I came across this fun recent post by Bill James, who writes:
[Before Moneyball] 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. . . .
I [James] 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 . . .
I love this partly for the content and partly because Bill James writes . . . just like Veronica Geng’s affectionate parody of Bill James. I can’t get enough of this stuff! It’s like visiting some country that specializes in a particular dessert, and when you’re there you have to have it with every meal.
James’s essay is also fun because he addresses two related issues: (1) how did the different players perform (as measured by wins above replacement) and (2) what was the mental model of baseball executives: how did they perceive player performance? It’s kind of like what we did in Red State Blue State, where we looked at how people voted, and we also tried to understand how the pundits could’ve kept getting it wrong.
When you have a new idea, it’s not enough to show that it works better than the old idea. You also need to explain why, if your idea is so great, people weren’t already doing it.
And this reminds me of my question when I wrote about James nearly ten years ago for Baseball Prospectus: Given all his writings about empty batting averages and how you shouldn’t take RBI so seriously, how come he provides the following four statistics for every player in his historical abstract: games played, home runs, RBI, and batting average. At the very least, why not give on-base percentage and runs scored?