By Michael Humphreys
Baseball fans love to compare the players of today to the players who came before, but one must wonder how great the margin of error in these comparisons is. Is there any way of knowing who the real baseball greats are, and whose legend should stand the test of time?
Let’s take Omar Vizquel as an example. So says Wikipedia, “Vizquel is considered one of baseball’s all-time best fielding shortstops.” It’s true, Vizquel “is considered” a great fielder. Of shortstops, he
-holds the highest career fielding percentage of those with a long career.
-has participated in more double plays (and his primary double play partner just entered the Hall of Fame)
-is third in career assists
-has played more games at shortstop than anyone in major league history.
On top of all that, Vizquel has received more Gold Gloves than any other shortstop except for Ozzie “Wizard of Oz” Smith. Indeed, writers have described Omar and Ozzie as the “graceful Fred Astaire” and “acrobatic Gene Kelly,” respectively, of shortstops.
Vizquel has something of a signature play—fielding ordinary grounders (not just bunts) with his bare hand and throwing in one motion. He was the starting shortstop for the most successful American League team of the 1990s, second only to the Yankees. He hasn’t been much of a hitter, even for a shortstop, so it’s not unreasonable to infer he must have been a great fielder to hang on as long as he has.
But, after all that, how do we really Vizquel actually is one of baseball’s all-time best fielding shortstops? With metrics.
Let’s start with the question: What is the job of a fielder? To help his team prevent runs. At shortstop, this mainly involves converting ground balls into outs and getting the second out on double plays—in other words, recording assists. (It is very rare that shortstops catch fly balls or pop ups that couldn’t be fielded by at least two and as many as five other fielders. Most of the differences in putout rates for shortstops reflect how much they ‘hog’ these easy chances, not how many marginal hits they help their teams prevent. And line drive putouts at short are mostly dumb-luck plays.)
It is not the job of a shortstop (or any fielder) to look “graceful” or make trick plays. It’s not even a fielder’s job to avoid errors. In fact, a fielder who makes ten more successful plays but also ten more errors has just the same value as the fielder who makes an average number of plays and errors, because an error is no worse than a play not made.
Any fielding metric for shortstop needs to estimate how many assists a shortstop generated above or below what an average shortstop would have, playing for the same team. My system uses some arithmetic and the statistical technique of “regression analysis,” resulting in what I call Defensive Regression Analysis, or DRA.
DRA estimates the number of assists the league average shortstop would have recorded in place of the shortstop you’re rating by starting with the average number of shortstop assists per team that year and adjusting that number up or down based on statistically significant relationships between shortstop assists and other defensive statistics of the player’s team that are
1. not influenced by the shortstop himself,
2. as little influenced by the fielding quality of his teammates as possible, and
3. independent (approximately) of each ot