24 November 2014

T.J. McFarland's Tricky Platoon Split

This post was written by Ryan Romano. Follow him on Twitter.

Platoon splits are weird things. They often don't reflect a player's true talent level, meaning we must heavily regress them to get a more accurate projection. (Like many things in baseball, this applies even more when the sample size is small.) I don't take a bold stand by saying that — all sabermetricians agree on it. But the means by which we look at a player, particularly a pitcher, and determine if he has a platoon split can lead to fallacious thinking. While most people utilize the numbers of opposing batters, they should probably rely on pitching-oriented numbers instead, because the former can deceive.

T.J. McFarland illustrates this conundrum pretty well. Does he have a platoon split? By some statistics, no. For his career, he's faced 244 left-handed batters, who have a .322 wOBA against him; the 342 right-handed batters to whom he's pitched only top that by three points (.325). That's really not much of a difference at all, and most projections would reflect that. So we can expect him to consistently retire batters of any handedness going forward, right?

Well, about that. Those same lefties have struck out in 21.7% of their plate appearances, while walking a mere 5.3% of the time; when they've put the ball in play, it stays on the ground 63.3% of the time. By contrast, the righties own marks of 11.4%, 8.2%, and 57.4%, respectively, in those regards. Put it all together, and McFarland has a career xFIP of 2.77 against same-handed hitters, to go along with a career xFIP of 4.32 against opposite-handed hitters.

His repertoire corroborates the xFIP side of the story: The two pitches on which he leans the most — his sinker (65.1% of career pitches) and slider (19.5%) — also have two of the largest platoon splits among common pitches. In other words, sinker-slider pitchers will almost certainly have a significant platoon split. His increased implementation of his changeup against righties (18.5% against them, 0.4% against lefties) can compensate some, but as the aforementioned plate discipline and batted-ball marks show, it can't do it all.

Why does this matter, though? Maybe McFarland's wOBA splits do a better job of reflecting his true talent level than his xFIP. A closer examination of the splits reveals that to be, in all likelihood, false. The discrepancy between the two splits comes almost entirely as the result of home runs. Of the nine he's allowed since debuting last year, left-handed batters have hit five, despite facing him far less often than right-handed batters. With lower fly ball rates, they've done this by making the most of the few fly balls they do hit: Their HR/FB% of 16.7% dwarfs the 6.2% figure their opposition has posted. Since HR/FB% is a notoriously fluky statistic, especially with a sample as minuscule as this, this almost certainly doesn't have anything to it.

So what does this mean going forward? For the past two campaigns, McFarland has served as a solid long reliever, someone whom the team could trust to pitch effectively against any batter. In the coming seasons, however, that might not hold true; if it doesn't, McFarland could see his role reduced to a one-inning guy, or even a lowly LOOGY. He shouldn't decline notably overall — Steamer projects a 3.46 ERA and 3.64 FIP, in line with his career numbers of 3.58 and 3.59, respectively — but the good fortune that he's had so far, and that has given him the veneer of an equal-opportunity hurler, will probably come to an end.

Photo via Keith Allison

No comments: