October 28, 2009
Ahead in the Count
The Mashers and the Mashed
As we approach the first game of the 105th World Series tonight, baseball fans everywhere are having a familiar conversation about who the National League team will use as a designated hitter, which in a few days will segue into a conversation about how the American League team will fare without its DH and how well the American League pitchers will hit and bunt in the National League city.
This year is no different, as the Phillies will need to decide whether to add Ben Francisco and his .257/.332/.447 line to the lineup or Matt Stairs and his .194/.357/.379 line. In a couple of days, the Yankees will likely subtract Hideki Matsui and his .274/.367/.509 line from their lineup. The dilemma frustrates both Phillies and Yankees fans; Phillies fans know that Matsui was paid $13 million to be the Yankees' DH for a full season of baseball, and that their team's choice between Francisco and Stairs will not give the Phillies lineup a better hitter than Matsui. For a National League team, hiring a guy to DH for occasional interleague games and a possible World Series is simply not cost-efficient, and the result is that they will be at a disadvantage in American League stadiums. Yankee fans are frustrated that Matsui will not be in the lineup, despite the fact that Cashman constructed the team around including him in the lineup. With the exception of CC Sabathia (.261/.269/.391 career), the Yankees will be dealing with pitchers who are not great hitters and who have not been practicing bunting at batting practice for an entire season like the Phillies pitchers have. An interesting curveball here is that the Yankees will have a very potent pinch-hitter available in high-leverage plate appearances during Games Three through Five in Citizens Bank Park, and this adds a layer of complexity to the analysis.
This debate has been going on for a long time, of course. The American League introduced the DH in 1973, although American League pitchers hit for themselves in the World Series until 1976, when DHs were permitted to hit in all World Series games in even-numbered years. This practice was abandoned in favor of the current rule in 1986, and has been the same in every year (excluding 1994, naturally) since then. During that time frame, we have 22 World Series, consisting of 119 games to study.
Although that is not an incredibly large sample size, it is large enough that we can start to look at some statistics during those years and see if we can draw some meaningful conclusions. The first thing that jumps out is that DHs do terribly in the World Series in general. Although National League DHs do far worse, with an .220 EqA at .211/.278/.344 in 241 PA, American League DHs only managed a .250 EqA, hitting .239/.332/.385 in 246 PA. The National League DH numbers are tricky, because it is not uncommon for weak-gloved fielders like Raul Ibaņez to be displaced by stronger fields like Francisco. In that case, the National League DH numbers may actually be overstating the ability of the player actually added to the lineup-but by how much, considering the poor performance of the actual DHs the NL teams used? This is clearly an advantage for the American League, but by how much of a margin? The difference is only about six runs over 119 games when you add it all up, meaning that the effect is not terribly large and may or may not have changed the outcome of even one game in recent World Series history.
Of course, that ignores the effect of pitchers batting. Unsurprisingly, pitchers are bad hitters in the World Series, just like they are in the regular season. However, the AL pitchers are simply awful, as they have an EqA of .100 with a line of .098/.112/.121 in 145 PA, contrasted with the NL pitchers EqA of .159 with a line of .151/.197/.193 in 140 PA. National League pitchers are slightly better bunters in the Fall Classic as well, sacrificing 13 times to the junior circuit's 11 times. In fact, the difference in this run total is about 11 runs over those 119 games as well, meaning that the NL may be benefiting from this-if anyone is. Looking only at the era of interleague play (1997-2008), we see that the difference is even more extreme-the NL and AL pitchers have respective slash lines of .148/.212/.197 and .116/.116/.130 in 76 PA each, "good" for EqAs of .170 of .106, respectively.
The last piece of the puzzle is pinch-hitting, and the results here are quite significant. Determining who was actually pinch-hitting for the pitcher was a bit tricky, as teams frequently double-switched and manipulated their lineups to avoid needing to bat their pitchers. I followed a couple rules of thumb to determine who officially pinch-hit for the pitcher. Obviously, anybody who pinch-hit for a pitcher directly counted, but I also counted double-switched players under certain circumstances as well. The exception was when the pitcher's spot in the lineup actually came up after the double-switch an equal number of times as the new nine-hole hitter's spot came up. In that case, I counted the new pitcher's spot as the pinch-hitter, and I followed the same rules if the relief pitcher was removed for a pinch-hitter. If the nine-hole hitter's spot came up one more time than the new pitcher's spot in the lineup, I counted the first PA of the new nine-hole hitter, and the rest of the PA came from the new pitcher's spot.
Using these rules, the American League hitters simply blew the National League hitters out of the water. They hit for an EqA of .268 with a line of .265/.299/.482 in 88 PA, while the NL pinch-hitters managed only an EqA of .206 with a line of .179/.233/.308 in 88 PA. This is an approximate difference of 11 runs, despite so few PA available for pinch-hitters.
The conclusion is probably that the National League is at an ever so slight a disadvantage due to the DH rule. Of course, this gets complicated when we consider that subtracting the DH hurts the AL team in the NL stadium, but given the general poor performance of the DH, this may not be the most sizable effect. The total run difference of six runs over 119 games actually probably changes the outcome of less than one game, and probably only has a 0.5 percent chance of changing the outcome of an individual game. Given that, the odds of this rule changing the outcome of the series are about 0.2 percent... incredibly low. But with so much at stake, it certainly is a fun discussion just in case.