Curves. They’re all kinds of good. Whether a well thrown hook induces an embarrassing swing and miss, a fine, Italian sports car hugs them on a mountain road or Scarlett Johansson is wearing something tight, curves are a beautiful part of life. The curve that thrusts itself into my existence most often is this one:
That’s a graphical representation of the normal distribution that makes up the 20-80 scouting scale. I doubt the readership of this site needs an elaborate explanation of this so I’ll be brief. Scouts grade the tools of players of all ages and skills from 20 to 80 in increments of 5 or 10 (I know of at least one team that splinters it even further than that) with a score of 50 representing the major league average and every 10 away from 50 represents a standard deviation away from that mean. So, when scouts talk to one another about players, these numbers help to paint a picture of his skill set even if the inquiring scout hasn’t seen the prospect at all.
“Haven’t seen Ben Revere yet, Hank? He’s about 5’9”, 160lbs, 65 hit, 20 power, 70 runner at least. I don’t know if I’d go so far as to put an 8 on the glove but it’s a magnificent 7 while the arm is a 3 at best, and I’m being generous, Hank, he can’t throw worth a shit.”
Now I grade everything on this scale. I saw Looper on Wednesday night, comfortably put a 50 on it. The water coming out of the fountain at the theater was a 30 but the Neapolitan coconut candy I snuck in was a delicious 60. Branch Rickey is generally credited with its creation (the 20-80 scale, not the coconut candy) but we’re not totally sure where it comes from.
Scouts, who are not the xenophobic, math-hating dipshits they’ve been made out to be lately, communicate their evaluations to their bosses and each other with these numbers. Most of what scouts are grading they are grading subjectively, using skills honed over years of keen baseball observation. There are, however, a few things evaluators can quantify and grade objectively. Two that come to mind are fastball velocity and, of course, speed. Mostly, speed is measured by hand with an Accusplit stopwatch, the timer ignited when the hitter makes contact with the ball and snuffed out when he makes contact with first base. Here are the times and their corresponding grade on the scale, the way it has been since….. forever:
Grade Time (R)/Time (L) in seconds
20 4.6/4.5 and the Molinas
My stopwatch and I just sort of accepted this reality and the measurements within it and went happily about scouting minor leaguers and high schoolers all over the place until the dulcet voice of Astros Pro Scouting Director, Kevin Goldstein, blew it straight to hell. Goldstein, on several occasions in several mediums, has stated that this scale, especially at the major league level, is likely incorrect. Guys get to the big leagues mostly because they can hit, and speed is just icing on the cake of major league relevancy. As such, “average” or better runners are rarer than the curve states they should be, according to Goldstein’s hypothesis. At Baseball Info Solutions this past season I watched baseball for about 7 hours a day and timed every full effort sprint to first I could to see if Kevin is right and, if he is, what the curve actually looks like. This is what I found:
There are a few differences. First, the results aren’t distributed in a beautifully even curve, they’re skewed. Second, the .1 second buffer built into the scale to separate right handed hitters from left handed ones (since the lefty batter’s box is closer to first base) is a little light. Third, left handed hitters are, on average, a little faster than the scale would indicate while righties are a little slower.
Some important logistics stuff about how I gathered data:
We use DVRs at BIS so I was constantly rewinding, timing everything a few times to make sure I got accurate results. I got about 280 times (not all from different players, I have multiple times for some guys) which probably isn’t enough to be statistically significant, but it’s a nice start. Some players for which I recorded multiple results displayed inconsistent times. David Freese, for example, has a few times in the 4.5s but one 4.23 dash that had me constantly questioning my own existence. Some players were remarkably consistent. I’ve got several times for Angel Pagan, all of them between 3.98 and 4.02. Variances like Freese’s can occur from all sorts of stuff. Maybe the guy slipped off camera or took a poor path to the bag to slow his time. Some players’ times are not accurate representations of their speed at all. Munenori Kawasaki has a jailbreak element to his swing that has him starting toward first much sooner than other players who take forever to get going. This alters his times in context. Rickie Weeks’ weight transfer is so odd that he also gets out of the box very quickly. There are plenty of caveats involved with this data but also tons of possibilities. Are there correlations with speed and defensive metrics? Would I see trends if I sorted players by position or by the team that drafted them? I rounded everyone’s time to the nearest five-hundredths, just so you know. Ben Jedlovec, who busts his ass along with the rest of BIS’s full time staff, took time out of his day to help me with Excel so I can churn out histograms now. I once saw Ben arm wrestle Bill James. Go buy a Fielding Bible.
Even if my curve is right and the model being used by scouting departments across the globe is wrong should we care and adjust the scale? Hell no. The value in clear communication far outweighs whatever value increased accuracy provides the scouting community. Of course, I’m open to all sorts of debate about that.
If anyone wants to google doc with everyone’s individually recorded time or has a request for an individual’s time, drop my a line in the comments section and I’ll see if I’ve got it.