Saturday, July 6, 2013

EXTINCTION OF THE EXTREMES: NEO-SABE TWO-DIMENSIONALITIES/1

So WTF are we free-associating here? This: those
missionaries in war-torn China are analogous to the
"insider neos"--their agenda deviates more than a bit
from what they claim their mission is. [Oh, and--of
course--as is usually the case even in pre-Code films,
Babs is not nearly so svelte nor so (sigh...) diaphanous
as we are led to believe is the case.]
It's not fun to file negative reports, despite what some may perceive from these pages as barely restrained glee packed between our highly-charged dependent clauses. All too often the laughter, the parody, the semi-sotto voce sleight-of-hand is employed simply to keep from crying crocodile tears. (But don't cry for us, Argentina, or whatever your name is...we're brave enough to keep quaffing The Bitter Tea of General Yen.)

Today we bring more evidence that sabermetrics, as practiced and parsed in the front office of baseball teams, provides more heat than light. Research that leapfrogs theory in order to rush into application is the hallmark of what we've taken to branding (and we choose that word knowing its reverberations in the math-meets-MBA enclaves of shiny careerists) as the "neo" phase of baseball numberology.

The troubling truth is that the result of these applications, appearing in so many directions as the length of the "neo" incursion into the ongoing chaos of front office "thinking" begins to enter its own "institutional" phase, we are seeing signs of this analysis producing changes that promote increasing two-dimensionality in the game as we see it on-field.

Some of this is, in fact, positive. But, as we've noted before, many of the "sins" of pre-sabermetric theory aren't quite so heinous or indiscriminate as what's been portrayed. While the so-called "cutting edge" is being bandied about in questionable applications of WAR and dubiously-conceived attempts to further compartmentalization in modeling (witness the "pitch framing" quantification movement, which we'll get to later in the year), many of the original ideas that informed sabemetric notions of value have been allowed to atrophy. We find that instead of promoting extremes in performance, the current trends indicate that these approaches are moving performance extremes toward extinction.

We're not talking about HR records, or batter's strikeouts, or any of the trends that tend to ebb and flow in tandem with the forces that contribute to the fluctuations of runs/game. In this instance, we'll focus on an area that is an indirect but significant contributor to offense--the batter's base on balls.

As we've noted before, the trends in offense in the 90s (and that carried into the second half of the 00s)--a virtually uniform focus on isolated power (ISO) have, as offensive levels continue to decline, resulted in a loss of high-walk hitters. This decline actually presaged the eventual downturn in offense, and, like batter's strikeouts, is threatening to become monolithic.

Let's have at the numbers and see what has happened with high-walk hitters (HWH) over time. Our definition of HWH is a walk pct. at 15% or higher in a batting-title qualifying season. There have been 945 of these player-seasons in 215 league-years from 1901 through the present. We will look at this data in two different ways to demonstrate how different perspectives can shift one's perspective on an issue.


The chart above shows the number of players per year with 15% or higher BBP. It's a wild chart with three major spikes--two of which seem related in some way to expansion. Player spikes occur in the late 40s, the early 70s and right at the fin de siecle (which, when you pronounce the phrase properly, conjures up Adam Dunn as a genie with a giant hole in his swing).

The spikes, as you can see, were short-lived, and you can see that since 2009 the remedial uptick has collapsed in a fashion disturbingly reminiscent to what happened to the stock market in 2008 (the operative phrase is: "going south").


But, in fact, this chart is misleading. Not wrong, per se, but inaccurate in how it represents what's happening. Why's that? Because there are far more hitters in the post-expansion world of baseball (basically the right half of the chart) than there were before. There should be an increase in the number of HWH (high-walk hitters, in case you've forgotten).

To see what's really happening, we need to see the percentage of possible hitters who are HWH. There are twice as many "possible hitters" by our definition in 2013 than was the case from 1901-1960, and our next chart reflects that reality.

Quite a difference, eh? The two subsequent spikes now pale in comparison to what happened in the late 40s (and we still don't really know what was going on in those years, save for things like the presence of Tommy Byrne). The totals flatten out in the mid-70s and are much closer to the historical average (shown, as in the first chart, as the breakpoint between the black region and the color region).We can now see that the 90s only created a few HWH around a power boom and its spike was very modest.

And when we examine it in this context, we can see that the title of our blog post ("extinction of the extremes") is no overstatement. As of this year, HWH are more endangered than they've been since the deadball era.

So are the neo-sabes really to blame for this? How can they actually be influencing on-field behavior and actively contributing to this precipitous decline? Consider it a failure of theory combined with a set of distractions in practice: if anyone drawing a salary to be an "advanced baseball thinker" had actually thought about this, they would have realized that hopping on the ISO bandwagon would eventually create a series of changes in the possible relationships of OBP and SLG that could remain in existence once ISO became such a monolith. By encouraging and condoning a single approach to hitting, the possible relationships between OBP and SLG would narrow and calcify, and would very likely result in just what we're seeing in the chart.

So, yes, the front office brain trust should have been able to see this coming--or at least some of it. What will they do when they discover that someone (even a sardonic semi-curmudgeon) is holding their feet to the fire for such passivity as hitters manage to take more pitches per PA on average than in the years when we actually had a robust total of HWH? Well, they'll shift the focus to the pitchers, claim that the trend toward five-inning starters and the "long-sequence" bullpen is the reason for this.

But much of it is based on the fact that no one is thinking about the training issues involved in batter strike-zone judgment. As "essentialists," the front-office neos have to rely on their models more and more conspicuously, and their ability to study such issues as how to train hitters to draw more walks is curtailed once they start believing that the actual offensive value of a walk (in a strictly Linear Weights model) is relatively slight. Once the idea is cast aside, put in the bottom drawer, shoved in the filing cabinet as they run computer models and "go out for coffee" (Rany J.'s "brave new world" circa 1998, now all too prevalent a mindset from front office to mom's basement), it atrophies.

That atrophy, that extinction, is what's shown above.


(Looking at this from the vantage point of rolling five-year averages mitigates things just a little bit, but not enough to take the sting out. Players who can work the count for walks are simply disappearing from the game, right before our eyes.)

Now it's also possible that someone will attempt to rebut by suggesting that walks have been leached out of the "market inefficiency" and are perfectly fine with increasingly constricted variances. That kind of thought, however, produces a rolling wave of two-dimensionality that is likely to creep into other areas. The so-called "opening of knowledge," at least as it's being practiced, may ironically make baseball into something two-dimensional. Twenty or thirty years from now, should these types of trends continue, we may well have found that sharp minds have produced a very dull object--namely, a game that is mind-numbingly uniform.

Are we on that road? Could be. In fact, it's quite likely.



Can sabermetrics reverse course and find a way to advocate for greater diversity in the face of its lockstep-inducing methodology? Hard to say. The signs don't look good, however.

While it's too soon to simply start crying, odds seem to favor that the end result of the peculiar type of compartmentalized tracking that's taking hold will simply be a detailed record of the tracks of our tears--for what we've lost, and for what might well be unrecoverable.