Image credit: D. Ross Cameron-USA Today Sports
Translated by Jose M. Hernandez Lagunes
For the past two weeks, I’ve been writing about the Good Face effect: good-looking players tend to play worse, but paradoxically they play more often in the Major Leagues, because they tend to be over-promoted in the minor leagues to levels they can’t play. They are qualified. Presumably this is because managers and executives consistently, albeit subconsciously, view them as slightly better than they are, similar to the Halo effects that have been found virtually everywhere else. Researchers have conducted studies on facial attractiveness.
There’s a little caveat to those earlier studies that you may not have noticed. For both analyses, I focused on a very particular era of MLB history, the time period from 2000-2019. In the latter part, that’s because of the pandemic and the major shakeup of minor league baseball that occurred in 2020. But in the first part, it had a different reason. It turns out that the Good Face effect worked a little differently before the year 2000: then, good-looking players received a significant benefit for being good-looking.
To recap those previous studies, I used a deep learning model trained with human-reported survey data to rate the facial attractiveness of every player in MLB history. I found that more attractive players consistently perform worse than less attractive players. That was true for earned runs allowed, strikeouts per nine innings, walks, and just about any other metric you can think of.
Early in the research process, I discovered that very old photos tended to produce misleading scores, in part because they are so grainy and distorted that not even a human could provide an accurate reading. The other factor is that criteria for facial attractiveness change over time: I doubt that a player who looks like this, proudly sporting a handlebar mustache, would score well by current criteria, but from what I know, he was very good-looking for there from 1900. So I didn’t dare to go back too far in time.
In an effort to avoid both of these problems, I focused on the most recent time period before the present, 1984-2000. I used a five-year moving average to model the relationship between attractiveness and various performance statistics, for the sake of simplicity, focusing only on MLB pitchers. And in this era, the pattern I discovered earlier—handsome players are worse—is reversed. In this period, better-looking players have more strikeouts per appearance (or per nine innings, red line in the graph below), almost the same number of walks (blue line) and fewer hits (yellow line) in the Major Leagues. (These effects extend at least until 1979, but the regressions start to be quite imprecise and inconsistent around this date.)
This effect gradually fades throughout the 1990s before reversing around the year 2000. Each point on the line incorporates two years of data before and two years after, so that the point in the year 2000—when beauty becomes more of a curse than a blessing, associated with fewer strikeouts and more hits—based on data from 1998-2002.
What happened in the late 90s or early 2000s? We can only speculate. The turn of the millennium brought with it numerous changes, including the end of the steroid era, the rise of sabermetrics, and the redefinition of the strike zone. Any combination of these or other elements could be at play.
One possibility is increasing the classification of umpires. Around 1998, MLB introduced the Questec system, a precursor to PitchF/X that tracked the location of pitches as they passed through (or around) the strike zone. Around 2001 (different sources indicate different dates), long before PitchF/X or Statcast data was leaked to the public, Questec began to be used as a tool for grading umpires, flagging calls that were “wrong” according to with a precise and geometric definition of the strike zone. Previous academic and sabermetric research has shown that the Questec system drove greater accuracy and precision in ball and strike decisions as umpires became more accountable for their decisions. The heyday of very large, player-specific zones awarded to guys like Tom Glavine were over.
I find it hard to believe that the disappearance of an attractiveness effect at the same time is not related. I doubt that umpires consciously awarded any extra strikes to pitchers with more symmetrical facial features. But there’s a difference between calling the strike zone when you know you’ll be graded on calls and when you won’t.
One piece of evidence against Questec’s hypothesis comes from minor league data. I did some research with some experts about when Questec or some form of umpire classification came to the minor leagues, because it must have been at least a little later than in the majors, but I didn’t get a clear answer. However, the timing of the change in the pretty effect is almost identical in the minor leagues. If it was Questec’s fault and he was implemented a little later in MiLB, this is not what I would expect. On the other hand, if minor league umpires saw or heard about target pitch tracking systems being used in leagues above them, perhaps they subconsciously adjusted their own behavior a bit before Questec filtered down to their level.
Or maybe there’s something else to blame. Images pose their own analytical problems: lighting conditions can vary, image quality can vary, standards of facial attractiveness can vary. There may be other factors that explain this trend.
Regardless of the underlying cause, it’s interesting to revisit the old scout line about preferring players with “good looks” in light of this data. (It is important to note that this “good face” does not refer only to beauty, but also to some nebulous quality of masculinity or virility, which I assume is at least to some extent associated with beauty.) In the most memorable reference to “good face,” a fundamental passage of MoneyballBilly Beane questions the logic of ranking players by their looks, and orders his scouts to look at stats and on-field skills.
Maybe the scouts were right and Billy Beane was wrong from the start. If, for whatever reason, pitchers with good looks or certain facial features received preferential treatment in the strike zone in the years before 2000, as this research shows, then it would be entirely rational to take their appearance into account when to evaluate them. An ugly player, or perhaps one with a less defined jaw or less robust features, could suffer his entire career receiving a few fewer calls here and there, allowing more hits as a result.
Michael Lewis’ book covered the 2002 A’s, just as Questec debuted and the less attractive players saw their fortunes reverse, so in the end Beane was right: having a good face was not and would not be useful to the players drafted that year. But scouts were also right to prefer players with a certain look, because that’s what was often associated with strikeouts. They may not have known why or been able to explain it, but perhaps there was more logic to Good Face than Beane or his scouts knew.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now
2023-12-05 11:20:24
#Beauty #matter
