BP Comment Quick Links
August 2, 2001 Aim For The HeadReaching on ErrorsThis week's question comes from T.W., who asks:
Does anyone track errors against for a hitter? It would seem to me that this might be a consistent skill (due to speed, an ability to consistently hit hard ground balls, and continued hustle). If it is a consistent skill shouldn't this be tracked and shouldn't certain players get recognition for this? I would think that historically Pete Rose would have "caused" more errors than someone like Ken Griffey Jr. due simply to hitting style and hustle. So, my question is, can this be measured? If so, who are the great players historically and today? If so, shouldn't this be accounted for in a players OBP?
Thanks, T.W. This is one of the most common questions I've gotten over the
years between BP.com and my own Stathead.com
Web sites. Fortunately, with the availability of play-by-play data over the past
couple of decades, it's possible to investigate this properly.
I have play-by-play data for 1978-2000, so all of the analysis in this
article will cover just those years. My first step was to look at the yearly
totals, to see what the highest seasonal totals for reached-on-error (ROE)
have been. Only four players have reached base via error 20 or more times in
a single season, but some of them probably aren't who you'd expect:
Willie McGee you might have guessed, as he fits the speedy profile
you naturally suspect of having an ROE ability. But Jack Clark? Not
only is he not thought of as particularly swift afoot (though he did have 11
steals in 1979), but he's also one of the Three True Outcome demigods, with
over 37% of his career PA ending without him putting the ball in play. And
while there's more speedsters on the list, there are some surprising names
like Clark, Tim Wallach, and Vinny Castilla, too.
Of course, we're looking at raw totals, and not rates. Getting a lot of
plate appearances will help you rack up the ROE totals, regardless of
ability. So next we'll look at the highest ROE% (percentage of plate
appearances ending with the batter reaching base on error):
Limiting the set of players to those with 500 or more plate appearances, the
list of seasons with a ROE% or 2.5% or greater is:
All of the players on this list are also represented on the previous list.
Only 86 out of the 1704 player-seasons represented topped 2% ROE, so we are
talking about a fairly small impact on OBP if we were to include it,
probably comparable to or slightly less significant than hit-by-pitches.
The next step is to ask whether the tendency to reach base on error persists
from year to year (thus indicating some level of ability innate to the
player). I've plotted ROE rates in consecutive seasons for players with 500+
plate appearances in both seasons in the chart below:
The correlation between year 1 ROE% and year 2 ROE% is 0.2183, which means there's a slight relationship there, but not terribly strong (less than 5% of the year-to-year variance can be explained by differences in ability among players). Of course, we're talking about an event with pretty low frequency anyway, so it's easy for a true effect to get lost in the small sample sizes of a single season. What if we borrow a technique I used in analyzing Voros McCracken's DIPS research, and split each player's career into even- and odd-numbered halves (1000+ PA in each half) and see if the evidence for ROE ability is any stronger?
It may not be clear at first glance, but there is actually a significantly more distinct linear trend in the chart above versus the year-to-year chart presented earlier. The correlation between the ROE% in the even-year half-career and the odd-year half-career jumps to 0.4114, with more than 16% of the variance now attributable to ability. So there's decent evidence that reaching base via error is at least partially due to the ability of the batter, ranging from about 0.75% to 1.75% or so for most players. It's difficult to spot using just a single year's worth of data, but using career ROE% to date is not a bad way to set an expectation for the future. As for its effect on OBP, there's a reasonable argument to be made to include it, as it does partially reflect the skill of the batter to avoid an out by inducing an error. However, since all players would get somewhat of a boost from including it, the importance of the effect is how much we might overrate or underrate a player's OBP relative to his peers by not including ROE%. As it turns out, the effect is rather small. The standard deviation in career ROE% for our half-career subset (about 500 players) is 2.7 points of OBP, meaning that more than 90% of players would be within five points of the OBP they would have as currently measured plus the league-average boost in ROE%. So while it makes some logical sense to include ROE% in OBP, we're probably not missing a huge piece of the sabermetric puzzle by not having it in the standard calculation. However, we can't let a topic like this go by without at least sharing the career leaders. So we'll close with a couple of tables, one of the players with the highest career ROE%, and the other of the lowest career ROE%. Highest career ROE% (min: 2000 PA), 1978-2000
Player ROE PA ROE% Bob Horner 86 4213 2.041% Rudy Law 53 2647 2.002% Johnny Bench 47 2365 1.987% Wil Cordero 71 3635 1.953% Billy Sample 52 2798 1.858% Garry Maddox 63 3431 1.836% Glenn Hubbard 94 5122 1.835% John Castino 47 2578 1.823% Rey Sanchez 64 3557 1.799% Derek Jeter 64 3565 1.795% Ricky Jordan 39 2221 1.756% Rafael Ramirez 103 5887 1.750% Willie McGee 143 8188 1.746% Milt May 37 2124 1.742% Otis Nixon 101 5800 1.741% Bob Dernier 48 2757 1.741% Mark Carreon 38 2190 1.735% Johnnie LeMaster 55 3179 1.730% Al Wiggins 44 2553 1.723% Jose Vizcaino 73 4244 1.720% Luis Salazar 75 4375 1.714% Ron Jackson 42 2466 1.703%
Lowest career ROE% (min: 2000 PA), 1978-2000
Keith Woolner is an author of Baseball Prospectus. You can contact him by
clicking here.
Keith Woolner is an author of Baseball Prospectus. 0 comments have been left for this article.
|