Tuesday, June 26, 2012

CHEEKY REVISIONISMS/1: THE DECLINE OF THE "THIRD SABERMETRIC AGE"

We know that the following will not fly in the little world of baseball numberology...the reasons are both short and long. We know that to identify an era in time by a phrase that did not exist in that time frame is to incur the wrath of the historian--or at least that portion who wish to serve vigilantly as knee-jerk guardians of methodology.

And we know that the perspective of those who "man the barricades" in that world will never accept the idea that their concepts, murky and muddled as they may be, were in any way existent prior to the accepted lineage of "advanced statistical concepts" as they are now commonly accepted--and, indeed, taught in a smattering of college courses across this fair land.

But you know what? None of that bothers us. (As you would doubtless expect, if you are familiar with the landscape here.) None of that, in fact, matters one little bit.

We are going to suggest that despite any and all lack of articulation of the principles, ideas, and latter-day ideologies that have come to be known under the clustered term of "sabermetrics," there did come to exist a series of primordial intersections with basic concepts that in the past thirty years have been written about at an increasing pace. (The term ad nauseum comes to mind, but it's probably a tad too negative...even for us.)

Bill James has made a yeoman effort to capture the pre-history of sabermetric concepts in his Historical Baseball Abstract. Much of that work, which mercifully avoided the more recondite mathematics that the "neo" phase of the movement has mostly mired itself in, centered around managerial tendencies and the lineage of concepts passed through realms of influence (what James has occasionally referred to as "managerial family trees").

We're not here to talk about that, of course. To put it in football parlance, we're here to run a reverse, a maneuver that will take the original play (or concept) in a completely different direction and use the principle of surprise to make a gain downfield.

To do that, we'll look at one feature of baseball offense--the base on balls--and we'll see how often the teams in the National League reach a certain threshold over the league's history (beginning in 1901 and continuing to the present).

Why just the NL? Two reasons. First, of course, the lack of the DH gives us a unified data set. Second, the AL was more thoroughly revolutionized by Babe Ruth--if for no other reason than the Babe's own presence. He skews the data: if we were to use him, we'd be making claims about the 30s that, in the cold light of analysis, would clearly be premature.

So it's the NL, and here--at long last--is what the bald, blustery, oversimplifying revisionism is all about. When we look at the history of the NL in terms of its teams, we have 1140 team seasons (from 1901 to 2011). Of those, exactly 119 teams have managed to draw 600 or more walks in a season. That's just under 11%.

What we posit here (submitted, as Rod Serling might have said if he were a standup comic, for your disapproval...) is that any time the three-year average for the league exceeds 11%, we are, de facto, in a "sabermetric age."


When we chart that, as you can see above, we see that there have been three so-called "sabermetric ages" in NL history--the first immediately after World War II; the second beginning in conjunction with baseball's second expansion (the expansion, in fact, that was the only one in which four teams were added in a single year--1969); and the third commencing in 1996 and lasting throughout what Eric Walker (not Walt Davis!) termed the "sillyball era" (also the steroid era).

Note that we placed the crossover value at the historical NL average (11%) so that the many years of "zero" (600+ walks was virtually non-existent for an NL team prior to the 1940s) will show up more dramatically. (A veritable sea of red, in fact, stretching below the line, all the way to zero.)

Now we've already suggested that some readers will quibble--perhaps violently--with this cheeky revisionism. What they should realize, however,  is that these are time frames where one of two things happened: offense exploded and took walks with them, or some other set of forces, conscious or unconscious, permitted teams and their managers to include strike zone judgment in their offensive approach.

Those formulations will allow the quibbler to suggest that the last fifteen years, when stat analysis and its associated proselytizing has increased at a rate that is somewhere to the north of the term "exponentially", is simply an unconscious reflection of the offensive explosion. However, what we see when we look at the actual NL teams that exceeded 600 walks in the 1996-2010 time frame (a total of 59 teams, 38 of which occurred from 1996-2002 and only 21 from 2003-10) is that they run the gamut of run scoring.

In short, they are not all skewed to teams that had such potent offenses (relative to their overall offensive context) that their walk totals were simply inflated by any sort of "fear factor". (Such a factor, of course, must be granted for several of the Giants squads in the early 2000s, when a player--Barry Bonds--with a skewing factor analogous to that of Babe Ruth was present.

The upshot is that this was a time frame in which certain teams embraced this concept, and it had its most sustained occurrence over that fifteen year span.

However, what the far right side of the chart is seeming to indicate to us is that this age has past.

All of which suggests that our claims about the misplacement of emphasis in the neo-sabe movement might well be on-target: namely, that misplacement has resulted in a series of its core tenets atrophying.

Do we have any NL teams in 2012 on pace to join the ranks of 600+ walk teams? No. The current leader in walks--the Los Angeles Dodgers--is on pace to walk 567 times.

Now we know that simply drawing a lot of walks is no outrageous guarantee of success. However, the 119 teams that walked 600+ times have an aggregate winning percentage of .533 (10147-8904). NL teams that have walked 500-599 times average out to .509; those that walk 400-499 times have an aggregate WPCT of .490. And those that walk less than 400 times in a year (leaving out 1981 and 1994) have an aggregate WPCT of .468.

So--as sabermetricians used to passionately argue--there's some basic, intrinsic advantage in drawing more walks (for teams as well as for individuals). More importantly, however, it's something that adds dimensionality to baseball--something that, as we noted awhile back, is slowly but noticeably decaying. The increasing lack of dimensionality in the recent game is attributable--at least in part--to a lack of dimensionality in neo-sabe thinking as it has infiltrated the front office and remained oblivious to the features of its own "sabermetric age" as manifested (alas, temporarily) in higher walk totals. The "consultancy culture" has rushed past its earlier precepts, and has overlooked one of the movement's original rallying cries. What grand irony...could it be that we've lived through our great age without knowing it, and that it's slipped through our fingers?