Revisiting The Second Base Platoon

gordon_2014-03-08

In February, I wrote about the potential for a platoon at second base after Don Mattingly stated that it was a possibility. I was a bit dismissive of the idea, citing the lack of combinations that made sense based on past hitting performances. My conclusion was that it was fairly likely that Alex Guerrero would start the season at second, so the platoon didn’t matter much.

Now that it’s pretty clear that Alex Guerrero will start the season in the minors, it’s worth revisiting the platoon idea. The Dodgers seem convinced that they should use Dee Gordon in some capacity, and he has hit just .221/.267/.232 against left-handed pitching in 204 plate appearances. Finding a platoon combination that reduces that impact is important.

Assuming that Guerrero is demoted to the minors, the Dodgers have three players capable of playing second base: Dee Gordon, Justin Turner, and Chone Figgins. What would a platoon among these players look like? First, I took each candidate’s past performance and split up their batting runs against each handedness of pitcher. Given a full season of 600 plate appearances, 173 will be against left-handed pitching and 427 will be against right-handed pitching (using the league-wide plate appearance ratio from last season). The batting value is also park-adjusted to account for each player’s different offensive environment.

Since Dee Gordon adds a bit of offensive value with his speed, base running value is also included in the platoon value calculations. Some scaling is applied* to get base running value against each type of pitcher. The overall offensive value is a sum of batting runs and base running runs.

 Table 1: Platoon performance using past hitting only
 Player Batting Runs/173PA vs LHP Batting Runs/427PA vs RHP Base running runs/173PA v LHP Base running runs/427PA v RHP Offense runs/173PA v LHP Offense runs/427PA v RHP
Dee Gordon -10.85 -6.58 0.83 2.06 -10.02 -4.51
Justin Turner -3.43 0.00 -0.13 -0.19 -3.55 -0.19
Chone Figgins (career) -3.04 0.00 0.32 0.72 -2.73 0.72
Chone Figgins (2010-2012) -4.95 -15.50 0.16 0.33 -4.79 -15.17

The first thing that sticks out is how awful Dee Gordon looks against lefties. Even after accounting for his baserunning, he’s worth -10.2 runs on offense. Chone Figgins is the best option on both sides, but only if his entire career is included. If only the last four seasons are used (sensible, given his age), then Justin Turner is the best option on both sides of the ball. Choosing Dee over Turner on the left side of the plate costs 6.47 runs, and choosing Dee over Turner on the right side costs an additional 4.32 runs.

The method used to find platoon value in table 1 has a flaw. It only looks backwards. It doesn’t take into account the instability of platoon splits or projected future performances. Luckily, Fangraphs outlined ways to apply regression to future platoon split forecasts using methods found in The Book. If we use those methods for the three options and the mean projected wOBA for each player from Steamer, we get slightly different results:

Table 2: Platoon performance using split regression and Steamer forecasts
 Player Batting Runs/173PA v LHP Batting Runs/427PA v RHP Baserunning runs/173PA v LHP Baserunning runs/427PA v RHP Offense runs/173PA v LHP Offense runs/427PA v RHP
Dee Gordon -8.17 -7.49 0.50 1.50 -7.68 -5.99
Justin Turner -1.46 -7.43 -0.15 -0.27 -1.61 -7.70
Chone Figgins -7.68 -12.41 0.20 0.24 -7.66 -12.17

The adjusted numbers are pretty interesting. Justin Turner’s reverse platoon split flips**, predicting that he will hit better against left-handed pitching than right-handed pitching. The change in methodology results in a clear platoon pairing: Justin Turner against left-handed pitching and Dee Gordon against right-handed pitching. Choosing Turner over Gordon against lefties results in six fewer offensive runs lost over the course of a season.Choosing Gordon over Turner against right-handed pitching results in 1.71 fewer runs lost. Figgins does not appear to be viable in either half of the platoon based on his Steamer projected wOBA.

The methodology isn’t perfect. Steamer is a very good projection system for batting value, but it is fairly conservative on baserunning value estimates, which hurts Gordon a bit. Additionally, this methodology completely ignores defensive value. This is intentional, since we know almost nothing about how Gordon will handle second. Defensive metrics don’t like Turner’s play at second, but the sample is too small to come to any definitive conclusions.

I really like it when data surprises me. Before this article, I was convinced that Turner would be a better choice for second than Gordon on both sides of the plate. After doing the math, using Gordon against right-handed pitching makes a lot more sense. However, using him against left-handed pitching is a mistake. As I was writing this, news broke that Turner will be playing in tomorrow’s game against lefty Wade Miley. Maybe the platoon talk wasn’t just talk. In any event, it’s probably only temporary, until Guerrero gets a bit more playing time in the minors.

Footnotes:

*Baserunning value scaling: There is a penalty for stolen base success rate if a left-handed pitcher is on the mound. That was accounted for when scaling stolen base value (wSB on Fangraphs) to apply to platoon values. The scaling factor for wSB was the percentage difference in overall stolen base success rate against each type of pitcher. Results also had to be scaled based on batting success rate (more times on base means more positive value from baserunning). In order to account for this, total baserunning value (wSB and UBR combined) is scaled by platoon OBP changes.

**Reasoning for Turner’s platoon split flip: Justin Turner has had a reverse split (hits better against same-handed pitching) during his career. However, right-handed batters show less platoon variation than left-handed batters, so the Fangraphs/Book regression method gives a higher weight to the league-average split. The regression brings Turner’s split far enough towards the league average platoon split that it passes the break-even point.

This post uses the following statistics:

  • wOBA: Weighted on-base average, a statistic used to calculate overall offensive value by using unique weights for different types of hits. Explanation here
  • wSB: Weighted stolen base runs, a statistic used to calculate the value gained by stolen bases only. Explanation here
  • UBR: Ultimate base running, a statistic used to calculate the value gained by all other events on the bases. Explanation here

About Daniel Brim

Avatar photo
Daniel Brim grew up in the Los Angeles area but doesn't live there anymore. He still watches the Dodgers and writes about them sometimes.