Miller’s Tale

Spoiler Alert: We’re going to tell you two sentences from now where we come down on the subject of Brad Miller, enigmatic Mariner. Jump to the next paragraph if you don’t want to know until the end of the article. Spoiler: We have no idea. The point of this piece is to call your attention to what we discovered when we started investigating this subject: we found that new, exemplary, state-of-the art, microscopic, persuasive explorations of Miller and his stats, in sickness and in health, by undisputed Fantasy All-Stars, wind up in opposite corners, and can’t be reconciled.

We Fantasy addicts often focus on the statistic du jour rather than looking at a player’s entire performance. Sabermetricians are producing ever more sophisticated analyses of ever more granular data—xBABIP, xK, xISO, Contact Quality. Reading these analyses dazzles us outsiders, and it’s very easy for both sabermetricians and their readers to be sucked in by them: “Beshrew our souls,” we say, or something to that effect; “we thought Damon Rutherford’s [to pluck a name from the ether] negative OBP signified that we should trade him away/jettison him/eschew him. But, learning now that his xBABIP is way higher than his BABIP, we see that he’s just been unlucky and is certain to improve, and we should trade for him/purchase him/draft him.” This leaping to conclusions usually isn’t the fault of the indefatigable S-metricians who are doing this investigative work while others sleep. The SMs tend to be careful not to let the language of prediction infect their analyses. But we readers, being human (and owning going-nowhere teams that we’re desperate to improve), sometimes take it that way, and occasionally even the SMs present their studies as if the factors must balance and (for example) an xBABIP >BABIP is a sign of good things to come.

And you know—why the hell shouldn’t they? An “expected” statistic is meaningful from a Fantasy standpoint, it seems to us, only if we’re persuaded that reality and expectation merge at some point. But that may be an illusion, just like the illusion that parallel roads merge in the remote distance. In fact, all that can be said when Rutherford’s xBABIP is higher than his BABIP is that Rutherford’s been unlucky on balls he’s hit into play. It’s not necessarily predictive of overall player performance. Why is this? Because, for example, we have no idea if the mix of balls in play was lucky or unlucky, if contact rate is up, down, regressing, progressing, and so on. Here’s a homely hypothetical example: We’ve noticed that Rutherford’s hard contact rate has doubled, but with no effect on any of the statistics we care about. Things will even out and he’ll have a breakout second half, correct? But suppose Rutherford’s been swinging harder and therefore making less contact of any sort. Is the increased hard contact good or bad? An analysis that looks at only one part of a big picture—in other words, what an economist would call a partial equilibrium analysis—may prove misleading. That is, it may prove wrong, as in “I’m drafting Matt Kemp because his hard contact rate in 2014 was his lifetime best; he’s not over the hill.” And an honest analysis that looks at multiple aspects of a problematic player is almost always going to be inconclusive, however deeply the analyst has dived into the lightless abyss of neutrino-level stats, and however fascinating the samples of new species with which she has returned.

Case in point: Mariners multi-position guy Brad Miller, who keeps winning and losing and winning again (and, since he is on our NFBC team, we hope not losing again) a starting job. Last week, Alex Chamberlain of Fangraphs, whose shoe’s latchet we are not worthy to unloose, Fantasywise, observed that Miller’s xBABIP was .055 higher than his BABIP and suggested (while noting that Miller’s high strikeout rate makes him a risky play) that, if one needs a shortstop (as one usually does), then Miller might be “Ian Desmond lite.” In Baseball Prospectus on the selfsame day, Greg Wellemeyer, the hem of whose garment we are not worthy to kiss, Fantasywise, highlighted Miller’s atrocious in-zone contact rate—in other words, Miller’s failure to hit balls thrown in the strike zone–and slyly recommended emphasizing to some clueless fellow-owner the very statistics on which Alex relies in order to trade Miller “before his value completely craters.”

We don’t know who’s right, and we don’t have a big original point of our own to make here. The main conclusion is excruciatingly obvious: look at all the data, be judicious, don’t follow leaders, watch the parking meters, etc. We’re rooting for Alex, partly because he plays for the home team and mostly because, with Martin Prado hurting, we’re obliged to start Miller this week. We nonetheless fear the worst, because—doing a pitch-type analysis, which is one of our favorite partial-equilibrium can openers—we think we see in Miller a guy who’s helpless, and now known to be helpless, against major league breaking balls and off-speed pitches. But maybe we’ve got it backwards: maybe Miller’s success against fastballs has caused opposing pitchers to retrench, and Miller’s catching up with them—he’s just been unlucky so far. We don’t know—we don’t know what to do ourselves (other than wish we’d obviated the issue by making a bigger bid on Carlos Correa) or what to suggest you do. But this stuff is fun, isn’t it?

One other thing, while we’re thinking of it: we very seldom play daily games like DraftKings or FanDuel, but since we do play in a Head-to-Head league with daily lineup changes, we occasionally notice a player we think will have a very good (or, if a pitcher, very bad) day. Henceforth, when we see one, we’ll mention it on Twitter (@birchwoodbroth2). Bear in mind that, if we’d started doing this yesterday, we’d have recommended Luis Valbuena (2 HR). On the other hand, if we’d started doing this the day before, we’d have recommended Carlos Rodon (3 2/3 IP, 9H, 1W, 7ER), pitching against the Pirates.