It tastes bad and I’m still hungry

Warning: What follows might sound like pseudo-intellectual tripe. Worse still, it might be pseudo-intellectual tripe. I’ll take my chances; will you?

That awkward introduction out of the way, I’m drawn to the work of people who think about the way people think. I’m interested in learning more about how humans perceive and respond to their world. I’ve been reading people like Massimo Piattelli-Palmarini and Nassim Nicholas Taleb and finding my mind opened in strange ways. And seriously, this is on just a few cups of coffee, but thanks for asking.

I can’t claim a deep understanding of much of what these writers discuss (I’m not smart enough to grasp it all, but I’m smart enough to recognize this, if you follow); but I have been able to digest enough to realize that we probably ought to be less sure of things we think we know. There is danger in certainty.

As Taleb, who specializes in markets, notes: “Loyalty to ideas is not a good thing for traders, scientists—or anyone.” If our patterns of thinking grow stale, then so do the insights that those patterns produce. Well, not always, but it’s at least a risk, and more of one than we might like to admit.

One aspect of sabermetrics that I’ve always found appealing is the ability of its practitioners to formulate questions that are worth trying to answer. At its best, inquiry is made with an aim to achieve greater understanding of the way baseball works—and maybe too, if we’re real lucky, of the way human beings work.

Our species likes to build and implement tools. We might use a hammer to pound nails into a piece of plywood, a megaphone to address a large group of assembled people, or regression analysis to determine the optimal construction of a major-league lineup. Where we run into problems is when we treat regression analysis like a hammer or megaphone.

Although the latter two tools, if implemented properly, will deliver what we expect, statistical analysis is a little different. Here we’re dealing with probabilities and developing theories based on what we think the numbers tell us. To perhaps overextend our metaphor: There is no theory behind pounding a nail; you just hit the thing.

Here’s an example. One popular notion in sabermetric thinking (and it doesn’t matter which one I choose, I’m just using this to illustrate a potential blind spot) is that, in general, it is desirable for a batter to see more pitches in a plate appearance rather than fewer. The idea is that more pitches suggests greater selectivity at the plate, which in turn typically leads to greater offensive production at the individual and team levels. (If you read this paragraph carefully, you noticed that I’ve already hedged my bets with the liberal usage of modifiers that frustrate those who covet the proverbial bottom line: “in general… suggests… typically…”)

Indeed, studies have indicated (ack, another modifier!) that there may be (and another!) a correlation between P/PA and run production. On a macro level, that makes perfect sense; however, when we examine specific cases, the situation becomes less clear. Look at a guy like Ruben Rivera (please, you go first; I had to watch him for two whole years):

Year P/PA OBP BB/PA OPS+
1999 4.19 .295 .116 82
2000 3.94 .296 .092 80

Rivera’s “ability” to see a lot of pitches didn’t translate into the more useful ability to reach base or otherwise engage in activities conducive to winning baseball games. Neither does there appear to be any predictive value inherent in his 1999 performance (e.g., “He saw a lot of pitches this year so maybe next year he’ll be able to put that skill to better use”).

There are dozens of counterexamples to Rivera, so don’t bother. I’m perfectly aware of guys like Bobby Abreu and Adam Dunn and Todd Helton and Jim Thome and Kevin Youkilis. They aren’t the point, because if their performances are more typical of what we would expect given the demonstrated tendencies, then the fact that they are excellent run producers should come as little surprise and therefore teach us nothing beyond what we already suspected.

But what of Rivera? What about the anomalous cases that either (depending on our bent) drive us to distraction or cause us to dismiss them as mere noise?

Forget about Rivera. (Lord knows I’ve tried.) How do we explain the career dynamics of, say, Rick Ankiel, or Josh Hamilton? Both are extreme outliers that defy any simple analysis. And yet, as analysts, we cannot ignore them; we can’t wave our hands and say, “Yeah, but you just don’t expect that to happen.” Although this may be true, it’s not very instructive.

Here I should interject that I don’t have a solution in mind. I’m merely pointing out the dangers inherent in reaching conclusions based on something less than a complete picture.

I should also note that most folks aren’t interested in biases or knowledge gaps, they just want cheap answers wrapped in a neat package. I know many of these people: They go to McDonald’s for the food and buy USA Today for the articles. In their defense, they need solutions (“I’m hungry, what’s open?”), not theoretical constructs (“I’m hungry, what’s the optimal eating strategy given infinite amounts of time, resources and tolerance for starving while deciding such things?”).

Anyway, the point of all this (and I’m fairly certain I have a point, though the evidence remains inconclusive) isn’t to fixate on Rivera’s approach at the plate, the freakish career paths of Ankiel and Hamilton or even the relative merits of eating at McDonald’s. Rather it is to remind people (myself as much as anyone) that much of the value of sabermetrics lies in asking intelligent questions about meaningful problems and not settling for simple answers that smack of lazy dogmatism. In general (there I go again), if you find yourself reaching unassailable conclusions, you are probably on the wrong track.

Of course, there may be exceptions. I can’t be sure.

References & Resources
This article was inspired by the work of Massimo Piattelli-Palmarini (specifically Inevitable Illusions: How Mistakes of Reason Rule Our Minds) and Nassim Nicholas Taleb (specifically Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets), as well as discussions with my esteemed colleague John Brattain. Please don’t blame any of them for what I think or the way in which I choose to express those thoughts.


Comments are closed.