Is the Dodger Rotation too Left-Handed?

Even after the signing of Kenta Maeda, the Dodgers are projected to have an extremely left-handed starting rotation. As of now, the starting pitcher depth chart has Clayton Kershaw, Scott Kazmir, Alex Wood, Brett Anderson, and Hyun-jin Ryu‘s bionic shoulder along with Maeda. It’s not all left-handed pitchers anymore, but it’s still five of the top six. This has created some understandable concerns about the Dodger starting staff being too left-handed, and thus overly vulnerable against right-handed batters.

There is absolutely merit to that concern. However, the actual pitchers on the staff are important too. If the Dodgers have left-handed pitchers who are less vulnerable to platoon splits, then their handedness matters less. There are two ways to look further into this: each players’ past performance, and their projected splits going forward.

Past Results

This method of investigation is pretty straightforward. How have the Dodger lefties fared against left-handed and right-handed batters in the past? Below are a selection of platoon statistics. K%, BB%, and K%-BB% are self-explanatory. GB% is a quick shorthand for contact quality. I’ve also included wOBA, which is the end result that actually matters in a huge sample, but for platoon splits it needs a lot more time to stabilize. These stats are from the years 2013-2015, a period when all of the Dodger lefties were active (except Ryu last year).

Before getting to the individual pitchers, it’s important to know how the entire league has done against left-handed pitchers. This gives a baseline to compare the Dodger pitchers against.

League Average

TBF		K%		BB%		K%-BB%		GB%		wOBA
LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB
44669	103133	22.2%	18.9%	7.3%	9.1%	14.9%	9.8%	48.4%	43.6%	0.294	0.327

Against southpaws, left-handed batters strike out more, walk less, hit more ground balls, and have worse overall results than they do against right-handed pitchers. The magnitudes of the differences are important to keep in mind: 3.3% difference in strikeout rate, 1.8% difference in walk rate, 5.1% difference in strikeout/walk differential, 4.8% difference in ground ball rate, and 0.033 points of wOBA.

Now, onto the Dodger pitchers.

Clayton Kershaw

TBF		K%		BB%		K%-BB%		GB%		wOBA
LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB
516	2031	37.4%	28.5%	4.1%	5.1%	33.3%	23.4%	45.1%	50.1%	0.226	0.234

Here’s the thing about Clayton Kershaw: he’s Clayton Kershaw. The peripheral splits are bigger than the average left-handed pitcher, but Kershaw has managed to get a lot more grounders from right-handed batters and the resultant wOBA split is basically non-existent. Kershaw kills left-handed batters, but he kills right-handed batters too.

It’s also important to realize that managers have already gone out of the way to not let lefties face Kershaw. On average, a left-handed pitcher will face a right-handed batter 70% of the time. Over the last three years, Kershaw has faced right-handed batters 80% of the time. Kershaw’s ridiculously dominant numbers over the last few years already account for facing the worst platoon situations managers can throw at him. It’s yet another reason why he’s so great.

Scott Kazmir

TBF		K%		BB%		K%-BB%		GB%		wOBA
LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB
556	1656	19.8%	22.4%	4.9%	7.8%	15.0%	14.6%	53.7%	46.1%	0.299	0.304

Newcomer Scott Kazmir has pretty weird splits over the last three years. He has actually struck out fewer left-handed batters than right-handed batters. However, his walk rate split has been larger than the league’s average, and the resultant K%-BB% split is basically even. Kazmir seems to get worse contact quality against left-handed batters like one would expect, and the overall wOBA split results in something just about even. Overall, the complete package looks mostly platoon-neutral, or at least with a smaller split than average.

Alex Wood

TBF		K%		BB%		K%-BB%		GB%		wOBA
LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB
463	1359	24.7%	20.0%	5.2%	7.9%	19.5%	12.2%	53.7%	46.1%	0.267	0.316

Given Alex Wood’s famously low arm slot, one would expect his splits to be larger than league average. So far through his career, this has been the case. His K%, BB%, K%-BB%, GB%, and wOBA splits are all more extreme than the average left-handed pitcher.

Brett Anderson

TBF		K%		BB%		K%-BB%		GB%		wOBA
LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB
320	820	17.9%	14.3%	8.5%	7.4%	9.4%	7.0%	70.7%	62.6%	0.320	0.320

Brett Anderson isn’t a power pitcher, relying more on bad contact and grounders to get batters out. As a result, one would expect that his split would be mostly neutral. Anderson’s strikeout split is smaller than average and his walk rate split is reversed. However, he has gotten more grounders against lefties than right-handed batters. The end result (which is the smallest sample of any pitcher listed), has been a split that has been dead even over the past three years. The peripherals show that it might not always be that way, but it makes sense given his pitching style.

Hyun-jin Ryu

TBF		K%		BB%		K%-BB%		GB%		wOBA
LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB	LHB	RHB
346	1068	18.2%	21.6%	5.2%	5.6%	13.0%	16.0%	47.3%	49.8%	0.311	0.285

Ryu is interesting because his splits have been almost entirely reversed so far, which is extremely rare for left-handed pitchers. He has struck out more right-handed batters than left, has no meaningful walk rate split, and gets more grounders out of right-handed batters. The end results are reversed, too. This is quite unusual, and probably wouldn’t continue to this degree (small sample, and unknowns relating to how he would recover from his surgery), but it seems pretty notable.

Projections

There are some flaws with looking at the above numbers alone. It’s important to know how a pitcher has done in the past, but platoon splits are extremely volatile. While peripherals will stabilize faster than wOBA, it still takes years to have confidence in just how large a player’s split is.

Jared Cross, one of the people behind the Steamer projection system, attempted to address this in an article at the Hardball Times last year. The article is extremely math-heavy, but it’s also very interesting. Here’s a summary of what factors Cross found that influenced splits the most:

• Pitcher handedness: left-handed pitchers tend to have more severe splits than right-handed pitchers.
• Pitch usage: Pitchers who throw more sliders have a higher chance to have severe platoon splits. Pitchers who throw more change-ups and curveballs tend to have smaller splits.
• Arm angle: Lower arm slots tend to lead to higher splits.
• Past performance: To an extent, depending on how other factors have changed.

Overall, Cross found that the median projected platoon split for left-handed starting pitchers with over 100 innings pitched last season (34 total) was .028 points of wOBA worse against right-handed batters than against left-handed batters.

We’ve already talked about past performance, but the usage, arm angle, and forecast split using these methods are all worth looking at. Before listing these, it’s important to note that the projections listed below are from last July (with the exception of Ryu, who I will discuss in his section). The overall projected split wouldn’t have changed much since the article, unless one of the other factors listed above has changed significantly. Cross is working on getting these splits updated daily but given the amount of factors taken into account that sounds like an arduous process.

Clayton Kershaw

• 2015 pitch usage: 53.8% four seam, 27.6% slider, 18.2% curve, 0.5% change. Obviously, Kershaw uses a lot of sliders, but he mixes in more curves than the average pitcher as well.
• Arm angle: 67 degrees (very over the top)
• Projected splits: .009 wOBA worse vs RHB
• Overall: Kershaw is projected to have the sixth-smallest projected split among last year’s left-handed starting pitchers. His arm angle is the highest among that group, which offsets his high slider usage. Basically, he’s Clayton Kershaw and is mostly immune to platoon splits.

Scott Kazmir

• 2015 pitch usage: 31.7% four seam, 26.0% two seam, 12.8% cutter, 3.9% curve, 7.4% slider, 18.1% change. Kazmir mixes in a lot of different pitches, though his low curveball usage slightly increases his projected split.
• Arm angle: 49 degrees (three quarters)
• Projected splits: .024 wOBA worse vs RHB
• Overall: Kazmir is forecast to have nearly a median split. This is counter to some of his past performance (neutral), but his average arm angle and slightly unfavorable pitch usage for platoon use make up for it.

Alex Wood

• 2015 pitch usage: 62.3% four seam, 21.9% curve, 15.8% change. This is a very good pitch selection to reduce platoon splits.
• Arm angle: 34 degrees (very low)
• Projected splits: 0.046 wOBA worse vs RHB
• Overall: Wood is projected to have eighth-worst split among the 34 pitchers in the sample group. He has a favorable pitch usage (no sliders), but he also has the second-lowest arm angle among the pitchers sampled. The projection matches up with Wood’s previous platoon split, which isn’t really a surprise.

Brett Anderson

• 2015 pitch usage: 20.4% four seam, 32.2% sinker, 12.0% curve, 25.8% slider, 9.6% change. High slider use, but high curve and change use as well.
• Arm angle: 46 degrees (three quarters)
• Projected splits: 0.009 wOBA worse vs RHB
• Overall: Anderson is projected to have the fifth-lowest split among the 34 pitchers sampled. His pitching style and pitch usage, along with previous neutral performance lead to such a small projected split.

Hyun-jin Ryu

Ryu is a special case because he was excluded from Cross’ projections for the remainder of 2015, since he was already out for the season when the article was published. However, he was nice enough to calculate Ryu’s arm angle and split projections (without the “past performance” regression applied).

• 2014 pitch usage: 52.3% four seam, 13.4% curve, 16.1% slider, 18.2% change. Slider use was trending down as the year went along, and high usage of more favorable platoon pitches.
• Arm angle: 57 degrees (high three quarters)
• Projected splits: 0.021 worse vs RHB, prior to regression
• Overall: Ryu is projected to have a lower platoon split than average. Additionally, that forecast is before applying regression based on his past performance, which has been a reverse split. Part of the projection is based on such a high arm angle, but it’s fair to wonder if it will remain the same upon his return.

Summary

Since that information is hard to read in bullet form, here’s a summary of the projections:

	League median LHP starter	Kershaw	Kazmir	Wood	Anderson	Ryu
wOBA split (RHB-LHB)	0.028	0.009	0.024	0.046	0.009	0.021*

*prior to regression

Four of the five left-handed starters have projected splits below the median, and two are projected to be among the league’s best in being platoon neutral.

The magnitude of the overall projection can be examined further using a weighted average of projected 2016 innings totals. Unfortunately, the Fangraphs depth chart, the usual go-to for information like this, has not been updated to include Maeda’s signing. As such I had to arbitrarily assign projected IP totals. Here are my best estimates on what the chart will look like after Maeda is added: Kershaw 225, Kazmir 160, Maeda 152, Anderson 150, Wood 130, Ryu 75, others (McCarthy, Bolsinger, minor leaguers) 85. That’s a maybe optimistic total for Ryu, though it’s about a third less than the pre-Maeda projection. For Ryu, I used a total split of 0.018, accounting for some of the “regression” which was not applied to the provided projection.

The weighted average projected platoon split of the Dodgers’ left-handed starters using those innings totals is 0.020 wOBA worse against right-handed batters than left-handed batters. As a group, the Dodger lefties are projected to have a 30% smaller platoon split than median.

However, that number is incomplete. We can go further, adding the platoon splits of the right-handed pitchers in the rotation. It seems safe to “assign” Maeda a league-median split for RHP (0.010 wOBA better vs RHB) because he has a 3/4 arm slot and throws a lot of sliders. McCarthy’s projected split is 0.013 wOBA better vs RHB in 30 innings. Bolsinger is projected to have a 0.002 wOBA reverse split in 25 innings.

Combine all of that, and the Dodger rotation as a whole is projected to allow a platoon split of 0.017 wOBA better against LHB than RHB. That’s still a potential weakness, but it’s not as much of one as what would be otherwise expected out of the starting staff as a whole.

Conclusion

Yeah, the Dodger starting pitcher depth-chart is very left-handed. That carries some extra risk compared to an inherently balanced rotation. However, when looking at projected split magnitude, the Dodgers carry two of the six pitchers who are projected to be the most neutral (Kershaw, Anderson). Additionally, they carry two more starters who are projected to limit splits better than the typical left-handed starter (Kazmir, Ryu). The only left-handed starter the Dodgers have who is projected to have a high platoon split is Alex Wood. These numbers are also backed by the players’ previous performances.

More importantly, though, the Dodger pitchers are also good at pitching. Signing Zack Greinke would have helped with the lopsided handedness of the rotation. However, once the big three pitchers came off the board, the team did a pretty good job assembling a core of a competent staff. As a result, the risk of the lefty-heavy rotation is reduced even further, and when combined with the small splits, the composition doesn’t seem to be worth worrying about very much.