In the Astros' wild card clinching victory yesterday, Dusty Baker pulled one reliever in favor of another to face pinch hitter Orlando Palmeiro. He wanted the left-lefty matchup. At the time, the Cubs were trailing 6-4 with two out in the bottom of the eighth inning. Of course, the pitching change meant several minutes of warmup, stalling the inevitable celebration. "C'mon Dusty," I thought, "get on with it! None of that matchup stuff matters anyway."
I may be wrong. Check out these stats on one-batter throwers, from Alan Schwarz in today's New York Times:
Relievers have been used for just one batter 1,013 times so far this season, roughly the norm for the past five years, according to Stats.
...The one-batter strategy usually works because such relievers are saved for the single batter they should be the most effective facing. This season, they have yielded a cumulative .170 batting average and a .282 slugging percentage, both remarkably low. In 2004, they were even lower, .131 and .209.
There are statistical biases that work into these stats (though the case of Joe Torre, discussed in the article, cuts the other way). The guy who gets the third out of the inning, for example, should probably be thrown out of the calculation since its not clear he was in for just one hitter. Regardless, biases notwithstanding, those are pretty impressive numbers.