6-4-3: Adventures in Consulting, Part Three

So, it’s time to check back in to see how the story ends, and my thanks to those of you who have stuck around for the full ride. In case you missed it, you can check out part one of the series here, and part two right here. In short, there’s kind of a Dilbertesque house of cards that I’ve built as an illustrative abstraction, with a consultant, some middle management, budgetary constraints, and uneven communication, all taking place in the front office of a club that wants to find a way to be competitive and outperform reasonable expectations. I realize this is a little bit like asking new viewer to start watching Lost for the first time about 50 episodes in (and insisting that they do so while sucking down an unholy blend on Jagermeister and Oxycontin), but bear with me for a bit. If you can, please take a couple of moments to go back and read at least one of the previous two parts to the series, and hopefully, things will become a bit more understandable and interesting.

In the last installment, you may remember that I left the following chart for all of you to peruse and consider prior to this delayed follow-up:

          0    1    2    3    4+
0-14    111  127  103
15-29    94  112   33
30-44         70   82
45-59
60-74                       209
75+                          61

To refresh your memory, each cell of the grid is a performance index. The pitcher’s average performance over the recent past is equal to 100. The left-hand column indicates the number of pitches thrown in the most recent outing. The top row indicates the number of days of rest since the last outing. So, for example, the upper left cell populated with the value ‘111’ indicates this pitcher was 11 percent more effective pitching on consecutive days when held to 14 pitches or fewer than he was overall. You can check out the previous articles for more information, but the key things to understand is that we’re working with a baseline where 100 is normal performance for a given pitcher, and that the cells represent pairings of pitch counts and days of rest.

When we left the case study, I had not been in regular communication with anyone from the club. My liaison with the organization had created cards like this for pretty much everyone in the system, from the majors ranging all the way down to Low-A. To make things slightly more Hitchcockian, the entire concept was in its infancy, with me believing we were going to work backwards from the decisions they’d be used to inform. Instead, they were rolled in, in their current state, to an amalgam of information passed up to senior management. Meanwhile, back at the batcave, I was over-patiently waiting to connect with my liaison, to get to work on this. Whoops. Consider this an attack of rookie mistakes.

Reader AB writes in with this observation:

I don’t understand this pitcher. It looks like you never want him to go more than one inning, at the very most. It must be a left-handed reliever, used in very short bursts, right?

And RS follows up with an insightful rejoinder:

Please tell me that you weren’t collectively stupid enough to not go into significant detail on the power of sample size. Did these oafs try to put this Tony Fossas clone into the rotation based on one emergency start two years ago?

To their credit, no, they did not. Instead, there was a meeting with about six people, including myself, the liaison, the pitching coach, one assistant who had actually created the cards for each pitcher, an Assistant General Manager, and a scout. At this point, I’ve been in the area for a couple of days to meet with people, nail down a bunch of details, and weigh in on a couple of topics. It’s a very informal setting, and the conversation does turn to the cards, and that’s when I had a day not unlike that bird had when it happened to cross the path of that Randy Johnson fastball a few years back.

The tone was immediately not one I would have chosen. It was clear after about ten seconds of cursory conversation that the expectations were that I had created this wonderful new tool that would allow better use of the pitching staff, and might even allow a weak pitching staff to become a strength. That wasn’t where I had wanted to go at the outset; I had instead prepared for more of a brainstorming session aimed at addressing big-picture questions like “What do you need to make these decisions that you don’t currently have?” and “Here’s some interesting findings that I think we can find an advantage in if we’re lucky-which ones do you want to examine first?” My recollection of the specific events here gets hazy, probably due to the human body’s remarkable penchant for protecting itself from scarring pain.

I was torn between two very disparate paths-on the one hand, wanting desperately to distance myself from using a strawman deliverable as something to base million-dollar and multiple-win decisions on, but on the other not wanting to throw my liaison under the bus, thereby making both of us look like incompetent yutzes. There weren’t a whole lot of good options, and my instinct to say “No habla Engles!” would be neither credible nor helpful. So instead I fell back on, “Holy crap. I apologize for not being in touch as much as I should. This isn’t meant as something to use in the fashion you’re talking about. It’s only the basis for us to start talking about some options that you really want. I didn’t make this clear to all of you, and this is all my bad. Let’s start over here, and I can show you some findings from some studies that I think can be used to steal a win here and there without spending any more cash.”

Mo< The groveling and apologizing took a fair amount of time, and at the end of it, I expected to hear righteous indignation and to receive a hearty chastising. Instead, it went something like this: "Thank God. I was wondering what I was missing, because I saw this to say that we should keep using [Tony Fossas Clone] as a lefty specialist, which isn't exactly too shocking." That was from the pitching coach, and things really moved forward from there.

One of the key things I learned (or, more accurately, re-learned) is that one of the great things about working with dedicated people is that you’re throwing at a moving target. It’s not throwing through a tire at 40 feet like some highly creepy Levitra ad; it’s like throwing to a gifted wide receiver who will contort their body eight ways from Sunday and stretch out to make sure the throw is caught. The result of the engagement was actually very productive, and ended up significantly different from what any of us thought we were getting into in the beginning. I can’t go into the types of specifics that would satisfy everyone, but we ended up developing a slightly different method for examining pitching performances for the purposes of forecasting. Without giving away too much, there’s some interesting data that points to a greater probability of a significant performance drop-off among pitchers, and it’s something that I hadn’t considered before getting in the room with a bunch of people with different approaches to how to answer these kinds of questions.

I still get emails from people all the time talking about stats vs. scouts in terms of “ideology,” for lack of a better term. As has been stated many times, there never was a beer vs. tacos question. Dayn Perry may not be an attractive man, his morals may be questionable, and he may pay slightly too much attention to his hair care, but he’s always stated that truth very well: it’s beer and tacos. My fear and panic in this would have been lessened significantly had I kept that in mind and had a little more faith. But not on issues of testable fact, of course. X-Rays do work.

Thanks for your forbearance. Next week’s column will be a conversation with two people who make their primary living gambling on baseball. If you have any specific questions you’d like me to ask, drop me a note this weekend.