clock menu more-arrow no yes mobile

Filed under:

Building A Bullpen - Part II

I'll start off with a friendly two-part request. 1) Hot stove chatter for the day has it's own dedicated diary found here. 2) If you still want to discuss the Mitchell Report, let's keep it confined to this thread. Thanks in advance on both accounts.

Now last week we looked at who had constructed the best bullpen over the course of the last two years. I used DIPS, which is a stat that looks like a pitcher's ERA but takes into effect the quality of the defense behind them. We know that once a ball is put into play, a pitcher has no tangible effect over what happens to it. If they have a good defense behind them, more BIP will be converted to outs but that's a credit to the defense not the pitcher. Obviously, we want to find the best pitchers for a bullpen so isolating what is actually a contribution on their part is important. I also adjusted for park effects since places like PETCO can make mediocre relievers look awfully good.

One suggestion was that I look at the issue with a stat like Baseball Prospectus' WXRL. That statistic, which seems unpronounceable, is a measure of expected wins above replacement level added by relievers with an adjustment for the quality of the opposing lineup. In English, if each reliever were to face the same caliber of hitters, how many wins over a baseline would a reliever add. I looked into that statistic and here's why this is the last you'll hear of it. First, it's a counting stat. Reliever A who gets more opportunities than reliever B can have a higher WXRL while being a "worse" pitcher. We could turn it into a rate stat using the number of innings pitched but I'm not sure that's an accurate denominator. I'd prefer something like total batter's faced but I don't really know if that's a good denominator either since that's a proprietary stat. Second, it doesn't appear to be park adjusted. An flyball out recorded in PETCO is not necessarily the same as one recorded at Coors. The articles that give the technical explanations are in the BP 2005 and Baseball Between the Numbers -- neither of which I own. I'd be shocked if it wasn't park adjusted since that seems like such an obvious adjustment but BP's glossary doesn't list it. The last and most important issue for me is that it uses leveraging. I'm not a fan of leveraging the value of the ninth inning greater than that of the first inning. It's something that has never made intuitive sense to me and I try to avoid it when I can.

There's another statistic that I looked at called adjusted runs prevented (ARP) which takes care of park effects, doesn't use leveraging and is based on the run expectancy matrix. Unfortunately, it doesn't have any adjustment for defense so it's not something I'm really pleased with either. Some kind of ARP adjusted for defense or a de-leveraged WXRL adjusted for defense would be worthwhile looking at. That's not to say that either of those statistics are particularly bad but I'll take my park adjusted DIPS data over both of them.

Whew, got all that. For those of you with a glazed-over look, I've come to the conclusion that the data I started out with last week is what I want to use going forward. One quick note, I don't make any adjustments for league or division and in a perfect world I would. I do not, however, know the impact that the AL vs. NL issue has on relievers much less what adjustments would look on a division level. It's likely that the AL teams in places 4-6 would vault ahead of the Cubs but I'm going to stick with what I've got unless I find some reliever specific translations.

So, as promised, let's take a glance at the top three teams from my list last week: the Dodgers, Twins and Cubs. I'm going to confine my look at the bullpen to relievers who threw 30 innings. That's the core of the pen. Everyone has some circulating marginal talent in the back but that doesn't really speak to why teams succeed. It's the 30+ inning guys that are taking the brunt. The raw data I'm working with can be downloaded here.

The 2006 Dodgers had 5 relievers throw more than 30 innings: Takashi Saito, Jonathan Broxton, Joe Beimel, Danys Baez, Tim Hamulack. Saito is a Japanese import and the lone free agent in this group. Hamulack and Beimel are both lefties -- Beimel, if you remember, was a critical part of the Dodgers success that year before he sliced his hand up on some broken glass in a bar just before the post-season. Broxton is the youngest of the group at 22 and Saito is the only one over 30. Despite being somewhat elderly by baseball standards, Saito threw 78 innings that season with a FIP under 2. The 2007 Dodgers had holdovers of Saito, Brocton and Beimel and additions of Rudy Seanez and Scott Proctor. Proctor came over in a trade from the Yankees and Seanez proved to be a nice free agent signing.

I'm really partial to how the Twins have built their bullpen the last few years. They've got some shutdown relievers in that pen. In 2006 they logged quite a few relief innings and had 7 relievers pitch over 30 innings. Joe Nathan and Pat Neshek are (or should be) well known names at this point as they're both excellent relievers. Juan Rincon, Matt Guerrier, Jesse Crain and Willie Eyre rounded out the right hand side of the pen and Dennys Reyes was the lone lefty. In 2007 there wasn't a single lefty that qualified but Nathan, Neshek, Rincon and Guerrier all appeared again. Only Reyes in 2006 could be considered a free agent and he signed an extension to remain a Twin.

The Cubs have a bit of a different look from the other two clubs. They got a majority of their innings from a troika of free agents in Ryan Dempster (signed an extension), Bobby Howry and Scott Eyre. In 2006 flamethrowers David Aardsma and Roberto Novoa qualified along with Michael Wuertz and lefty Will Ohman. In 2007, the emergence of Carlos Marmol (who qualifies for the 5th best single season FIP between all three teams) saved the Cubs pen with a spectacular year.

Are there any conclusions that we can draw at this point? Well, I don't see anything that I'd be willing to etch in stone but most of the pitchers that are making this list aren't free agents but cost-controlled players who have under 6 years of service time. Of the 34 seasons, only six are players with a K/9 under 7 including 2 lefty seasons. Walk rates in this group aren't awe-inspiring with a full third walking over 4 per 9 innings -- maybe that can help allay some of my concerns about Chris Perez and his walk rates. Lastly, you'll note that only a third are from players over 30 and just 5 that are 35 or older. Those strike me as sensible qualities to look for in your pitchers but nothing particularly surprising.

Next week, I'll look at the Cardinal pens in depth from the last few years.