Sports graphics

April 2, 2005  |  Ric Werme
55 Comment(s)

Runnings – Baseball Standings on the Run

(ET edition)

I recently created some simple graphs that display baseball
standings over an entire season and then came here to look for more
ideas. I was surprised to find sparklines of baseball win/loss
records. My graphs have no innovation beyond our common idea of displaying a
season’s worth of standings on a single graph. Until a better term
comes along, I call my graphs “runnings” since that implies position
change over time and “standings” imply just positions. My traces are
simply games above .500 vs. game number, which is pretty close to
time. A win goes up and to the right, a loss continues the trace and
goes down and to the right. They’re just sparkline tics assembled
differently.

The Orioles spent most of the season playing just under .500 ball.
They last saw that standing 27 games ago after a 4-0 run gave diehard fans
hopes of a strong finish. No one anticipated that the team would win
just 4 more games this season, finishing a dismal 67-95. Only the
Tampa Bay Devil Rays (55-106) kept the Orioles out of the basement.

 image1

The key thing that is easier to see with runnings is the relative
position of one team vs. another over time. The Red Sox get off to a
good start, but the Yankee’s consistant play lets them catch the Red
Sox while the Sox have an extended stretch of .500 ball. It would be
interesting to make an interactive variant where selecting a team’s
trace would highlight it (e.g. use a wider line) and also highlight
the corresponding segments in the opponents’ traces. A pastel color
could denote the away games (mnemonic – the colors of distant
mountains becomes more pastel). A parallel line or tic marks could
denote shutouts. ET suggested it would be interesting to show teams’
performance vs. another, e.g. giving the reference team the X axis and
showing the others as games ahead or behind the reference.
At the very least, my graphs need
a couple improvements – the legend ought to be on the teams’ traces,
not in a stray corner, and at the right end of each trace should be
the win/loss/ratio data representing the final standing. I’ve
experimented with shifting each team’s trace slightly vertically so that
tie stretches don’t overlap but have’t tried it on a full season.
The overlaps aren’t necessarily bad, but there cases where ambiguities
can appear for a while when more than two teams have similar or identical
records.

Of course, a sparkline saves vertical space, but note that my graph
can overlay multiple teams pretty well, but the sparklines need
vertical separation as they can’t be overlaid. Perhaps the graphs
belong in the Sunday newspapers and the sparklines belong in the rest
of the week an inside articles any day.

The data behind the graphs is excerpted from WWW pages like

http://www.baseball-reference.com/teams/BAL/2002_sched.shtml
which
got it from RetroSheet. Pertinent
data was extracted by a Python program which wrote a gnuplot control
file to produce the traces. I chose the colors by picking out
representative pixels from team logos at
mlb.com
.

I have graphs of the AL
2004 season
including a couple key post-season series of some
interest to Boston’s fans, but there’s little reason to include them here.

-Ric Werme

Topics: 3-Star Threads