Happy Holidays from Baseball Prospectus! We'll have new content on December 26th.
April 6, 2010
Checking the Numbers
Last week we took a look at SHINOs, switch-hitters in name only, who were defined as batters that swung a mighty broomstick from one side of the plate but fell below the league average from the other. They certainly switched, but the handedness of the pitcher determined if they were to hit like Willie Mays or Willy Taveras.
OK, maybe that is a bit too extreme, but the goal of my research involved determining whether or not certain players should have pulled a J.T. Snow and given up on hitting from their poor side altogether. Overall, very few players fit this bill, which was to be expected for a few reasons, primarily that batters do not face lefties all that much to begin with, which leads to plenty of noise and fluctuation, and an inherent selection bias exists in that, theoretically, only those capable of producing from both sides of the plate are given the opportunity to prove themselves in that capacity.
One of the more fun and thought-provoking questions with switch-hitters revolves around the breakeven of when one should stick to his dominant side. After all, the benefit of switching is derived from exploiting the platoon effect of performing better against pitchers of the opposite throwing hand, but there comes a time where this benefit is negated by what might occur if the hitter were to simply bat from his more productive side in every trip to the dish. Some of these players are easier to diagnose than others; a .350 TAv as a righty and a .025 TAv as a lefty leaves little room for discussion, but the reality is that the question is nowhere near cut and dried either quantitatively or qualitatively.
In the prior piece, my methodology for identifying SHINOs was contingent on the idea that player-splits are best served on a dish garnished with the league split and not that of the individual in discussion. In many cases, this rings true, but switch-hitters follow different rules for the simple reason that they don’t have to hit from both sides of the plate. If Lance Berkman so desired, he could become a permanent lefty. A discussion with Mitchel Lichtman that centered on this idea led to a new line of inquiry: compare the regressed self-split of the individual to what the expected split would look like if he stopped switching. In other words, to determine the point of no return, we need to develop a reasonable estimate of what the split performance might look like for the hitter in question as a pure lefty.
To accomplish this quantification, three preliminary bits of tid must first be known:
Over the last decade or so, and using OPS for ease of use, lefty hitters have a 1.15 ratio; they perform 15 percent better against opposite-handed hurlers. Additionally, recall from last week that the distribution of pitcher handedness tends to be around 72 percent righties and 28 percent lefties. I’m going to use Berkman as the switch-hitter to discuss since he is renowned for mashing from the left while resembling Aubrey Huff from the right; good, not great. Berkman’s career OPS was 967 entering the season, and his estimated lefty-only split is going to rely on the 1.15 league average, widened based on the idea that he hasn’t faced same-handed pitching in over a decade. As will be discussed a bit later on, the qualitative muscle-memory and psychological aspects of ceasing to switch looms large.
Assuming Berkman’s split ratio as a lefty would be 1.21, two equations are set up that must be solved simultaneously:
OPSvRHP*0.72 + OPSvLHP*0.28 = .967
OPSvRHP / OPSvLHP = 1.21
To solve, start by isolating the second equation, setting OPSvRHP = 1.21*OPSvLHP, and then substitute into the first equation:
(1.21*OPSvLHP*0.72) + OPSvLHP*0.28 = .967
This formula can be simplified to:
.8712*OPSvLHP + .28*OPSvLHP = .967
The result informs us that the estimated OPSvLHP is equal to .839. Knowing then that the OPSvRHP has to be 21 percent better than that mark, we can determine that the estimate equals 1.016. What does this information explain? It suggests that, if Berkman were to bat from only the left side of the plate, knowing his overall talent level and the league average ratio from that side, he would OPS the numbers above. To make sense of these numbers we must compare them to the regressed version of Berkman’s actual self-split. Over the same span of time, switch-hitters do not truly favor a particular side of the plate—which makes sense give the idea behind switching in the first place—but the ratio does slightly favor lefties facing righties at 1.01.
Berkman’s career splits include a 1.024 OPSvRHP and a .792 OPSvLHP, producing a 1.29 ratio. Of course that is higher than the prior estimate, because Berkman was not randomly sampled; he was chosen for these purposes due to his proclivity for favoring the left side of the batter’s box. Regressing his ratio towards the 1.01 ratio for the league brings us somewhere around 1.19. Running through the same simultaneous equations setup, his regressed self-split as a switch-hitter is a 1.013 OPSvRHP and a .851 OPSvLHP. If we operate under these posited assumptions, Berkman is still benefiting from the switch, but slightly. If we felt more confident in his plate appearances and regressed less, the results would obviously change. So too would they change relative to the assumptions of his estimated splits as a permanent lefty.
It is logical to assume that not facing same-handed pitchers for over a decade would render his platoon ratio larger than the league average, and that the methods presented above are more than acceptable. However, logic does not always win out; I thought it was perfectly logical that a film starring Johnny Depp and Christian Bale, directed by Michael Mann—all three of whom are in my top ten in their respective categories—would be fantastic, and yet Public Enemies was one of the most disappointing movies I’ve seen in the last decade. Relative to the switch-hitting conundrum, the variable left out of the equations is that we simply cannot trust logic or assume that if Berkman were to stop switch-hitting that his performance as a lefty facing righties would remain perfectly intact.
Aside from the fact that the brain operates contra-laterally—the left side of the brain controls the right side of the body and vice versa—the issue of muscle memory rises to the forefront. Berkman is so used to his current setup that the body knows how to react in certain situations out of habit. If he sees a curveball down and away from a lefty his body knows exactly how to react from the right side. He has never executed this muscle memory as a lefty against a lefty, however, so while he knows what a curveball looks like, reacting to it in this foreign circumstance is a different animal altogether. If he really wanted to stick to the left side, he could train his neural pathways to develop the new muscle memory, but this isn’t a mere matter of making the decision; the re-training could take years.
Taking the league split and penalizing him several percentage points might satisfy quantification urges but it ignores the neuroscience of the issue, which matters a whole heck of a lot more. Now this is not to ultimately suggest that the formulas shown throughout this article are moot, but rather that the neuroscientific aspects need to fuel the assumptions and should be treated with extreme care. Perhaps we still think of Berkman as having a 1.21 split in a 1.15 league, but that the overall number to use as the baseline is lower given the potential for his overall performance to be severely worsened while adjusting to his new digs. Or perhaps we assume the overall number remains intact but the platoon ratio is much, much larger than the league. Assumptions fuel this type of analysis, and unfortunately, the goal of such research is to measure an aspect of the game that rarely ever occurs.
Perhaps this isn’t really even a question that should arise for a 12-year veteran and should instead be addressed when the players are being developed, since their neural pathways are not yet set in stone. Switch-hitters are valuable to teams because they can provide platoon production from two players for the roster spot of just one, but the fact that they can ably swing and judge pitches from both sides of the plate should not automatically remove them from discussions of platooning with other players if they happen to favor one side. There is much left to learn with regards to switching and stopping so this sort of discussion marks merely the starting point, because the qualitative foundations from which the idea to stop switching is derived is ill-defined in its own right. When qualitative information fuels quantitative assumptions, yet both are on shaky ground, we have to be very careful with the outputs. Even so, this is definitely a question teams should be asking, whether at the big-league or developmental levels.