Do Batters Learn to Narrow Their Splits?

Every position player in baseball is, in a sense, two distinct players on offense. He is the batter who faces right-handed pitchers and the one who faces left-handed pitchers. The difference between his performances in those situations—his platoon split—is usually significant and sometimes so stark that he is effectively unplayable in the adverse match-up.

Platoon splits complicate baseball strategy. Opening lineups are different depending on the handedness of the starting pitcher. Aggressive use of the bullpen, probably personified best by Tony LaRussa, can force opposing managers to burn players on an already short bench, leaving them without the resources to react to the next change on the mound.

How much easier strategizing would be if those platoon splits were narrow or nonexistent. How useful it would be for batters to learn how to narrow their splits, to become more consistent players who can’t be exploited as easily by a certain type of pitcher.

Can players learn to do that? Do they, as their careers progress and they gain experience, find ways to close the gap in their batting game?

Setting Up, Plus a Tip

The question requires a long-term view to answer and also requires one to decide how to measure that long term. Is experience best counted by years in the majors, facing the best competition and working with the best coaches to improve one’s batting? Or is age a better measure, since active players are always (barring injury) playing baseball, either at the majors or at lower levels, and piling up chances to improve themselves? I decided to look at the matter from both angles.

Another decision was how to measure batting performance. wRC+ is a preferable metric to OPS+, but records of splits for the former go back only to 2002, leaving some early years in my data set uncovered. I gathered data for both and will give results for both.

For my subjects, I chose position players who debuted in the majors between 1998 and 2003. Their careers needed to span at least 10 years, though they didn’t have to be in the majors every year of that span. This roped in a few marginal players who had intermittent and widely spaced stretches in the bigs. I accepted this, as it allowed me to look at platoon evolution by age and by years in the majors a bit more separately.

I gathered data on players’ platoon splits every year they played through 2017. This cut off several careers before their actual ends, for players like Miguel Cabrera, Adrián Beltré, José Reyes, and Albert Pujols. The lost data is thankfully minor and also comes in categories of age and experience where the sample sizes are so small I did not strain to draw any conclusions from them.

There were other occasional hiccups in the data. Shane Victorino, who debuted in 2003 as a switch-hitter, began batting exclusively right-handed midway through 2013. I cut off his data at 2012. Likewise, Ichiro Suzuki and Hideki Matsui, transplants from Japan, had long experience at a high level of competition before joining coming to America’s major leagues. How high that level is remains a contentious issue. To prevent my having to guess at a proper adjustment, I treated their NPB careers as equivalent to the minors.

Before I descended into the platoon numbers, I found something tangential that interested me. I had lists of long-tenured position players and the ages at which they debuted in the big leagues. This means I could calculate an average age at which players destined to play for a long time reached the majors. This I did, and I learned a lesson. If you want to break into the bigs young and play a long time, learn to bat both ways.

There are caveats. Giving early-20s debuts to Suzuki and Matsui—reasonable for them if they’d gone directly to the majors—would lower the debut age for lefties by a few tenths of a year and raise the career length likewise. Adding in Victorino’s righty-batting years adds a tenth of a year to switch-hitters’ career length.

Even after these adjustments, switch-hitting remains the superior career foundation by both measures. Broken down by year, switch-hitters had the earliest average debuts in five of the six rookie seasons I examined; the advantage didn’t arise from one fluke year. It’s not shocking that an extra skill set should make a player likelier to rise faster and endure longer, but as with so many things in baseball that seem to make natural sense, it’s better to have the numbers backing you up.

In my main work, I tracked the size of players’ platoon splits each year they played, and for one-way batters, which type of pitcher they batted better against. For switch-hitters, tracking the direction of the platoon advantage made less sense. They would not share the same “dominant” handedness, and some splits would cancel out others. For them, I took the absolute value of the split.

I weighted each player’s season by the harmonic mean of his plate appearances against left-handed and right-handed pitchers. This was needed in order to avoid players used in extreme platoon roles from skewing the figures, especially since the short ends of their platoons could have extreme batting values from the tiny sample sizes.

(To help to orient readers, a batter who faced lefties 200 times and righties 500 times—a very full season of hitting—would have a harmonic mean of 285.7 plate appearances. A year with 100 and 400 would come out to a harmonic mean of 160; 10 and 100 would produce 18.2 as its harmonic mean. The lower value—usually meaning plate appearances versus southpaws—has the greater influence on the result.)

To show results, I’ll display graphs of the data by handedness and year/season method, showing the platoon splits as measured by OPS+. (wRC+ will diverge early on, but it becomes identical to OPS+ in time.) A negative value to the splits means the batters do better against left-handed pitching; positive split values mean the batters hit better against righties. The split values for switch-hitters are absolute, which doesn’t indicate superior hitting against lefties or righties. Rather, the values simply show the magnitude of whatever split there is, whichever way it faces.

Be warned: not all points on the graphs are created equal. Both ends of the line will suffer from small sample sizes, which would be very complicated and likely confusing to show on the charts themselves. This is true in all cases, and even more so in one case I will highlight when it arises.

To handle this problem, I will measure the trends not only for the full data set but for narrower sets requiring larger samples, per year of age or season played. My cutoffs will be at 500, 1,000, and 2,000 harmonic-mean plate appearances. That can be roughly approximated as two, four, and eight full seasons of batting. This will lower the number of years covered, from a range of 17 to 23 for the full set down to 10 to 15 for the most restricted one.

I’ll cover the platoon trends in the text while giving tables to show the direction and magnitude of the trend lines more precisely, both for OPS+ and wRC+.

What Do They Learn?

We’ll begin with right-handed hitters, counting by seasons in the majors.

The sharp widening of the splits in very late career years exaggerates the apparent overall movement. At the 500+ level of sample size, widening is much reduced, and at 1,000+ or 2,000+ becomes a marginal narrowing. Using wRC+ numbers, things swing a little more toward narrowing, starting at the 500+ level and growing decidedly stronger at 1000 and especially 2000.

This hint of improving splits with experience, however, disappears when we look at righties by age.

 

(Notes: Numbers are in points of wRC+/OPS+ per year or season. “N” stands for splits narrowing; “W” stands for splits widening.)

While the small-sample reverse splits at age 19 skew the full-sample numbers, even at the higher sample thresholds the trend is toward widening with age. The magnitude is less pronounced with wRC+, but the pattern is the same. The “maybe” seen by season becomes a “no” by age.

That could be the pattern: narrower platoon splits coming through major-league experience rather than the more generalized experience of age. If we look for this pattern in the numbers for left-handed hitters, though, we find the opposite.

Every sample size, with both measures of offensive output, show lefties’ platoon splits widening season by season. The larger samples show the greater widening. The sudden plunge in Year 16 is an outlier that doesn’t change the overall numbers.

Looking by age shows a dramatic difference—and a different kind of outlier.

 

(* The Ichiro effect)

Here we see the danger of working with small samples, taken to its extreme. The huge swing into reverse platoon splits beginning at age 39 comes about due to one player: Ichiro Suzuki. Suzuki had a career OPS+ 11 points better against lefty pitchers than against righties, and this reverse split was much stronger in the early and late seasons of his North American career. (I’d love to know what his splits were when playing with the Orix Blue Wave.)

Take the Ichiro-dominated years out of the equation, and the massive tilt flattens out. There’s some narrowing at the 500 level, some widening at the 1,000 level, and nothing at 2,000. The pattern holds for the wRC+ numbers, except that the 2,000+ sample shows widening.

Lefties and righties end up something like mirror images of each other, which is ironically apt.

The pattern by season for switch-hitters comes close to following that for righties. An overall widening seen with all samples tends to switch to narrowing with larger samples, except that the 2.000+ level turns back to widening. Looked at by age, the switch-hitters mirror the righties again.

 

Every level of sample size shows the platoon split growing.

This completes the pattern for all three types of hitter. By one measure, there is some indication hitters can learn to narrow their platoon splits with time. By the other, the split does nothing but get bigger. The overall result is the same. Not only is there generally little evidence batters learn how to hit better from their weak side, the preponderance of evidence is that the split grows with the combination of age and experience.

What Have We Learned?

My opening hypothesis was so wrong I am forced to rethink the matter entirely. I thought players would want to close a hole in their game, become more complete performers. Maybe the reverse is true. Maybe players are incentivized to strengthen the better parts of their games to give them a greater peak performance. Improving their ceiling might keep them in the game longer, if only in part-time roles, than raising their floor.

It might also be that players put better effort into the type of batting that gives them more positive and satisfying feedback. The platoon split would effectively be a rut, one that deepens as it’s worked through year after year—though to a small extent compared to its original size.

It’s also possible that using a sample of long-term players produces a kind of survivor bias. Those hitters who scrape along for a couple extra years as platoon players may tip the numbers, even with their effect diluted by my methodology. (The right-handed short side of the platoon ironically would have the greater impact, as their “bad side” PAs would tend to be greater than those of lefties and boost the harmonic mean.)

Whatever the explanation, my question is answered. Hitters do not narrow their platoon splits over time. Whether it is because the disparity in skills is too physically ingrained to budge or because they are encouraged to do things a different way, the numbers do not tell. If some other baseball writer with a knack as an interviewer wants to ask a few players how they deal with their platoon splits, I would be very interested to hear about it. I—we—might learn something.

References and Resources

• Baseball-Reference, for OPS+ data
• FanGraphs, for wRC+ data