1953 (actually accomplished.)  You can decide for yourself how persuasive you think this is. 

From 1950 to 1953 there were 269 players who qualified for our study—that is, they had either 15 Win Shares or 2.5 WAR, or both.  Of those 269 players:

198, or 74%, met both standards,

58 had 15 Win Shares, but did not have 2.5 WAR, and

13 had 2.5 WAR, but did not have 15 Win Shares.

So 2.5 WAR is a little bit higher standard than 15 Win Shares, I guess; you live and learn.  Anyway, I pulled out the Win Shares book to start building the list of players with 15 or more Win Shares.   On the very first team that I looked at, the 1950 Boston Red Sox, there is a very dramatic counter-example (counter to Neyer’s point; consistent with mine.)  Billy Goodman on that team hit for a .354 batting average, but with a .229 secondary average, putting him in Group 1.  He finished second in the MVP voting, getting 180 points and drawing a 54% vote share. 

Goodman had 16 Win Shares and 2.6 WAR.  He had no less than SEVEN teammates who had (a) more Win Shares, and (b) a higher WAR.   All seven had lower batting averages than Goodman, but higher secondary averages.  All seven of them had more Runs Scored + RBI than Goodman.   Among the seven of them, they had 120 points in MVP voting—60 less than Goodman had by himself.

Of course, a sample of one doesn’t prove anything.  Goodman had special traits.  He was the Ben Zobrist of his time, playing all over the field, filling in wherever he was needed today.   There is value in that, and his MVP vote can be seen as a recognition of that value. 

But going back to the data, we don’t have to go very far to find another almost equally dramatic example.  The Cleveland Indians had two stars from group 3—Larry Doby and Al Rosen.  Doby hit .326 but with a .429 secondary average; Rosen hit .287 but with a .446 secondary average.  Both had MVP-level totals of Win Shares (30 and 29) and WAR (6.7 and 5.8).  Doby was 0.1 away from having the highest WAR in the league for a position player.  They finished 8^th and 17^th in the MVP voting.   

This, however, could plausibly be explained by racial or ethnic prejudice (Doby was black, Rosen Jewish), and also, there are other ways to interpret the data, so let’s move on.  That was not the almost-equally dramatic example I promised you.   That came on the fourth team on the list, the Detroit Tigers. 

The 1950 Detroit Tigers were a surprise team, a team that came out of nowhere and almost won the American League, winning 95 games (and then moved back to nowheresville.)  Three players tied for the team lead in Win Shares—George Kell, Hoot Evers and Vic Wertz, 26 Win Shares each.   Kell hit .340 with a secondary average of .251, which would put him in Group 1, the empty-average group.   He was fourth in the MVP voting, with 127 points.  Evers and Wertz would both be Group 3 in our study, hitting .323 and .308 respectively but with Secondary Averages of .373 and .388.   While both were mentioned in MVP voting, they had 38 and 50 points.  Between them, they had 39 fewer points in MVP voting than Kell had by himself. 

Again, there are many other possible explanations for this.  Kell was a very good defensive player, and he drove in 101 runs, which is unusual for a player with only 8 home runs.  Kell had the same Win Shares as Wertz and Evers, but slightly more WAR (4.9 vs. 4.6 and 4.3).  Kell had essentially as many Runs Scored + RBI as Wertz and Evers did, although he did this while making many more outs.  A sample of one doesn’t prove anything, and a sample of two doesn’t prove anything, either; only a systematic study of the data should convince us. 

In 1951 teammates Kell and Wertz were tied in Win Shares again, 22 each, but this time Wertz had more WAR, 4.4 to 4.1.  Kell still pulled in much more support in MVP voting—30 points for Kell (a 9% share) and zero for Wertz.  

In 1950 the AL batting champion was Billy Goodman.  In 1951 it was Ferris Fain.  Fain had 19 Win Shares, 4.6 WAR.  A teammate, Eddie Joost, had 25 Win Shares and 6.1 WAR. 

In secondary average, Joost beat Fain, .382 to .315.

In runs plus RBI, Eddie Joost beat his teammate Ferris Fain by the slim margin of 185 to 120.  

Joost was the shortstop on the team; Fain was the first baseman. 

Fain, however, beat Joost in batting average, .344 to .289; Joost was better than Fain in basically everything else, but Fain beat Joost in batting average—and therefore, beat him by a wide margin in MVP voting, 103 to 32. 

In 1950 Larry Doby was 3^rd in the American League in WAR, but 8^th in MVP voting.  That was better than 1951, when Doby had 29 Win Shares again and 6.4 WAR, second-highest of any position player in the American League, and was not mentioned in MVP voting. 

In 1951 the top three position players on the Cleveland Indians were Larry Doby, Al Rosen and Bobby Avila.   Larry Doby had 29 Win Shares, a .295 average, 20 home runs and 101 walks, and was not mentioned in MVP voting.   When I tweeted about that, people explained it away by saying that Doby drove in only 69 runs, which is true and relevant, but Al Rosen had 25 Win Shares, a .265 average but 24 homers, 85 walks and drove in 102 runs.  He wasn’t mentioned in MVP voting, either. 

Teammate Bobby Avila hit half as many homers (10) and drew significantly fewer walks than either Rosen or Doby (60).  He drove in and scored fewer runs than either Rosen or Doby—but he was the only .300 hitter in the Cleveland lineup.  He drew 49 points in MVP voting, while Rosen and Doby had none.  

Understanding that these are still episodic, rather than systematic. . . .

In 1952 there were three White Sox who were good enough players to qualify for the study:  Minnie Minoso, Eddie Robinson and Nellie Fox. 

Fox, who had 3.3 WAR and was in group 1, the empty average group (.296 batting .130 secondary average) drew 59 points in the voting.

Robinson, who had 3.7 WAR and essentially the same batting average as secondary average, drew 47 points in the voting.

Minoso, who had 4.1 WAR but a secondary average 25 points higher than his batting average, was not mentioned in MVP voting. 

OK, I’d better get to the systematic data.  Here’s one.  In my data, 269 player-seasons, there are 26 players who hit .300, but with a secondary average below .250.  Here’s the list:

That’s the empty-average group, although it should be noted that Richie Ashburn was actually a valuable player—more valuable than anyone else in this group.  Over half of those players are outfielders or first basemen, by the way, which I point out in case somebody tries to argue that the players with low secondary average were middle infielders whose value was in fielding.  Anyway, of those 26 players, 18 were mentioned in MVP voting for that season, and nine drew more than 20 points in MVP voting (23 or more.)

On the other end there are, by fortunate coincidence, 26 players in the study who had batting averages of .265 or lower, but secondary averages of .300 or higher.   In terms of the quality of their play, there is minimal difference between the two groups.  The high-batting-average group had 533 Win Shares among them, 95.3 WAR.  The low-batting average group had 527 Win Shares, 90.3 WAR.  

But of the low-average group, only ten were mentioned at all in MVP voting (as opposed to 18 in the other group), and only one had more than 20 points in the voting (as opposed to nine.)   The one lower-average player who did draw a little MVP support (40 points) was Luke Easter, who had driven in 97 runs in 127 games; he was followed by Gus Zernial, 1952, who had driven in 100.  A low-average, high secondary-average player can attract some MVP support if he drives in runs; that was the one other thing that really counted.    But despite the fact that there were several more players in the low-average group who had driven in 100 runs or very close to that (Hank Sauer, Gil Hodges, Ralph Kiner, the first Frank Thomas). . .despite those RBI guys, the singles hitters received more than four times as much MVP support as the low-average hitters got, based on essentially equal performance.   The Group 1 hitters received 538 points in MVP voting; the Group 3 hitters, 131.  

On a more systematic level. . . I sorted all 269 players into 9 groups by the number of Win Shares, like this:

So, trying not to confuse you, we now have 27 groups of players, 27 "cohorts", with the players sorted 9 ways by the value of their performance, and 3 ways by the type of player they were.  There are about ten players in each of the 27 groups.  Group 5-1 is high-average singles hitters with 22 or 23 Win Shares; Group 5-3 is lower-average, more power hitters with 22 or 23 Win Shares. 

Then, for each LEVEL of performance, I figured the average number of MVP votes for all players at that level.   There are 34 players in the study who had 22 or 23 Win Shares (Level 5).  As a group, the 34 players at Level 5 received 685 points in MVP voting, or 20.15 points per player, so 20.15 points is established as the norm for a player of that quality (or value). 

Ten of those 34 players were Type 1, meaning high-average singles hitters.  We thus have an expectation that those 10 players would receive 201.5 points in MVP voting—ten, times 20.15.   The number of players in the group, times the expected number of MVP vote-points for a player of that quality.   In fact, they received 257 points in MVP voting, 55.5 more than expected.  They over-performed in MVP voting by 28%, and we’ll also record that as +55.5.

This chart summarizes the average number of MVP vote points for each Level of Performance.  As you can see, it makes pretty much perfect sense; the higher the level of performance, the more votes you draw in MVP voting:

In a moment, I’ll show you a parallel chart for WAR.  It doesn’t work quite as beautifully, but it is similar.

There were 11 Level-5 players who were Type 2, meaning that their batting averages and secondary averages were about the same.   A typical player in that range would be Willie (Puddin’ Head) Jones in 1950, or Willie Jones in 1951: 

Jones his .267 with a .297 secondary average in 1950, .285 with a .301 secondary average in 1951.  He was a good, valuable player with an even mix of skills, but not a serious MVP candidate; he was in cohort 5-2 both years.

The 11 players in Cohort 5-2 could have been expected to receive 221.65 points in MVP voting, or 11 times 20.15.   In fact, they received 228 points in MVP voting, or 6.35 more than expected.  We’ll record that as +6.35 for Cohort 5-2.

There were 13 players in the study in Cohort 5-3; that is, Level Five players with higher secondary averages than batting averages.   Those 13 players could have been expected to receive 261.9 points in MVP voting, or 13 times 20.15.  In fact, they received only 200 points, which we record as negative 61.9.   The low-average players underperformed in MVP voting, relative to their actual value as measured by Win Shares.

They usually do.   They normally underperform; that is the conclusion of the study.  This chart summarizes all of the data like that from the study:

What that means, if you were to take the time to study those charts, is that the 90 players in the study who had notably higher batting averages than secondary averages over-performed their MVP voting expectations, based on the quality of their play, by 488 points, or 35%.    Those whose batting averages fairly reflected their offensive value over-performed in voting by 213 points, or 11%.  Those who were batting-average impaired underperformed their actual value in MVP voting by 701 points, or 17.5%.  

And what that means is that given a certain level of value, a high-average hitter will outperform a low-average hitter of the same quality in MVP voting by 64%.   135, divided by 82.5, is 1.64.

We can do, and I have done, the same study, but using WAR as the measure of "true" value, rather than Win Shares.   This is the "levels" chart for that data.  

Level 5 performance in the Win Shares Study means 22 or 23 Win Shares.  Level 5 performance in WAR means 3.6 to 3.9 WAR.  I use both methods as a check to make sure that I am not simply validating my own expectations, and also because I don’t want to argue about which method I should have used.  Level 9 performance in the Win Shares study means 30 or more Win Shares.  Level 9 performance in the WAR study means 6.3 WAR or higher. 

Players at Level 9 in Win Shares averaged 139 points in MVP voting.   Players at Level 9 in WAR averaged 110 points in MVP voting:

The totals are lower at the top end in this portion of the study because, in this study group, Win Shares correlates more strongly with MVP vote performance than WAR does.   We can’t generalize that; we don’t know whether that is generally true or not.

Anyway, I then repeated all of the same steps I had used to measure the "singles hitters’ preference" in the Win Shares study, only based on WAR.   I won’t report all of the steps, because I don’t know whether that has any value to you, but the 90 players who were identified as Type 1, higher batting average than secondary average, had an expectation in this part of the study of 1,668 MVP voting points.  They actually received 1,866 (the same number reported earlier), which is, in this study, 11% above expectations.  In the other study, it was 35% above. 

The Type-2 players in this study (Willie Jones) had an expectation of 1,997 MVP vote points, and actually received 2,190 (as before), so they also over-achieved by 10%, whereas in the other part of the study they had over-achieved by 11%.   And the Type-3 players in this study, the players with walks and homers instead of high averages, under-achieved in this part of the study by 10.6%, whereas in the other part of the study it was 17.5%.   Whereas singles hitters compared to "productive" hitters were overvalued by 64% in the Win Shares study, they are overvalued by only 25% in WAR study.  The lower values, again, are due to the weaker correlation of WAR with MVP voting.  

So far, I think we have demonstrated that players with "empty" batting averages were significantly over-valued in MVP voting in the 1950 to 1953 era.   That is, of course, a substantial distance from disproving Neyer’s claim, which was that productive vs. empty batting averages were understood by executives at some point in the space between Ralph Kiner and Moneyball. 

Transitioning now to talk about money, because salaries are more clearly a statement of the belief system of executives of that time, we go back to George Kell.  You may assure me that Home Run hitters drove Cadillacs, but here’s a data point for you.  

George Kell made his major league debut on September 28, 1943; Andy Pafko, on September 24, 1943.  Both men were able to play through the war—WWII—so through 1950 both men were 7+ in terms of experience.   Kell had played 943 major league games; Pafko, 911.   Through 1949, both men had earned 104 Win Shares, or each man had earned 104 Win Shares, whichever is correct there.  In 1950 Pafko earned 27 Win Shares, Kell 26, which put Pafko ahead, 131 to 130.   In WAR, Pafko was ahead by a much wider margin, 25.8 to 21.2.  

In 1950 Kell hit .340 with 8 homers, 26 Win Shares.   Pafko hit .304 with 36 homers, 27 Win Shares.

In 1951, according to Baseball Reference, Kell earned $42,000.  Pafko earned $29,000.

Matches like that are hard to make, in that era, because only about 20% of salaries from the 1950s are reported in Baseball Reference.   You have to have TWO salaries to make a match, so that’s 4%.   If you can find two players who match and illustrate the point, you’re only going to be able to find the salary data to pay off the work 4% of the time.  But here’s something.     

George Kell was in the major leagues from 1943 to 1957.  A high-average hitter, he earned 37.6 WAR, and 229 Win Shares. 

            Vern Stephens was in the major leagues from 1941 to 1955.  He earned 46 WAR, and 265 Win Shares.   Stephens, of course, hit for a much lower average than Kell, but with much more power.   Stephens drove in and scored far more runs than Kell, and Stephens was a shortstop, while Kell was a third baseman. 

            Stephens in his career earned, according to baseball reference, $291,300. Kell earned $426,600.  Per unit of WAR, Kell earned 69% more.  Per Win Share, he earned 79% more. 

            That line of approach—and I agree that Kell is only one player, so it’s a sample of one and a half—but that line of approach attacks the possible argument that baseball writers may have rested their evaluations heavily on batting average, but baseball executives knew better.   

Here is an interesting one.  Bob Allison came to the majors as a late-season callup in 1958 and was a rookie in 1959, actually Rookie of the Year.  Tommy Davis came to the majors as a late-season callup in 1959, and was a rookie in 1960, so he was always one service year behind Allison.  (He was not Rookie of the Year; a teammate was.  More on that in a moment.)

            While Davis did hit 27 home runs in 1962, he was essentially a high-average hitter.  He never hit more than 16 home runs in any other season, but won the National League batting titles in 1962 and 1963.  He hit .300 four times before 1967--.346 in 1962, .326 in 1963, .313 in 1966, and .302 in 1967. 

            Allison never hit .300, and had an average higher than .271 only once in his career, but had much more power than Davis, hitting 29 to 35 home runs every year except one from 1959 to 1964, then slipping mostly to the low 20s.  He also walked much more often than Davis, and had more defensive value.   (Davis was a very limited defensive outfielder who was tried unsuccessfully at third base.   Allison was basically a right fielder, and played about 200 games in center.)  But all things considered—power, average, walks and defense—Allison was a significantly better player.  By Win Shares, Allison was the better player almost every season:

            That’s by Win Shares; Allison leads Davis 201-149.  By WAR, Allison’s advantage was much larger than that.   While Win Shares credits Davis with a high peak in 1962-1963, WAR believes that Allison’s best seasons were better than Davis’, and that Allison overall was twice as valuable as Davis through the same seasons.

            We need not quarrel about whether Allison was twice as valuable as Davis, as WAR argues, or merely 35-36% more valuable, as argued by Win Shares.  Both Allison and Davis had one serious injury, Davis in 1965, Allison in 1966.  I will add to this information that while Allison stayed with his original organization until the end of his career and kept his job almost to the end, Davis by 1969 had been dumped by the Dodgers, the White Sox and the Mets, and was playing for the Seattle Pilots, a first-year expansion team.

            And yet, incongruously, Davis by 1964 had exploded past Allison in annual salary—and remained far ahead of him until the end of Allison’s career:

            Why did that happen?  Perhaps it was only that Allison was a poor contract negotiator.   But I remember those years, as if they were yesterday, and I remember that Tommy Davis was a STAR.   Allison was not.   Davis earned more money—a LOT more money—because, in that era, hitting for average was what it about. Allison was a .260 hitter, and .260 hitters just didn’t make big money.

            I noted that Davis did not win the Rookie of the Year Award in 1960, a teammate did.  The teammate was Frank Howard.  Howard, like Allison, was a lower-average hitter, but Howard had a LOT of power.   Like Tommy Davis, Howard had a couple of good years for the Dodgers then had a couple of years of injuries and slumps, and the Dodgers moved him out.   Unlike Davis, who bounced around after the trade, Howard after the trade became the Washington Senators biggest star.   By Win Shares or WAR, Howard had more value than Davis through 1966 or 1967 or 1968.  Davis still earned far higher salaries.   It wasn’t until 1969, after Howard hit 44 homers and drove in 106 runs in 1968 (in a pitchers’ park, in the worst year for hitters in the last 100 years). . .it wasn’t until 1969 that Howard’s salary moved beyond Tommy Davis’.

            Or how about this one:  Matty Alou, and Dick McAuliffe.  You talk about an empty batting average, Matty Alou was pretty much the poster boy.  He had a career average of .307, and won the NL batting title in 1966.  But he had near-zero power, didn’t walk, and was a break-even percentage base stealer. 

            McAuliffe was the exact opposite of that.  He had a career batting average of .247—60 points lower than Alou—but he hit 22, 23, 24 homers in a season, and twice drew 100 walks in a season.  His career on base percentage was only two points lower than Alou’s; his slugging percentage was 22 points higher.  And he was a middle infielder. 

            We have salary data for McAuliffe in only one season, 1972.  He made $50,000.   Matty Alou, who was a year older than McAuliffe, 33 to 32, made $100,000. 

            Or here’s one. . .I apologize for the spotty/selective nature of the comparisons, but this is just the data that we happen to have.  Two third basemen, both born in 1951:  Buddy Bell, and Bill Madlock.   Madlock won four NL batting titles, and had a career batting average of .305.  Bell hit .279, which is a good batting average, but it ain’t .305 and he never won a batting title.

            Nonetheless, Buddy crushes Madlock in the career value stats, 66.3 to 38.2 in WAR, and 301 to 242 in Win Shares.  We have salaries for both players in three seasons—1985, 1986, 1987.  Would you want to guess who earned more money?       Of course he did.  Madlock, $2.67 million; Bell, $2.49 million. 

            Or how about this one:  Ron Cey and Steve Garvey.  Same team, right?  Long-time teammates.   Steve Garvey was slated to be the Dodgers’ third baseman of the future.  Ron Cey took that job away from him, pushing Garvey to first base.

            Steve Garvey had a career average of .294; Cey, only .261.  Cey, however, had an on base percentage a whopping 25 points higher than Garvey’s, .354 to .329, and, because Cey packed more power into his .261 average than Garvey did into .294, essentially the same slugging percentage, Garvey being one point higher.  In terms of WAR, Cey leads 53.8 to 38.1.   In Win Shares, Cey leads only 280 to 279, but that is in 259 fewer games. 

            We have salary data for both of them in six seasons:

            I was working in baseball salary negotiations at the end of that period, and I vividly remember trying to convince baseball professionals (agents) that Steve Garvey was really not a great hitter.  The reactions ranged from skepticism to scorn.  An agent with whom I was working explained patiently that Garvey was a fearless hitter.   An offense was built around that fearless hitter who could drive in runs.  

            That’s my personal experience, and my personal experience is really the reason that I feel it is important that the statement should not go unchallenged that executives before Moneyball understood the difference between a productive hitter and an empty batting average.   In the years that I worked for the Boston Red Sox (2002 to 2019), I’ll bet that we had 15 to 20 former General Managers who worked in our office.  Theo was young and inexperienced when he came into the position; he surrounded himself with people who had been there and done that.  Lee Thomas, Mike Port, Lou Gorman, many others.  (It may actually have been the owners who brought these people in, I don’t know.)  Anyway, I know those men.  I have sat and talked to them, some of them for hours.  I know how they thought; I know what they understood and what they didn’t understand.   

            Bill Lajoie.  Bill’s gone now, been gone for years.   Bill was a crusty old fart and sometimes acted like a child, but I genuinely liked him and tremendously respected his judgment.  In his first year as a scout, Bill recommended Johnny Bench to the Reds, although the honor of signing Bench went to a veteran scout, which still irritated Bill 40 years later.  Bill had worked for the Reds in the 1960s and 1970s, the Tigers in the 1970s and 1980s, the Atlanta Braves in 1990s, and for us—the Red Sox—in the early 2000s.  He had more World Series rings than he had fingers—not all of them winners, but you get a ring if you get into that seven-game show. 

            Bill got off a lot of good lines; maybe you had to be there.  One time he and I were leaving Fenway about 4:45 in the morning after a long game followed by a long post-game discussion session, and I asked him "You ever leave the park at this time of morning?" 

            "Oh, yeah," Bill said in his raspy growl.  "Back when I was drinkin’, I did this a lot of times."

            I never kidded myself that Bill had much use for my opinions, but we got along very well, for a series of odd reasons.  One is that we were the only two guys in the office who were raised in deep poverty.   Poverty gives you a certain set of shared values, an intolerance for people who whine about stuff that doesn’t amount to anything. 

            Another thing was, one time Theo asked me to do a report on the most successful drafting organizations ever.   I studied it and reported that the most successful series of drafts ever, at that time, was by the Detroit Tigers in the mid-1970s, when they drafted somebody like Kirk Gibson, Lou Whitaker, Alan Trammell, Jack Morris or Lance Parrish every year for several years.  I reported that back to Theo, but I did not realize when I made that report that Bill Lajoie had been the farm director who headed that operation.  Bill appreciated my unintentional flattery. 

            And the really odd thing was, Bill—and I shouldn’t speak for him—but Bill had a lower opinion of Sparky Anderson as a manager than I did.  When Bill took over as General Manager of the Tigers in 1984, he inherited Sparky Anderson as his  manager.  Bill had worked with Sparky years earlier, got along with him alright, but General Managers and managers often have differences of opinion as to what should be done now, especially if the GM did not hire the manager.  I’m sure Bill spoke nicely about him in public, but in private, Bill could rip on Sparky at a pretty good pace. 

            But here is my point.  Bill had no more concept of an empty batting average than he did of the geography of Venus.  You remember Deivi Cruz?  Deivi Cruz was a shortstop who, in 2000, hit .302—with 13 walks in 615 plate appearances.  

            Well, Deivi Cruz was a free agent basically every year beginning about 2001, and one of those years, I don’t remember which, Bill Lajoie really wanted to sign him to play second base.  In one of our discussions about this, he said that Cruz gets on base.  Deivi had a career on base percentage of .293, but because he had hit for some good batting averages, Bill believed that he was good at getting on base.   He could not process the concept that Cruz’ batting average was an empty shell.  He did not have the supporting concepts.  It was like you were talking to him in a foreign language. He had been taught to think, as a child, that a "walk" was something a pitcher did, and if a batter had a lot of walks, that was just because he was a great hitter so the pitchers wouldn’t throw him strikes.  He never looked at on base percentage. 

            He wasn’t stupid.  He was as smart as any of us.  He was then what I am now—an old man who thought about things the way that he had been taught to think about them decades before. 

            Well, no sooner had we signed somebody else and moved on from the Deivi Cruz debate than Randall Simon tripped the sausage.  Simon went from a regular in Detroit to a player who was available, and, a year or two later, one of our other ex-GMs, who will not be named because he is still alive, wanted to acquire him.

            Randall Simon had hit .302 in 2001, with the sort-of-solid looking numbers of 19 homers and 82 RBI.   And, like Deivi Cruz, 13 walks.   So now we had two ex-GMs, Bill Lajoie and the other guy, who were pushing us to solve our first base problem, whatever it was, by signing Randall Simon.   You couldn’t explain to either one of them that Randall Simon was a bigger problem than the guy we had. 

            You can’t tell me that these guys understood the concept of an empty batting average. 

            The real import of this study is that it introduces a method which could be used to study almost any question about MVP voting—whether center fielders are appropriately evaluated in MVP voting compared to third basemen, whether first-year regulars are treated fairly compared to veterans, how much of an advantage there is to playing on a team which has a good year compared to expectations, if not a great year in fact, whether there is racial bias in MVP voting or whether there was such bias in the 1950s, whether Cleveland Indians players have done as well in MVP voting as they should have or whether New York players have an edge in MVP voting (I don’t believe that they do, or ever did.)  Any question you can think of in regard to MVP voting, you could study by this method if you are willing to do the work.  This is all that I have left to give you; I have just had another birthday, and I am probably too old to do a lot of meaningful work in sabermetrics, but I can still show you how studies could be done, if you have the energy to do them.   Thanks for reading.