Happy Thanksgiving! Regularly Scheduled Articles Will Resume Monday, December 1
March 14, 2008
Adventures in Consulting
I think it was Einstein who said, "Answers to questions should be as simple as possible, but no simpler." Almost every job I've ever had involved finding answers to operational questions. One of the the most interesting projects I ever worked on was for a major league organization that didn't just want an answer--they wanted the question, too. The engagement was pretty simple; I can't identify the client, nor can I use specific data, but I think this is still an interesting case study. I hope you'll find it worth your time to read; if not, you might as well skip the next 6-4-3, too, since it's going to be Part Two. So let's dive inů
We'll cover how we arrived at the question in another column, probably during the Thanksgiving season, when only the truly deranged are reading or writing for Baseball Prospectus. But let's address the question at hand, which is: How can a club best use its pitching staff?
It's a fairly simple question, but one that dominates a fair number of games, particularly once you get past the first three or four innings. Traditionally, teams use a rotation of starting pitchers, usually five, who go out there, throw somewhere around 100 pitches or so (hopefully), then give way to the guys in the bullpen, who work somewhere between one batter and perhaps 10 or 12, depending on the circumstances of the game, pennant race, or acute team need.
By and large, this structure of pitching staffs drives a surprisingly high number of decisions within an organization. It also drives the market for free agent pitchers, the roster structure for most clubs, and even team staffing decisions, like trainers, coaches, etc. On a more tactical level, managers work really damn hard to eke out every little matchup advantage they can get, resulting in a lot of left-handed relievers pitching something like 80 games and 45 innings during a season. Much to the screaming horror of people like me, multiple factors, mostly human frailty and hardwiring, lead clubs to use a closer in games that they're winning by an arbitrarily small margin very late in the game. (I'm narcissistic enough to believe that most managers do this just to piss me off and vent my spleen at random people with no interest in the topic.)
That's tradition, and it probably works better than I'd like to admit. But consider this--let's say you're running a mediocre club, one that just isn't going to win more than, say, 82 games, and that's only if a couple guys in the lineup go all Brady Anderson for you. Looking at reasonable expectations of your players' performance, you're just not going to finish very high, and realistically, you're going to win 70-something games, and your attendance after the All-Star break is going to be way too close to 10,000 a game for your taste. If you take a look at the last several decades, even teams in this circumstance stick with tradition.
So you're in that spot, and you ask someone to come in, provide a fresh, outside perspective, hoping to find some knob or lever you can adjust to get into the race in the short term, while you rebuild your talent through the farm system and draft. One of the things you ask them to look at is, vaguely enough, "pitcher usage." You do some really basic research, doing your level best to throw conventional wisdom and the sometimes weighty assumptions that come with it out the window. You come across a very basic question: What drives pitcher usage, specifically bullpen usage? Just then, you glance over at what happens to be a Cardinals game, and Tony La Russa is making his 17th pitching change of the night, bringing in the lucky fan in Section 206, Row 9, Seat M into the game, because there's a chance he might be left-handed. Hey, there's a big part of your answer: platoon advantages drive the bus, often along with conditional fatigue.
So you start to do some actual number-crunching, and as it turns out, there's actually a lot of advantage to be earned not just through platoon advantages, but by the specific performance histories of your pitchers, based on how recently they pitched, and for how long in that outing. You can get another one percent efficiency out of a complicated metric that involves more than just the last outing, but for a start, this seems a pretty good one. Being a weaselly consultant, you resist the urge to put together a lengthy PowerPoint presentation on the topic, and instead present the data to your liaison at the club, consisting entirely of a worksheet for each pitcher on his staff. It looks something like this:
Days Rest 0 1 2 3 4+ Pitches 0-14 94 98 104 100 15-29 83 92 112 109 30-44 80 96 97 45-59 98 102 130 60-74 70 75+ 78
It's a simple concept, simplified yet again for brevity's sake. Each cell represents a performance index for this specific pitcher, with 100 representing the three-year average. (For this exercise, the metric doesn't matter. It could be ERA, or WHIP, or VORP, or whatever.)
OK, so you've now had a little bit of an opportunity to take a look at some alternative ideas. The context in which this research exists is in a front office in the major leagues. So, to kind of prep the ground for the second piece of this column, I'd please like you to consider how this story actually played out, in the real world. If you're interested in sending in a note containing your speculation and reason for same, please do. I'll include the most compelling material in the "Aftermath" section of this column, which will run next week. I encourage you to take a few moments to review the organization of the various MLB clubs for some inspiration regarding potential pros and cons for this type of approach.
Some operational notes: 6-4-3 will be appearing weekly for a while, so please forgive me if I can't reply to all the emails that come in. All emails will be read, but unfortunately, I just can't respond to all of them. Second, and most importantly, I'll be hosting two BP Pizza Feeds in the Bay Area on March 18th (in Menlo Park) and March 27th (in Walnut Creek.) BP and I are springing for the pizza and soft drinks, and in return, we ask that you bring a minimum $25 donation (checks, please) made out to the Cystic Fibrosis foundation. It's a good cause, and is always a good time. I'm also in the process of setting up an Indianapolis event, which will be hosted by me and Will Carroll, hopefully sometime in late May. If you're in that area, and have a preference on dates and/or location, please drop me a note, as I could use the guidance.
Extra-super special note: We just had a large group (14) cancel on the Menlo Park Feed, so we need people to come on out next Tuesday, March 18th! Click here to email me at my BP address. Cystic Fibrosis is an awful affliction, and this is a chance to help out with some much needed funding while talking baseball and meeting other fans. Please, come on out.