Printer-friendly

Contact Author

May 8, 2009

Checking the Numbers

LOOGYing Around

by Eric Seidman

Over the last several decades, with the platoon effect becoming more widely understood and exploitable, a flood of pitchers with otherwise weak skills in the broad strokes have found employment in the major leagues. No aspect of the sport has undergone as thoroughgoing a transformation as bullpen usage, and while the idea of a structured relief corps may in fact be inefficient in terms of roster management, it has certainly paved the way for the rise to prominence of a number specialist relievers. Those referred to as LOOGYs—courtesy of John Sickels, who coined the catch-all term for Lefty One-Out GuYs—comprise the vast majority of this new group, entering into the latter stages of a game to try to face a same-handed hitter or two. Pitchers assigned to the role generally dominate lefty batters; approximately three-quarters of the pitches that a batter will see over the course of a season are thrown by right-handers, and a pitch delivered from a southpaw to a left-handed batter can be thrown from a more deceptive angle, altering the perception of the hitter.

Since these specialists meet and greet so few batters per appearance, evaluating their performances can be very tricky. As is, the stats for relief pitchers tend to be unreliable as stable summations of skill due to smaller innings totals; this only gets worse with LOOGYs, who accrue even less playing time. It's very common to see a specialist experience a large disconnect in his appearances and his innings pitched totals, logging, for example, only 40 2/3 innings in 65 games, as the pitcher records only one or two outs in most instances, and the resulting numbers can appear inflated and therefore non-indicative of actual performance. Many fans have already reached similar conclusions, and have shifted their statistical focus to slash lines, checking to see if certain pitchers hold an advantage against same- or opposite-handed hitters.

Unfortunately, such research proves to be meaningless because of its lack of context, which is key in any study or evaluation, but is often ignored as confirmation biases reign supreme and proper evidence which contradicts a previously determined point may be dismissed. The single-season home-run record of 73, set by Barry Bonds in 2001, becomes significant only when compared to all other seasonal totals. Without knowing about previous records, or the context in which the figure matters, his home-run total becomes nothing more than an ambiguous data point. The stats-related misstep of leaving context by the wayside occurs far too frequently, and can happen even more when evaluating LOOGYs. To begin rectifying this issue, we must first define specialist relievers for the purpose of our database classification.

One of the tables in my database contains a row for every game from each pitcher-season. I queried this table to count the number of times that a pitcher faced a maximum of three hitters in an appearance, and the number of total games for each season. Simple division, with the specialist count as the numerator, provided the percentage of total appearances in which a pitcher, well... specialized. Anyone with a specialist percentage south of fifty percent was removed, so for the sake of this study, a pitcher is considered to be a specialist if he saw three or fewer hitters in at least half of his appearances. The cutoff point can be debated, but it's far from arbitrary, as certain pitchers may not serve in the role all season long, while others may take on two jobs during the season, simultaneously serving as mop-up men.

Since the goal here involves evaluating LOOGYs and placing their numbers in the proper context, all right-handed pitchers were also removed from the table. Before delving deeper into the data, here is a breakdown of the specialist percentages relative to all games pitched, be they starts or relief appearances, going back to 1955:

  Span      SpecGP    Spec%
1955-59      3,811     13.5
1960-64      5,656     15.1
1965-69      6,922     16.0
1970-74      7,450     15.7
1975-79      7,066     14.2
1980-84      7,881     15.3
1985-89     10,072     17.1
1990-94     12,946     22.4
1995-99     19,860     25.0
2000-04     23,568     26.6
2005-08     21,750     28.9