|Mam'zelle Hepzibah to the rescue...|
|Browning 1: Pete...never saw a|
camera he couldn't stare down.
Clearly we should be getting ready to make a bunch of 2012 predictions, right? Great gobs of goo, it's just a few days before America's pastime resumes (though it must be some type of sign'o'the times that the season is once again opening in Japan--just why is it, anyway, that we never permit them to reciprovocate and start their season over here?) and we really ought to be trying to tell you what's going to happen before it happens.
|Browing 3: Robert...preparing|
to do his Bud Selig imitation.
|Browning 2: Tom...not the worst|
pitcher to ever throw a perfect game.
Wasting time reading their breathless math exercises disguised as theories (or is that vice-versa...) when we should be writing about film noir, we were stunned to discover that the time is once again "right" for a hard dollop of "timelining," that quaint old sabermetric custom that expects (and sometimes downright insists) that it can apply a linear value to the quality of play over the history of the game.
The most excruciating aspect of these types of exchanges (which we are proud to report that, for once, we refrained from wading into, either in or out of our new pair of Speedos...) is the eventual appeal to authority as opposed to reason, common sense, or compelling, verifiable research. But they are also instructive exercises in how one can be seduced by their own brain pan, swept up into a lexicon of terms that one's cronies (cohorts) and sycophants (proxies) find irresistible.
So we read on, glued to our seats, only to find out that the glue has seeped into our ears.
OK, let's get to the "theory" (or the "math exercise," depending on what type of glue you've been sniffing). The idea is that pitchers' hitting is the key to constructing a device for "timelining" baseball's quality of play. Working from a series of biological analogies cooked up by Stephen Jay Gould, it was noted that pitchers' hitting has decayed over the course of baseball history. Whereas the original usage of this fact was to demonstrate how baseball followed certain precepts of evolution, the late-post-neo vanguard now is bent on calculating the quality of play timeline from the change in the OBP and SLG of pitchers as hitters.
It's one of those over-ingenious "reach exceeding grasp" audacities that issue from the "counterintuitivity" wing of baseball numberology. And, as with many of these types of ideas, there's a surface excitement to the attempt. But, of course, there's something seriously (if not fatally flawed in it).
No, these aren't exhaustive data points for pitchers' OPS as hitters, they only show the values for years ending in "0" from 1920 until the present. But that's really enough to get the general idea.
Oh, and we also know that this isn't the best data--we need to relativize, which we'll do in the chart that's a bit further down. But the basic data--the original raw data--is here, and it shows that there's a bounce upward in the midst what is otherwise a clear trendline of decline. That bounce has to do with offensive levels (1930 a zenith and a noticeable step up from 1920; 1950 a big year for offense; 2000 a big step up from 1990). While the theorists want to present the data as a linear thing, we must admit to being a good bit more interested in why it isn't quite linear.
Now this shows a more consistent, linear pattern of decline.When the offensive levels are factored in (keep your eye on 1930), you can see that the relative hitting performance of the pitchers actually declined in 1930 even though their raw numbers were up a good bit from what they were in 1920.
Again, however, we become more and more interested in the big drop points on the chart (1930-1940; 1950-1960; 1980-1990) and wish that the math whizzes were more interested in these types of details than the lassoing motion inherent in their singular commitment to sweeping theories/(math exercises).
The chart shows that pitchers' OPS relative to the overall league OPS has fallen from just under 71% of league average in 1920 to around 48% in 2010. That works out to just under a 33% decline over ninety years.
The theory at this point goes on to postulate that if you take a 150 OPS+ hitter in 1920 and bring him forward to the present day, you can adjust according to the value of this decline in pitchers' hitting. That means that such a hitter would be just barely over league average today. (150 x .67 = 102.)
Sounds peachy, doesn't it? Linearity...check. Proxy...check. The rhapsodic power of analogy...check. Or is that simply strike three for this idea? Here's where the flaw comes in.
Simply put, pitchers are not hitters in the same way that position players are. Using OPS as the measure for this makes a huge assumption about pitchers' hitting that is almost certainly not true about regular hitters. It's important that the two components of OPS (OBP and SLG) get examined separately--both to determine how they relate to each other and to examine any differences in how they individually change over time.
The divergence in OBP and SLG relative values is crucial to understanding why the timelining method is seriously flawed. Regular hitters always have a much greater ISO (SLG-BA) than they have OBE (on-base extension, as Alan Shank calls it: OBP-BA). SLG is always higher on a league basis than OBP, but this is not the case when pitchers hit. They are sacrificing power in their hitting to a much, much greater extent than would be the case for a regular hitter.
Failing to adjust for this fact creates an overly simplistic timelining estimate. What we see is that pitchers' OBP has dropped 27% relative to league average over the past ninety years, whereas pitchers' SLG has dropped 36% over that same time span. That is simply not going to translate to how regular hitters would decline if they were "beamed forward" in time from the past.
sledgehammer effect on pitchers' hitting in the second decade of that rule change? We're just conjecturing here--but then again so is the mathex crowd.
Our guess is that pitchers' hitting, while an interesting idea for a proxy, is too glued to itself to be of much use in an actual timeline. There is another step that has to be infused into the process that creates a plausible proxy. Maybe it's simply balancing against the defensive position or batting order position that represents the weakest overall "regular hitting" performance. Maybe it's something much more complex. We simply don't know yet. What will be interesting to note, however, is what trend pitcher hitting takes in the next 20-30 years. Can it go even lower?
Some might argue that the OPS chart argues for jumps in league quality in 1930-40 and 1980-90, but they'd be arguing against the linearity hypothesis by doing so. Nope, we just aren't yet equipped to really isolate the factors that provide a way to quantify how the level of play has improved over time. Best not to keep reaching where we can't yet grasp...