But how does one go about making such difficult decisions? Since that's a large topic, tonight let's just play around with one possible factor, which is how the Yankees have matched up against Verlander.
Announcers and media folks hump these matchup stats all the time. For example, Brett Gardner's has been mentiond a lot as a game 3 possibility, with justification that in his career he's gone 5 for 11, with three walks, against Verlander.
However, as Tom Tango and his co-authors of The Book have demonstrated, the small sample sizes of such matchup stats render them of little meaning in terms of predicting performance.
To see how unreliable matchup stats are, the following table lists the matchup stats of Yankee batters against Verlander. The table shows Batting Average, On-Base-Percentage, and Slugging Percentage for each active Yankee who had at least 5 plate appearances against Verlander before 2011 and also faced him this season (stats from baseball-reference.com). The pre-2012 stats are in black, while the 2012 stats are in green:
|Yankee||PA (pre-2012)||PA (2012)||BA (pre-2012)||BA (2012)||OBP (pre-2012)||OBP (2012)||SLG (pre-2012)||SLG (2012)|
As you can see, the pre-2012 stats do not correlate with the 2012 stats. For example, prior to 2012 Verlander owned A-Rod; in 2012 so far, A-Rod has owned Verlander. Ichiro is the opposite. If these matchup stats were as meaningful as announcers often make them out to be, the 2012 and pre-2012 stats should not differ so much.
So while there may be an argument for playing Gardner, it should not be made primarily on the basis of his matchup stats against Verlander.
One good thing about these stats is that it puts numbers to the fact that Verlander is human, not some unbeatable robot. Even though Verlander dominated these hitters in the past, in 2012 they hit Verlander well enough to have beaten him twice. This to me is the most interesting point to be drawn from these numbers.