My goal over the next few weeks is to focus on bullpens. I'm going to look at them from a variety of angles and hopefully provide you with some insight (if I find any) regarding what makes a good bullpen. Let me be upfront about two things: I don't think smart teams build bullpens with a) older players or b) free agents. Those are my preconceived notions and I'm more than willing to put them to the test because I acknowledge them and think they're right. I've been meaning to give this a go for a while but time constraints have prevented me from doing so. Reading this article by Geoff Young over at The Hardball Times spurred me into action.

I was really frustrated reading that article because the logical framework that he starts up with becomes exceedingly muddy after about three paragraphs. He begins by citing the large contracts handed out to Danys Baez and Jaime Walker by the Orioles last year. He says:

I went and looked for some defense independent statistics that are cumulative for bullpens. I found that ESPN has team DIPS information that can be sorted for relief. Check out their stats glossary for a brief explanation. I didn't want to stop there though. I also wanted to adjust for parks. Places like PETCO can make pitchers look better than they are with their spacious outfields. In fact that turned out to be exactly the case as the Padres ranked first in DIPS before park-adjustment but only 13th afterwards. So maybe they aren't such a good example after all so much as they are in a fantastic pitcher's park. I don't like ESPN's park factors so I used Baseball Reference's yearly park factors. You'll see that I do my best to bastardize statistics as often as possible. There's typically some volatility in bullpens so the core relievers aren't always there from year to year. I didn't want to rely on a single year's worth of data so I decided to find the best bullpen's across 2006 & 2007 combined. You can download the spreadsheet here.

I knew that the Dodgers had a good pen (although not that good) built around Japanese import Takashi Saito and young flame-thrower Jonathan Broxton. Does it surprise anyone that the Twins are second? Not me. The Cubs surprised me quite a bit but I'm just following where the numbers are leading me. The Blue Jays at #4 bought BJ Ryan for an absurd amount in 2006 but he missed all of 2007 with an injury. If you look at the yearly data, their DIPS adjusted for park jumps by a half run from 2006 to 2007. Indians and Red Sox at 5 & 6 -- that seems like a pretty good model. The Cardinals come in at 21 (a suspicion I had heading into this was that they would rank poorly). Personally, I like the methods that the Padres employ to find their relievers (under-appreciated soft-tossers, quirky delivery guys, etc.) but the method works, in part, because of PETCO.

The Orioles really do suck -- fiercely -- but I think it's poor analysis to lay that at the feet of Baez and Walker. There' are a variety of reasons that they can be bad and while I don't think Baez or Walker were good signings, I want to know what makes them poor signings. Two teams in rebuilding modes (Nats, Rays) follow the Orioles and then we have the Reds coming in at number 27. Wayne Krivsky seems to have a reliever fetish most recently exhibited by the Francisco Cordero signing. Cordero is a great reliever but a quick glance at my other spreadsheet (you can't have this one yet) shows that the relievers over 30 are scarce among those top 3 relief teams. That's not even to speak of relievers 32 and over -- it's not many.

Looking at the DIPS list those teams at the end seem like they belong there and the teams at the beginning seem like rational examples as well. I haven't finished crunching all my numbers but next week we'll look at the statistics behind the relievers in each of the top 3 pens. There's a significant jump after the Dodgers and although the separation between 2 and 8 is minimal that's where I'm drawing my arbitrary line in the sand. In subsequent posts, we'll look at the composition of some of these bullpens and individual reliever data.