Luke Voit remains a mystery heading into 2019

I’ve been thinking a lot lately about a paradox in baseball. Projection systems have never been more sophisticated, and information has never been more democratized. That’s made it easier to project a player’s performance down the road. We know that a player like Aaron Judge is likely to have success in multiple, successive seasons, because the information we have available indicates so. The flip side of that is it’s sometimes more difficult to project a player’s future, because that player has access to even more information.

Take someone like J.D. Martinez. In 2013, he was cut from the Houston Astros, owners at the time of one of the worst rosters in the history of baseball. The Astros were only just beginning to cement a reputation for being an analytically-driven organization, and even THEY didn’t see Martinez as worthy of a roster spot on a team that had just lost 111 games. Martinez seemed to be on his way out of baseball, with a career 87 wRC+, terrible defense and inability to hold a major league roster spot.

Of course, Martinez isn’t out of baseball. He went to the Detroit Tigers and immediately became one of the best hitters in the game. In the five seasons since leaving Houston, he’s posted the following season wRC+s: 154, 136, 143, 167, 170. This wasn’t luck or a dead cat bounce; Martinez has famously gone all in on modern analytical trends, rebuilding his swing with private instruction to optimize launch angle. Not only has he privately improved himself, but one of the soft arguments made in favor of him winning AL MVP this season is that he’s spread the doctrine of better hitting analysis to the whole Red Sox team, and continues to be as diligent a student of hitting as anyone in the game.

Martinez never projected to be this good, because the most sophisticated algorithms don’t know, and therefore don’t take into account, sudden and dynamic improvement, especially when it’s done under private instruction. Even after his first stellar season with the Tigers, we just didn’t know enough to say whether it was a one-off career year or a real, sustainable change. The amount of information players can access, and the help they can afford, has made it more difficult to sort the real great players from the one-offs, hence the paradox.

A different but related example of this is Luke Voit. We all know his story by now, and are familiar with the fact that for 148 glorious plate appearances in 2018, he was the Yankees’ Mike Trout. Extrapolating Voit’s 194 wRC+ in his short Yankee stint over 600 PAs gives you 56 home runs and 133 RBI, and would make him the best hitter the Yankees have seen since prime Alex Rodriguez. He’s not that good, probably, but can we project regression for him in 2019?

Let’s start with what we know. Luke Voit’s distribution of home runs per fly ball is hilariously high. In 2018, more than 40% of all Voit’s fly balls left the park. That will never happen over a full season. In the last two years, 588 players have had at least 300 PAs. The HR/FB% average of that set is 13.8%. Aaron Judge and Christian Yelich have had the two highest rates, at 35% each. Voit’s pace is four and a half standard deviations above the average, and is bound to come down. Barry freakin’ Bonds’ highest recorded HR/FB% was only 28.8%.

Fewer of Voit’s balls, on a per-ball basis, will leave the park in 2019, but his overall fly ball rate overall is probably pretty sustainable. He posted a flyball rate of 36.6%, right on the average across baseball the last two years. This could indicate that Voit could hit even more fly balls going forward, as almost every player in baseball is making it their mission to do just that.

Fewer balls will leave the park on a per-ball bases, but if Voit bumps his overall FB% up to, say 41% - right in the range of players like Didi Gregorius, Yasmani Grandal and Cody Bellinger – his power numbers will still decrease but not nearly so sharply as to make him unplayable. If Voit is working this offseason on optimizing his launch angle, his regression may not be as hard as the conventional wisdom would offer.

There’s also the matter of his strikeout rate. Conventional wisdom holds that he strikes out too much, that all players who strike out at his rate – 26.4% with the Yankees – are doomed to see their production fall off. Unfortunately, among players who could be considered MLB regulars, strikeouts really don’t affect their production all that much:

There’s almost no correlation between offensive performance and strikeout rate. The correlation coefficient for this spread is -0.08, which is nothing. The R² of this distribution is 0.0073, indicating how hard it is to set a single formula the data fits. This may seem counter-intuitive, but the reason why strikeouts don’t affect performance of MLB regulars is because they produce enough to justify the outs made at the plate. The production is primary, the strikeouts secondary. It makes sense to have your 150 PA defensive replacement be a contact hitter, but players good enough to achieve full-time status reach base enough that it doesn’t really matter how they make outs.

That’s good news for Voit. As long as he can continue to do damage at the plate – which is tied closely to his ability to hit the ball in the air, as discussed above – the strikeouts shouldn’t bother anyone. The last thing that could be a clue to Voit’s sustainability is his Statcast data, the core of modern hitting philosophy.

Fortunately for us, Voit was kind enough to induce exactly 100 batted balls in 2018, providing a nice round cutoff. One of my favorite Statcast metrics is called “barrels per plate appearance”, which basically tries to establish how many plate appearances end with a hitter “barreling” a ball. A “barrel” is a batted ball that has an exit velocity and launch angle consistent with balls that yield a .500 batting average and .1500 slugging percentage.

Really good hits, basically. When you hear terms like “put a charge into one” or how the ball “sounds good off the bat”, you’re really hearing about old school ways of describing barreled balls, and those batted balls generally have an exit velo in excess of 95 mph and a launch angle between 15-45 degrees.

The ability to generate barreled balls is generally the most efficient way to do damage offensively, as they are the batted balls most likely to go for extra bases. Here’s the Statcast table for the top 25 hitters in barrels per plate appearance rate in 2018, minimum 100 batted balls:

Voit may not have been the single best hitter in baseball in 2018, but he may very well have been the most efficient. His plate appearances ended with barreled balls more often than any other hitter in baseball, and this list is peppered with the best and most consistent hitters in the game. Also, Voit’s average and max exit velocity hangs right with the best on the list, as he’s able to hit the ball harder more often. The difference in a guy like Voit compared to say, Tyler Austin, is that consistency in exit velocity. Voit ends a batted ball event with a 95+ mph ball 25% more than a hitter like Austin does, meaning Voit relies less on luck, or positive random deviations from the mean.

The last thing about Voit is the existence of the known unknown. We know that information works both ways. Pitching coaches, analysts and pitchers themselves are familiarizing themselves with what Voit can do, and are working on the best ways to get him out. We don’t know what those ways are, and how effective opposing pitchers will be at utilizing them, and that uncertainty will probably play a bigger role in Voit’s future success than anything else.

Voit fits perfectly into the baseball’s current paradox of information. He’s difficult to project because he has very little prior track record that says he’s capable of sustained success. However, the way he’s generated his short stint of success is generally pretty sustainable, which should make him easier to project. We don’t know how well he responds to pitching adjustments, again because of his lack of track record, but we don’t know what those adjustments are going to be.

In a lot of ways, the answers given to us by the reams of information we have access to just leave us with more questions, and sometimes the same questions. If Voit had this kind of success at the end of the 1960 season, there would be a number of people thinking he was a flash in the pan, a few people thinking he was the real deal, and there’d be a big chunk in the middle without any idea what he really was. At the end of the 2018 season, I think we’re all pretty much in the same boat we would have been in generations ago. We can explain why Voit was so good, but like Martinez, and so many others, we can’t really give a good projection on where Voit is going.