June 29, 2010
Between The Numbers
The Need for Spd
Our Equivalent Baserunning Runs have been making some picturesque rounds in recent weeks. In addition to producing several works of EqBRR art (EqBRRt?) that wouldn’t look amiss on the walls of one’s bedroom (where they’d doubtless increase one’s odds of advancing extra bases with the ladies), Beyond the Box Score’s Justin Bopp observed (in so many words) that Jose Reyes and Angel Pagan have been banking BRR as if a scantily clad Anna Benson had been perpetually beckoning them from just beyond the next bag. Reyes and Pagan rank 5th and 10th in EqBRR, respectively, making the Mets and the Rays (courtesy of Carl Crawford and Ben Zobrist) the only teams to boast top-10 tandems.
Reyes and Pagan both seem like speedy fellows, so it’s not surprising that they’ve done some damage with their wheels. However, commenter “jwiscarson” noted that he found it “stunning that the Rangers CF (Borbon) [was] so much worse than the guy who looks like a lumbering beast (Cruz).” A quick check of the numbers reveals that his comment was on the mark: Nelson Cruz ranks 3rd on the Rangers with 1.9 EqBRR, while Julio Borbon has amassed -1.4 EqBRR, the second-worst total on the team (ahead of only the -2.2 EqBRR subtracted by Vladimir Guerrero, who’s been hobbling along on knees haunted by the ghost of Astroturf past).
This apparent incongruity got me wondering about the nature of the relationship between speed and baserunning ability. Do the two cohabitate, or do they merely exchange pleasantries at weddings and funerals? I decided to see how EqBRR and Bill James’ “Speed Score” stack up at the team level. The four-component version of Speed Score hosted at FanGraphs (as opposed to the five-component version employed by PECOTA, but unavailable on our website) consists of an average of stolen base percentage, frequency of stolen base attempts, percentage of triples, and runs scored percentage, placed on a 0-10 scale with a mean of 5 (gory methodological details here). One could pick several nits with the way Spd is currently calculated, but any indirect statistical attempt to capture a quality so elemental is doomed to be, at best, an approximation, like a novel translated from its original tongue. A baseball man with a stopwatch or a finely tuned sundial could do better, but since I’m fresh out of scouts at the moment (all of mine are out on assignment), I’m willing to settle.
I pulled EqBRRs and Spds for each team from 2007-2009, and inserted them into a scatter plot to see how closely they were correlated. Remember that EqBRR boasts five separate components in its own right; if you need a refresher on the difference between EqAAR and EqHAR, or suspect that EqOAR has something to do with a baserunner’s performance on college campuses, I don’t blame you. The rest of this post will wait while you consult our glossary, if you’re so inclined.
Now that you’ve had time to glory in your preternatural statistical savvy, study up, or revel in your ignorance, we can move on:
There’s clearly some sort of linear relationship here, although the data points aren’t clustered all that closely around the line of best fit. Perhaps we're working with an incomplete picture: sabermetric lore holds that EqBRR creator Dan Fox was working on a sixth component of his stat, EqSUR (Equivalent Soiled Uniform Runs), before he was snapped up by the Pirates, who were eager to keep the knowledge of which players weren’t afraid to “get dirty out there” proprietary. Thanks to Pittsburgh’s quick thinking, we’ll have to make do with what we have. Here’s what I found for the correlations (r) between Spd and EqBRR, and Spd and each of EqBRR’s components:
It might seem odd that EqBRR would be more closely correlated to Speed Score than any of EqBRR’s components, but think of it this way: since random variance affects the strength of a correlation, and is largely a function of the number of observations in the trial, it makes sense that we’d find a stronger correlation for the aggregate of all baserunning events than for any one particular type of baserunning event. So what can we say about that .63? Well, it’s a fairly strong correlation, but it could certainly be a good deal stronger. That it isn’t owes something to a combination of two factors.
For one thing, Spd probably isn’t a particularly accurate indicator; if we substituted clocked running times for Speed Scores, we’d likely see a stronger correlation. For another, raw speed isn’t the only aspect of effective baserunning, as we could have divined without doing any work. Knowledge and awareness of pitcher motions, opposing arms, and defensive positioning, as well as sound judgment, efficient routes, leads, and slides, and capable coaching (and last but not least, a hefty helping of heart, guts, and grit) all play significant roles.
In honor of Bernie Williams, the swift center fielder whose chronic lack of Jeterian “baserunning instincts” became a go-to topic for Yankees broadcasters throughout the late ’90s and early aughts (in fairness to them, he retired—or did he?—with -18.6 EqBRR), here are the five teams who’ve made the least of their speed on the basepaths, ordered by the disparity between their EqBRR and Spd ranks:
And in honor of Nelson Cruz, who got us started by stunning “jwiscarson,” here are the five teams who’ve made the most of their speed on the basepaths, by the same rubric: