The surprises and disappointments in the new Statcast xwOBA metric

I wrote yesterday about the newest metric made possible by Statcast, xwOBA. You can read more about it in that article, but basically, instead of treating all singles the same, and all doubles the same, etc., it looks at the launch angle and speed of the batted ball, and assigns credit to the batter based on what that kind of ball usually turns into. Hammered line drive to the wall that Kevin Kiermaier somehow runs down? xwOBA treats it identically to a ball that’s hit in the same way yet gets past Yoenis Cespedes and turns into a inside-the-park-home run.

That offers a lot of potential for better understanding of offensive performance. But what we saw yesterday was that the top of the xwOBA leaderboard looked a lot like the top of the regular wOBA leaderboard. It’s possible that wOBA isn’t impacted that much by luck, and that while xwOBA strips out whatever bit remains, that’s just not that helpful.

To get at that question, I wanted to take a look at one of the other things xwOBA can tell us: who is over- or under-performing their expected offensive output, based on the exit speed and angle at which they’ve been hitting the baseball. While there are a number of reasons we could see that kind of over- or under-performance, this intro is long enough already, so let’s get to the first table.

Now this is promising. There are a few reasons someone could show up at the top of this leaderboard (or on the converse leaderboard, which is just below):

• Speed: xwOBA is concerned only with what happens just after the ball leaves the bat, and what the hitter does once he leaves the box doesn’t change it all. Of course, a speedy batter is much more likely to turn a gapper into a double or a triple than a slow batter is. Whether it’s a good thing that xwOBA doesn’t account for speed is an academic question that we could get really bogged down with; for now, it suffices to say that it’s something that could make a batter’s xwOBA different from his wOBA.
• Ballpark effects: I touched on this a bit yesterday, but right now, xwOBA also doesn’t take into account the shape of a ballpark. A flyball hit deep to left is treated the same whether it bounces off the Green Monster, lands in the Camden Yards cheap seats, or is caught in PNC Park’s enormous outfield. Again, it’s not totally clear if this is desirable, but it’s a source of a difference.
• Luck: This is the source of error that really interests us. Someone’s xwOBA could be very different from their wOBA simply because all their line drives have been hit right at fielders. Often, we can tell that someone is hitting better or worse than their true-talent level — their BABIP being too high or too low is the primary way we usually detect this — but we can’t tell what that true-talent level actually is. If xwOBA can strip out some of that luck, and get us closer to an individual player’s true-talent level, then it’ll be a fantastic tool, especially over small samples.

And this leaderboard gives real reason for optimism about the power of xwOBA. It’s not filled exclusively with lumbering sluggers who are getting dinged for getting singles where most people get doubles; Matt Joyce, at the top of the list, certainly isn’t fast, and James McCann is extremely slow, but one of Raúl Mondesí’s main tools is his speed. Maybe there’s less reason to be optimistic about McCann than there is about Mondesí because of the speed gap, but it’s not just running ability that’s driving this leaderboard.

Nor does it look like ballpark effects are to blame. There are lots of players on that leaderboard who play in totally ordinary parks, and that shouldn’t be a huge surprise. Hanley Ramirez plays in Fenway, sure, but how many fly balls to left can a single player really hit? It seems like, while this is a possible source of error, it’s not a substantial one.

So what the composition of that list demonstrates is that at least some of the xwOBA gap appears to be based on luck. It’s not hard to believe that Mondesí has been unlucky — it’s hard to end up with a .161 wOBA any other way — but it should be reassuring to Orioles fans to see Manny Machado here, and to think that his perhaps his slow start (a .299 wOBA) is not anything to worry about. The story is the same for most of the other guys, allowing this list to serve as a sort of reassurance about unimpressive Aprils, but that’s not the case for James McCann. The Tigers catcher is running a totally respectable .323 wOBA, and xwOBA suggests that should be even higher.

But what about the overperformers?

This is also a list that looks like it has some real insight contained within in. Speed might be the reason the famously fast Trea Turner shows up here (he made the triple portion of his recent cycle look very easy), but it seems highly likely that Bryce Harper and Eric Thames are not actually the baseball demigods they’ve looked like for the month of April. (Both of their xwOBAs are still are in the top 10 overall, so it’s not like they haven’t been ~incredibly good~, just not quite this good.) This is a good place to start if you’re looking for players who might see their offensive stats decline somewhat from their April levels.

Both these leaderboards display promising signs for xwOBA’s utility going forward. You maybe could’ve guessed some of the names that would appear, but probably not all of them, and what that might means is that xwOBA can tell us things other metrics (and our intuition) cannot. While Statcast isn’t perfect yet, already it can enable a lot of cool stuff, and I’m optimistic that it will only get better.