Monday, November 19, 2012


Shameless plug # 2112b: BTW, we've long since forgiven
Mr. Murray for not being much of a baseball fan...
We've been absolutely swamped this week with location work on our continuing labor of love, the decidedly non-baseball documentary Don Murray: Unsung Hero, which decamped to NYC for a week of interviews and verité footage in and around the Big Apple (the great but undervalued Mr. Murray is an East Rockaway native), so we've had to sit out the playbacks on the AL MVP.

And, frankly, that's just as well. Because the war of words that issued from Miguel Cabrera's selection was a watery bloodbath (save, of course, from a few folks misusing the radio in their time-honored fashion) that just might start to indicate to many that awards voting is not necessarily the best path for either proselyting or cementing ideological purity.

The big problem in the perspective of that vote, in fact, stems from those who've staked their own reputations (and their twilight status as "insiders" thanks to baseball's indiscriminate entrepreneurial approach to information technology) on what their One Big Number is supposed to tell us. While most of these folk (we won't name any names today, but you know--and they know--who they are...) like to make a few circling motions in the wind about how these stats aren't definitive of anything, such practiced acts of false modesty are just that.

And in the case of the 2012 AL MVP, virtually all of the intrigue in it stems from the fact that one system ranked Mike Trout's non-offensive value at such an extreme level that it made the comparison with Cabrera into something egregiously lopsided. Our suspicion is that the 28 voters entrusted to decide the AL MVP (and--hey, really? just twenty-eight?? is that not one of the most ridiculously small sample sizes ever for such an important honor???) may well have looked at the various "advanced numbers" as part of their decision-making process. Trout was touted heavily in the MSM while his performance was at its peak from late July-mid August, and at least a smattering of those voters are known to reference those numbers. kinda looks like a Presidential tracking poll, doesn't it??
We'll never know for sure unless we can get all of the two-dozen-plus-four to talk (so far, the number who've given a public rationale for their vote can still be counted on the fingers of one hand...), but we strongly suspect that there were another 6-8 voters who were on the fence between Cabrera and Trout who looked at the distance between the two players as represented by the version of WAR at Forman et fils and simply shook their heads in disbelief. It's possible that the idea that a player en route to winning a Triple Crown could be so inferior to another player in that same year--not slightly inferior, mind you, but massively inferior--just seemed so incredible to them that it pushed them back into voting for Cabrera.

And, really, you can't blame them for being incredulous. We admit to having the same initial reaction, and before we embarked upon our East Coast odyssey in November, we spent a few days in St. Louis with long-time pal Brock J. Hanke and revisited the old defensive methodology that had been in place during the days of the Big Bad Baseball Annual (BBBA for those of you who've worked hard to forget...) in search of some kind of perspective on all this.

Hey, if we looked as good as this little guy, we'd be on TV...
Craig Calcaterra said that statheads don't want to be God: while
that's true, Craig, you don't have to be God to engage in a holy war.
What we found was a bit baffling. It was clear that Trout, while clearly an above-average center fielder, was not the best in the league as measured by a variant of the old method that included baserunner advancement data (something not available to BBBA during its publication run). All variants that we generated seemed to indicate that a series of factors--the modeling assumptions in outfielder defensive value, the quirks in the zone rating systems being used, and the lack of an observational method that indicated how many discretionary put-outs existed where a corner OF could have made the play--all of these combined into a kind of "perfect storm" that elevated Trout's defense from the very good to the other-worldly.

Our various methods for calculating center field defensive value for the 2012 AL consistently suggested that Trout's projected value was being overstated. The five variations that were developed ranged from an over-estimate of 0.4 wins on the low end to 1.9 wins at the top end. Our overall delta was +1.1 defensive WAR (rounding up just slightly).

So you can pretty much halve those defensive WAR numbers, from 2.1 to 1.1 (2.14 to 1.06, to be egregiously exact). Numbers wonks will still say that Trout was more valuable than Cabrera, of course, and no one (at least no one here) is trying to get them to stop. They'll continue the argument, however, by claiming that Trout was more valuable offensively as well, which is where the baby and the bloodbath water start to splash into observers' faces and induce a case of pinkeye.

There are also assumptions and flaws in the park factors being used to calculate offensive value, and we spent some time on a recent airplane ride revisiting those issues, too. The standard method for calculating park factors is incapable of handling the schedule variations that have come into play thanks to interleague play and the flukish micro-climates that now proliferate in the detailed batting data of hitters. Faced with this fact about a decade ago, we looked for ways to adjust those factors for our own amusement, and discovered that several adjustments that move the park factors back toward the league average constitute the best possible way to prevent distortions.

What on earth do these two guys have to do with the topics
being discussed here? You may (or may not...) find out below.
For example, Trout's home park is though to be a reasonably extreme pitcher's park, but when you make adjustments for exactly where the Angels played in 2012 (including games in Coors Field) and when you make league-average run environment adjustments, the park snaps back a good bit closer to neutral (from 92 to 98). A smaller but similar adjustment in the opposite direction also affected Cabrera's home park (Comerica Field).

At issue as well the amount of actual value stolen bases and double plays should assume in such a system. We wish that Forman et fils would add stolen bases to their invaluable Play Index data, so that the context of these events could easily be captured and analyzed. Leaving it to the WAR formula, which applies a rote expected win value approach to the event, is not sufficient. A look at players' stolen bases in terms of eventual run scoring is feasible with play-by-play data: when we do the labor-intensive work to look at it for Trout and factor in the events that ultimately weren't needed to result in a run scored, we see that the total number of baserunning runs that are produced in the calculation are cut almost in half--5.3 vs. 10.1. (A similar measure shows that Maury Wills' 1962 season, with a similar success rate but with more than twice as many SB/CS, grades out at 12.9 instead of 19.2.)

When we do all that, and when we halve the run values assigned to GIDPs, the WAR value shifts a bit: the result of these adjustments effectively reverses the OPS+ value at Forman et fils as shown for Trout (171) and Cabrera (165). The gap in the Batting Runs stat is also widened (Cabrera gains a couple of runs there, while Trout loses one). All of that gives Cabrera an edge in the basic offensive WAR stat (7.7 to 7.3).

Now, we don't know that ours are the definitive values...we don't know if anyone has the "right" answer. But it's clear that these numbers have taken on a great deal of importance in the tone and shape of the argument, and we seem to revisiting something similar to the extremist psychology gripping the political landscape in this country--a psychology that stems, in large part, from an impatience on the part of those who are pushing toward reason which results in a conflation of method with meaning.

So what the eff does that last sentence mean, anyway? Let's contextualize from the election results. The fact that Nate Silver was right about the Presidential polls does not mean that it was right to vote for Barack Obama. (That was the right choice, in our not-so-humble opinion, for a whole host of other reasons). What's happened with the AL MVP race, however, is precisely that--a set of numbers were masquerading as reason itself, when those numbers needed aggregation, further interpretation, historical comparison--anything but the puffed-up certainty that accompanied them in certain circles during the latter stages of the 2012 season.

The best that can be said for Ocker's mystifying vote
for Adrian Beltre for #2 is that it was...anomalous.
It was, in effect, a variant of the "Romney spin" that much of the political punditocracy was playing out in the latter stages of the election campaign. And it proved to be just as (in)accurate in predicting the winner of the AL MVP award.

Let's close with a thought about what would have happened if the hard-sell harping on that outrageously high gap between Trout and Cabrera hadn't turned away some voters who may have been receptive to shifting their vote. If eight voters had shifted from Cabrera to Trout, we'd wind up in a virtual tie: that vote tally would shift from 362-281 to 322-321. Then, if the inestimably unmentionable Sheldon Ocker of the Akron Beacon Journal hadn't been the lone wingnut deviation from the Cabrera-Trout binary (either voted 1-2 or 2-1 by 96% of the voting population), we mighta coulda (shoulda?) had a tie in this MVP race, something that traditionalists and the various flavors of Stalinists in the little world of numberology might have found a way to live with--a kind of benevolent whitewash to the watery bloodbath that will now continue to divide rather than unite.

(In any event, we sure as hell would like to have seen a much larger BBWAA voting population involved in this result...with technology and analysis being what it is today, the notion of limiting the voting population to league beat writers and a few other random chimps is as outmoded as the Model T. We wonder what a full BBWAA membership vote on the 2012 MVP might have looked like.)

That day is further off, and while in the real world it's the "bad guys" who are keeping that from happening, here in this little sub-region of "reality" it's the "good guys" who are doing same. And the "good guys" should know that it's much more possible--and much more feasible--to have "co-MVPs" than "co-Presidents." Keeping that dichotomy in mind might keep things in better perspective.