January 20, 1998

Abstract Progress

Don't overestimate Bill James' effects on baseball

by Steven Rubio

In 1982, after several years of self-publishing, Bill James released his annual Baseball Abstract with a major publisher for the first time. Since then, the baseball world has never been the same. The influence of James' work during the 1980s cannot be overestimated; while some might argue that if not James, then someone would have arisen to begin a revolution that fueled fan interest in baseball analysis, it was in fact James and his Abstracts that did the job of making publications like Baseball Prospectus possible.

Arguably, baseball analysis has come a long way since 1982. The Prospectus is one of many examples of baseball fans following James' lead. Even the mainstream baseball media seems to have made advances: the work of Rob Neyer and John Sickels on the ESPN website, for instance, garners a fairly large audience for these old colleagues of James. A baseball analyst in 1998 might feel comfortably proud of the influence of Bill James and his inheritors on the general public. But while there is no denying that the caliber of baseball analysis has progressed in the last fifteen years, it is not as clear if that analysis is working its way into the public, or the minds of the average fan or baseball writer.

Take the following three items:

• In that 1982 Abstract, James wrote at length on the topic of "overrated" and "underrated" players. He offered eight factors that might result in a misperception of a player's value, three of which were statistically related. Those three were: performance in triple-crown categories (for instance, BA gets more attention than OBP); defensive performance (defensive statistics being less accessible to the average fan than offensive stats); and what I'd call a narrow focus of abilities (a guy who hits .300 but does little else gets more attention than a player who does very well in a broad spectrum of skills). This discussion took place as part of an analysis of Steve Garvey, who, as James noted, was good in the triple-crown stats, had more offensive than defensive value, and "does very few things well, but does them very well." In James' ranking of 1982 major-league first baseman in that book, Garvey finished 12th of 25, behind folks like Mike Hargrove, Willie Aikens, and Warren Cromartie.
  
  
• In his final Abstract in 1988, James listed the ten things he had learned over the years that would be most useful for a major-league organization. In one of these items, he noted that the on-base percentage of a team's leadoff man was the most important factor in determining how many runs a team would score. The year before, 1987, the league leaders in walks were Brian Downing (who scored 110 runs despite hitting "only" .272), and Jack Clark (who scored 93 runs while hitting .286).
  
  
• In the recent Hall of Fame voting, Steve Garvey received 195 votes. Jack Clark received 7. Brian Downing received 2.

Here are some career numbers for the three players:

          G     BA  OBP
Clark   1,994 .267 .383
Downing 2,344 .267 .373
Garvey  2,332 .294 .333

Or, since this is a discussion of public perception of a player's contributions in the wake of advances in statistical analysis, these numbers courtesy of Clay Davenport. EPEQA (or Effective Park Equivalent Average) is the park- and season-adjusted values where league "average" is set at .260, and whose distribution closely approximates that of batting average; EPER (or Effective Park Equivalent Runs) depends on EPEQA and playing time, and follows the distribution of RBI:

        EPEQA EPER
Clark    .309 1346
Downing  .288 1294
Garvey   .287 1394

First, these are all clearly fine ballplayers. Second, they were comparable in terms of their career value. Clark was the better hitter, but he also played in 350 fewer games. Clark was an underrated right fielder who degenerated into a poor 1B/DH, while Downing was a weak catcher, a hard-working outfielder, and finally a DH. Garvey won some Gold Gloves (for which there is scant statistical justification) at the easiest defensive position.

So, lets say you wanted to draw some conclusions about these players, and how they should be remembered in the context of their careers and the era in which they played. Let's give Garvey some credit for those Gold Gloves and take away a bit of credit from Clark for having a shorter, injury-plagued career. It still looks like these guys are close enough in value to make for some fun hot-stove arguments over which was better. So how do we explain why Garvey got more than 20 times as many Hall of Fame votes as Clark and Downing combined?

You could say Garvey had a better image, but Downing has generally been presented as an upstanding guy, while Steve's reputation took a few hits in the years after his retirement. You could say Garvey had a clutch reputation, but you could say the same for Clark. No, there's probably a much simpler explanation for those Hall of Fame votes. It's in the chart above: Clark and Downing hit .267, but Garvey hit .294 (and hit over .300 seven times in his eight prime years).

You would hope, sixteen years after that first readily-available Abstract hit the bookstores, that a good number of people would realize a 40-50 point difference in OBP is hugely important, or that a first baseman with ok-but-not-great power and an OBP of .333 is far from being a Hall of Fame caliber ballplayer. You could hope that. You'd also be wrong, and that just tells us we have a long way to go. Certain concepts that baseball analysts may take for granted are still far from gaining widespread acceptance.