« Weekend Photo | Main | Prepared »

October 22, 2011

Belated Thoughts On Washington's Game 1 PH Decisions

With two on and one out in the 7th inning of Game 1 of the Series, Tony La Russa removed reliever Fernando Salas so LHP Marc Rzepczynski could pitch to lefty David Murphy. Ron Washington pulled the veteran for relatively untested righty Craig Gentry. After Gentry struck out, Washington replaced the pitcher with righty Esteban German instead of veteran righty Yorvit Torrealba, one of three catchers on the roster. German struck out in an especially foolish manner on three pitches.

Fan and media reaction to both moves was mixed at best. Some wanted Murphy to bat despite the wrong-handedness of the matchup. For example:

Most were baffled at German's appearance. German has anchored Texas's AAA squad during the past three seasons but has 79 MLB plate appearances in that span and had never swung a bat in the postseason. Based on my scan of the Twitterverse, many fans see him a defensive specialist, which actually couldn't be father from the truth. While nominally capable of playing just about anywhere on the diamond, German is no better than mediocre at any position and poor at several. When Khalil Greene bailed out of the backup infielder spot in March 2010, German was among the potential replacements. At shortstop in a "B" game, he literally played himself out of the role in the space of two innings. I witnessed the carnage.

Texas under Washington has favored a defense-first fifth infielder, and German doesn't qualify. Instead, German offers on-base skills (career .280/.359/.386) and speed. Per Baseball Reference, German has produced 4.3 wins with his bat (in 1,170 PAs, equivalent to two years for a regular) and lost 3.0 with his glove. If German could defend just a little better, he'd be on a two-year, $6 million contract instead of a miserly split deal ($200k in the minors, $600k in the Majors in 2010).

So, with that background, does the math confirm Washington's decisions? A formula dating back to the 1981 Bill James Baseball Abstract answers just such a question. See here and here for two articles which tested and confirmed its accuracy.

Conceptually, the math says: Given a player's performance against the league, he's going to hit better against below-average pitchers and worse against above-average pitchers. This isn't rocket science. In the specific case of Murphy vs. Rzepczinski:

Given #1: Murphy hits poorly against LHPs relative to the league,

Given #2: Rzepczinski absolutely kills lefties,

Therefore: Against Rzepczinski, Murphy is going to hit even worse than he usually does.

Again, an obvious observation. But the math can be used to suggest optimal matchups. Was there a significantly better choice on the Texas bench than Murphy? If Washington is thinking about replacing Murphy with a righty, but La Russa might counter with a righty reliever, is the resulting matchup so bad that Washington may as well stick with Murphy vs. Rzepczinski?

In this instance, it appears Rzepczinski was going to face whomever Texas offered. I gathered batting averages for Murphy, Gentry, German, Torrealba, and the leagues versus lefties for 2011 and the last three years, and I did the same for Rzepczinski against both lefties and righties. Using one-year averages, here's how Murphy stacks up against Rzepczinski:

Murphy versus lefties: .215
Rzepczinski versus lefties: .166
MLB average, lefty on lefty (excluding pitchers batting): .239

Expected average = (.215*.166/.239) / { ((.215*.166/.239) + (1-.215)*(1-.166)*(1-.239) }
Expected average = .145

You don’t get much worse than that.

Here's the expected batting averages against Rzepczinski for Murphy, Gentry, German and Torrealba based on one-year and three-year stats:

How bad is the Rzepczynski-Murphy matchup? Pretend you're managing the Cardinals and are given the opportunity to select any St. Louis pitcher and any Texas batter (not just those listed above) for the purpose of getting an out. Who would you chose?

Mathematically, the answers are Rzepczynski and Murphy. No other combination produces a lower expected batting average. (Okay, technically it's not Rzepczinski but Eduardo Sanchez, who produced an .005-lower opposing average against lefties than Rzepczinski. But Sanchez had a stupid-low .204 BABIP combined with far fewer strikeouts.) An Octavio Dotel vs. Nelson Cruz is nearly as bad (purely in terms of expected batting average, not OBP or slugging) and is probably going to come into play at some point in the Series.

If I'm managing Texas in that situation, I PH Gentry without hesitation. The three-year stats bring Murphy and Gentry closer together (.201 vs. .240), and there's an argument to be made for Murphy's "proven playoff composure." On the other hand, in limited exposure Craig Gentry seems quite composed himself, hitting 4-10 and a walk before his Game 1 appearance. As for the pitcher's spot, that's trickier. The math says German, but he hadn't faced live competition in weeks and had never hit in the postseason. Honestly, I probably would have gritted my teeth and sent Torrealba to the plate. No option is palatable, as Texas's bench isn't populated with strong bats. Regardless of what I or any other armchair manager would have done, I think you have to give Washington a ton of credit for batting German. It's a gutsy, inspired decision, despite the dismal results.

Husker Du, "From The Gut," from Everything Falls Apart, 1982.

Posted by Lucas at October 22, 2011 02:58 PM