One question I’ve had about the coming deployment of “contact guys” Nick Madrigal and Nico Hoerner is whether they will consistently generate a batting average on balls in play (BABIP) that will support overall production.
We know that contact, on its own, can be a useful thing – a strikeout is a sure-fire out, but a ball in play can sometimes become a sac fly or an error or otherwise move a runner along – but isn’t really valuable unless it is part of an overall productive set of skills. Usually, for the high-contact types, that means they have to post a really high batting average (support their OBP and SLG), which in turn will require a really high BABIP. By the time these guys make it to the big leagues, though, that tends to be part of their game – lotta line drives, decent speed, spraying the ball all over so it’s hard to shift, etc. But also, when you’re in the big leagues, the difference between a .340 BABIP and a .300 BABIP for a high-contact guy can be the difference between a really good regular and a guy who isn’t even worth rostering. (Obviously there is play at the margins for guys who bump up their walk rate or their power, but the general principle holds in most cases.)
So, circling back to Madrigal and Hoerner, my question has looked something like this: can we feel confident that the BABIPs we see for them so far in their careers are reflective of a real ability to post higher-than-league-average BABIPs, as opposed to just small sample luck? Hoerner is at .319 in the big leagues, and had a .360 BABIP last year. Madrigal is at .338 in the big leagues, and had a .324 BABIP last year. So you can see the general range of where we are hoping they can legitimately be: something like .325 to .340 in an average year, which for each of them, typically *would* make them above-average overall with the bat.
Where I had developed some concerns – well, or at least some questions – is that the expected BABIP data at Statcast on both guys was really, really dour, generally indicating that their BABIPs should have been 20 to 30 points lower than they actually were. That’d be a screaming red flag in most cases, though I hadn’t freaked yet because I couldn’t help but wonder if there was just something about these level-swing, spray-line-drives, decent-speed types that was missed in the xBABIP data. Like, maybe something about the exit velo/launch angle combo (which is used to calculate xBABIP) that misses a skill these types have?
You knew I had to be setting something up.
At FanGraphs, Mike Podhorzer was studying this exact issue. Not with respect to Hoerner and Madrigal, specifically, but the overall concept of xBABIP underweighting certain types of hitters. He tried to incorporate more data into Statcast’s xBABIP calculation to create a better expected BABIP. Specifically, he worked in the batter’s sprint speed, the nature of the shifts against them, and the directionality of groundballs.
And, wouldn’t you know it: on the list of hitters whose new expected BABIP in 2021 was undercalculated – according to Podhorzer – by at least 15 points? You guessed it, Nick Madrigal and Nico Hoerner.
Now, keep in mind, that doesn’t prove that the duo’s BABIPs haven’t featured some random luck. That’s still a factor to be considered (Madrigal’s new expected BABIP for 2021 was just .300 (compared to .324 actual), and Hoerner’s was .347 (compared to .360 actual). So this isn’t about TOTALLY eliminating good fortune from the discussion. Instead, this is merely a suggestion that Statcast’s xBABIP calculation probably is missing at least a little something about these two guys’ ability to keep their BABIPs higher than the average hitter.
So if you were like me, and you’d seen those xBABIP numbers and worried that there was a lot of flukiness in their success so far in the big leagues, you can be at least a little encouraged. (And that’s to say nothing of these young hitters’ ability to, you know, improve!) Moreover, I tend to think there is still something missing with respect to the directionality of line drives and pop ups, but I will admit it’s probably pretty hard to suss that out in the data and to attribute it to skill in the same way you could with the other things.