Is Max Kepler Figuring Out Lefties?

The Twins' rebuild began with the disastrous 2011 season, and it has yet to truly produce a contender. Long-term improvement and short-term overperformance have taken the team to winning records in two of the last four seasons, as well as a Wild Card berth in 2017, but the Twins haven't surpassed averageness.

A significant reason the Twins have failed to break out thus far has been a struggling group of young players. Eddie Rosario and José Berríos overcame initial difficulties to become the team's most valuable hitter and pitcher, respectively, but the Twins are still waiting on a few others to find their ways.

Byron Buxton was MLB.com's top overall prospect in 2013, 2014, and 2015. Though he plays center fielder like a Gold Glover, Buxton has struck out too much and repeatedly hit the Disabled List.

Miguel Sanó, ranked seventh on the 2014 list and third the year before, has phenomenal power and takes plenty of walks. But he's been unable to make contact consistently, reported to spring training overweight, missed time due to injuries, and was accused of assault last winter.

Jorge Polanco has had moderate success since becoming the starting shortstop in late 2016, but a PED suspension kept him off the field for half of 2018.

And then there's Max Kepler. The German has provided excellent defense in right field and shown enough flashes of offensive prowess to be a decent regular, but one thing held him back in his first couple of seasons.

Through 2017, Kepler hit .261/.334/.475 against right-handed pitching. By wRC+, that's 12-percent better than the average MLB hitter. For comparable hitters in overall production, see Derek Dietrich and Jurickson Profar last year. It's not exactly what you want out of a right fielder, but it's fine.

Meanwhile, Kepler hit .176/.242/.279 against left-handed pitching. That's 64-percent worse than the average MLB hitter. By comparison, Chris Davis' season, the 30th-worst of all-time by wRC+, was 10 points better. That's absolutely disastrous for any major-leaguer.

Being worse against same-handed pitchers is common. Being so abhorrent against them is not.

Platoon splits take a little bit of time to stabilize, sure, but the sample Kepler provided was large enough to cause legitimate worry: Was he doomed to become Seth Smith?

Smith was a reliable platoon hitter for years, but the Twins undoubtedly want more out of a player they expect to start every day at least until he becomes a free agent. In his first two seasons and change, Kepler didn't show any evidence he could start every day.

This season, however, something weird happened.

Kepler started hitting lefties. In fact, he hit them better than he did righties, if only by a little. What exactly happened here?

There's a fairly simple explanation, but I want to get there gradually.

First: Does this look like a fluke? Sometimes, even over a full season, statistics can be weird.

Most of the time, this weirdness comes from batted-ball data. Though batters have control over what kinds of batted balls they produce, they have little control over the kinds of results they do. Unless they only hit home runs, their batting averages are at the whims of opposing defenses.

For that reason, sabermetrics-oriented observers look at batting average on balls in play (BABIP).

Generally, over a whole career, BABIP makes sense. If you hit a lot of line drives, your BABIP will be higher. If you hit a lot of popups, your BABIP will be lower. If you're fast, your BABIP gets a slight boost because you leg out a few extra infield singles. With time, it makes sense.

However, in small samples, BABIP can fluctuate wildly.

For example, in 2014, rookie Danny Santana hit .319, more than 20 points better than he had ever hit in professional baseball. The reason: more hits were falling than usual: his BABIP was an absurd .405. The next year, his BABIP predictably fell to a far more normal .290; his overall batting average fell to .215.

The lesson: If your BABIP is higher in a small sample than it normally should be, you likely aren't as good a hitter as your batting average says you are – assuming, of course, you remained roughly the same kind of hitter during that time. If you change your swing or approach, you might see a substantial change to your BABIP that can be explained by more than just luck. BABIP, then, works as a first test of flukiness.

Applying this to Kepler: What we need to find out is whether his batted-ball results are abnormal.

We'll look at BABIP first.

Well, that certainly looks weird to me. Kepler clearly must have gotten the weirdest combination of bounces possible that caused his splits to totally reverse, right?

We cannot say until we investigate further.

We'll look next at Kepler's batted-ball rates. Did his profile substantially change?

You bet it did. Kepler put far more balls in the air in 2018, and especially against lefties, he hit the ball harder. He went from a ground ball hitter to seventh among qualified hitters in overall fly ball rate. That's bound to create different results than he got previously.

To find out how different those results should have been, I'm relying on Statcast data from Baseball Savant. The site has detailed batted-ball data, including statistics meant to stand for the outcomes you would expect based on those data: expected batting average (xBA officially, but I say xAVG), expected slugging percentage (xSLG), and expected weighted on-base average (xwOBA).

(A quick digression on wOBA: It's the stat on which wRC+, cited above, is based. It essentially measures the value of a player's whole offensive output and expresses it as a number on an OBP scale; a good wOBA looks like a good OBP, and a bad wOBA looks like a bad OBP. It's worth noting that wOBA is not adjusted like wRC+ is, so if you play in a hitter-friendly park, your wOBA will be inflated. For more on wOBA, see this page from the FanGraphs glossary.)

I've included links explaining all three, but the short of it is that the "x" stats (which I'm calling xResults) look at what kinds of results different kinds of batted balls normally generate, and then it applies those general results to a hitter's tendencies to determine what his results "should have" been. If the differences between his actual results and xResults are significant, then something weird might have happened.

Here, then, are the differences between Kepler's actual results and xResults. A positive number means he did better than expected; a negative one means he did worse.

There aren't many differences. Kepler seemingly got slightly worse results than expected against righties, and his slugging percentage against lefties is somewhat higher than expected, but the differences are not so dramatic that we can call Kepler's season fluky.

It makes sense: Against lefties, Kepler started hitting the ball more squarely and harder, so his batting average on balls in play improved. Against righties, he got under the ball so much more that it killed his batting average. (Fly balls fall for hits less often than ground balls and line drives.) Though he probably didn't hit the ball as poorly against righties as his results indicate, he did hit it less well.

What's behind this change? To find out, let's look at Kepler's swing.

Here is his swing from the side. (The top GIF is from 2017, and the second is from 2018. Each is via the YouTube channel Baseball Swingpedia.)

And now from the front:

It looks like he made his leg kick quicker and less pronounced, but only by a little bit. I don't want to discount the possibility that this difference was a conscious tweak that helped change Kepler's profile, but it's not a considerable difference.

Still, the front view might reveal something. Let's look at his batting stance.

Kepler's hands are in the same spot, but his hips are more open than before. It may be just a meaningless platitude, but supposedly, open stances allow hitters to see the ball better.

If Kepler is seeing the ball better because of his stance, that doesn't explain his worse batted-ball data against right-handers. But it might help explain this:

(Via Baseball Savant)

As a batter, you want to make contact when you swing, you want to swing at strikes, and you want to not swing at balls. Regardless of handedness, Kepler did better at all three of those things than he did in his previous two full seasons. The results are undeniably positive:

I don't know if it's because of his revised stance, but Kepler is definitely seeing the ball better against either kind of pitcher. And when you see the ball better, not only do you take more walks and fewer strikeouts; you probably make better contact.

After entering the majors a hopeless case against southpaws, Kepler posted a 2018 that showed he can be an average hitter against them. Does that mean he'll go without such a massive platoon split from now on (or even have a reverse-platoon split)?

Not necessarily. We still have far more data suggesting he is worse against lefties than he is against righties. Before we say for sure what kind of hitter he is now, we need to see him play more.

However, there is still strong evidence this change is real. Based on how different his approach was this season, I believe Kepler can at least avoid being a black hole in the lineup against left-handers, which is far more valuable than a strict platoon guy.

Assuming he can straighten out his extreme-fly-ball tendencies against right-handers, he can be a reliable middle-of-the-lineup hitter for the foreseeable future.

All data, unless otherwise noted, are via FanGraphs.