The concept of “clutch” is so central to our understanding of sports that it needs little in the way of introduction. Simply put, a clutch player is one who performs better when the game is on the line. The usual criterion for recognizing “clutchness” is something along the lines of, “If your life depended on a jump shot/putt/hit being made, whom would you want to attempt it?” For most people, the answers would be Michael Jordan and Tiger Woods—but what about baseball? If you're a manager, whom do you want at the plate in the ninth inning with two outs and you're down by a run? Or if you're up by a run and the other team is batting in the ninth, whom do you want on the mound? Is there any way to answer these questions objectively? Let's take a look.

The most obvious way is to look at a player's history—compare his wOBA in high-pressure situations to his overall wOBA, and chalk up any difference to the player's ability to come through in the clutch. That much is straightforward, but what exactly constitutes a high-pressure situation? Obviously there is no concrete definition, but as long as we divide plate appearances into two groups, such that one set is mostly low-pressure and the other is mostly high-pressure, we're fine. We'll define a high-pressure situation as one in which runs are needed in the very near future but the game is not yet out of hand: i.e., any plate appearance in the eighth inning or later in which the batting team is trailing by one, two, or three runs. Again, there really is no perfect definition of a “clutch” or high-pressure situation, but this will do just fine for our analysis. All other PA will be classified as “non-clutch.”

Looking at all players with at least 100 high-pressure plate appearances between 2000 and 2004, we find that the best clutch hitter in our sample was Scott Spiezio, with a clutch wOBA of .416 and a non-clutch wOBA of .329. In other words, Spiezio is pretty much an average offensive player overall, but when the game is on the line, he seems to turn into one of the game's top players. That's saying something, right? Of course, one might expect that Spiezio would be getting some more attention (Scott who?) if this transformation could be counted on in the future. Well, maybe Spiezio is a statistical fluke. What about some other elite clutch performers? Aramis Ramirez is second-best with a clutch/non-clutch differential of .079, followed by Bret Boone (.075), and J.T. Snow (.067). And how about Derek Jeter, who is widely regarded as one of the game's great clutch hitters? His improvement is a mere .022. Perhaps something else is going on.

One of the pervasive themes of this book is the danger of inferring too much from too little by underestimating the influence of randomness. In the case of clutch hitting, clutch plate appearances, according to our definition of “clutchness,” typically account for 7% of a player's total, which means that a regular player will see approximately 30 clutch situations over the course of a full season. Perhaps what we're actually seeing is just random variation caused by the small number of clutch plate appearances? Recalling the Toolshed chapter, we expect that the typical random fluctuations (for the mathematicians, one standard deviation) to be around .050 in wOBA after 100 plate appearances, meaning that 16% of players in our sample will have clutch wOBA more than 50 points higher than their true clutch wOBA due to randomness alone.

As you have already seen a few times in this book, the quickest way to examine this is to determine whether or not a player's history of clutch performance is useful in predicting his future clutch performance. For example, in 2005, would a manager have had any reason to expect Spiezio's .416 clutch wOBA to continue, and thus make decisions accordingly? Or put differently, if you're managing a team, how important is it to your decision-making process that a player has done well in the clutch in the past?