This won't come as anything especially new or insightful for the people who have a good understanding of the volatility inherent in relief pitchers, but Keith Law has a good column out there right now (behind the Insider paywall) that takes a deeper look at the problem with trading young assets for relief pitching:
Within seasons, reliever performance can be volatile, both because of the nature of relievers and the small samples involved in that role. Consider some recent evidence: Of the 40 relievers who had at least three saves in the first half of 2010, 19 had higher ERAs in the second half, including Manny Corpas, who threw just 15 innings after the break before having Tommy John surgery. Of the 19 relievers whose second-half ERAs were higher, 13 saw spikes of over a run, and four -- Alfredo Simon, Matt Lindstrom, Jonathan Broxton, Jose Valverde -- saw spikes of at least three runs. Bobby Jenks, Fernando Rodney and Leo Nunezwere in the group with second-half ERAs a run or more higher than their first group, which at least shows that the phenomenon is not limited to garden-variety middle relievers, but can infect capital-C Closers as well.
It's too hard to predict a half-season of reliever performance no matter how good he is, even if you're staring at his basic stats while trying to do so.
I asked database guru Dan Szymborski to run some more rigorous queries to look at reliever volatility, and the results were pretty similar. In one query, we looked at relievers who changed teams midyear after throwing at least 30 innings with a sub-4 ERA for the first team. Of 166 pitchers, 108 saw their ERAs rise after relocating, with 68 posting ERAs for their new clubs a run above their ERAs for their former clubs, and 40 going up by at least two runs. Even eliminating pitchers who threw under ten innings for their new clubs doesn't alter the percentages. And there were a number of spectacular disasters on the list, like Eric Gagne in 2008 (ERA of 2.16 for Texas and 6.75 for Boston), Jon Rauch that same year (2.98 for Washington, 6.56 for Arizona), and Kyle Farnsworth in 2008 and again in 2010.
The philosophy that says not to invest heavily in relievers is about more than just avoiding overvaluing them in midyear deals, as we've seen plenty of absurd contracts handed out to free-agent relievers in recent years, including Lyon, Scott Linebrink, and Joaquin Benoit. (In fact, the idea for this article came from a very simple study I did in my last offseason with Toronto to show why we shouldn't give B.J. Ryan a four-year deal. I guess it worked, as we didn't give him four years. We gave him five.)
Dan looked at all relievers in the past ten years who threw at least 60 innings in one season with an ERA under 3 and then threw at least as many innings in the following season. In the group, 133 of the 172 relievers (77 percent) saw their ERAs rise in year two, from as little as 0.04 (Mariano Rivera, 2009 to 2010) to as much as 3.89 (Gabe White, 2000 to 2001). A small rise in ERA isn't necessarily a big deal, but 65 of these relievers (38 percent of the total) saw their ERAs rise by at least a full run from Year 1 to Year 2. And that sample doesn't include the pitchers who got hurt or lost their jobs in Year 2.
Expand the sample to pitchers who posted an ERA under 4 in the first year, still using the 60 IP threshold, and the results are better, but hardly encouraging. Of 340 pitchers who met the criteria in the 2000-09 period, 217 saw some rise in ERA, and 91 (27 percent) saw a rise of at least a full run.
That's a one-in-four chance that the reliever you just acquired will see his ERA spike, even if you know for a fact that he won't get hurt.
The takeaway here isn't just that teams should beware of overpaying for relievers in midseason trades or in free agency, but that noncontending teams should always look to trade high-performing relievers if they can do so for more predictable assets in other roles.
Now, of course, there are more variables injected into the equation depending on the relief pitcher that you're talking about ... a guy like Heath Bell, for example, should also recoup compensatory draft picks once he's played out his two-month string (and whatever post-season action happens to follow), which adds quite a bit of value to the relief asset that you're acquiring. If you dig beneath that layer and look specifically at the player, however, you're going to find that Bell isn't a very good bet to maintain his present clip of run prevention in a different environment ... and that's probably true of most of the relievers in this market.