May 30, 2001

The Daily Prospectus

More on the Walk Gap

by Joe Sheehan

While sporadically entering data for a follow up column on the Walk Gap, the following arrived in my inbox. Suffice to say Mike Rice does more justice to the topic than I can, so I'll turn it today's column over to him:

I read your article on the "Walk Gap" last week and decided to do a little research, using the data at Baseball-Reference.com. I used the 1995-2000 seasons--the post-strike era--and found that there is a very strong, positive correlation between Walk Gap and Run Differential. On average, the correlation was 72.5% in the American League and 78.8% for the National League.

What I did was figure the difference between walks and walks allowed for each team, then ranked them from greatest to smallest. I then figured the difference between runs scored and runs allowed for each team and ranked them the same way. I then ran regressions, with the team's wins as the dependent variable and Walk Gap rank as the independent variable. I did the same exercise using run-differential rank as the independent variable also.

There is a pretty good relationship in most years. The t-scores are significant and the R-squared is pretty high. The three seasons that don't work are the American League in 2000 and 1997 and the National League in 1996. However when you take a team's average Walk Gap rank and their average run-differential rank and their winning percentage for the six-year period, you see that the teams with the highest winning percentages have the best rankings in both Walk Gap and run differential. The only outlier is Oakland. Year by year, you can see the effect of Billy Beane and what he has been preaching and the year in which it started to take effect.

Mike, thanks for taking the time to do the legwork, and for allowing me to share it with our readers.

Mike included an Excel file with his research, and it's clear that over a period of years, the Walk Gap is a pretty significant indicator of success. As he points out, the A's have had good walk differentials in the post-strike era, but only recently have had success to match.

From the spreadsheet, here are are average Walk Gap ranks, run-differential ranks and aggregate winning percentages of teams since 1995:

             Avg. Rank
Team    Walk Gap  Run +/-   Pct.

NYY       3.00     3.17     .594
CLE       3.33     2.33     .589
BOS       6.83     4.67     .545
SEA       6.33     5.00     .525
TEX       7.00     6.67     .519
BAL       7.17     6.17     .512
CHW       8.50     6.83     .507
ANA       9.83     7.83     .492
OAK       5.00     8.33     .485
TOR      10.50     9.00     .483
MIL      12.00     9.00     .478
KCR       8.67    11.17     .447
DET       9.17    11.00     .425
MIN       7.83    12.33     .424
TBD      11.67    12.00     .414

             Avg. Rank
Team    Walk Gap  Run +/-   Pct.

ATL       2.50     1.83     .619
HOU       2.50     5.00     .538
NYM       5.67     5.67     .531
LAD       9.17     5.67     .526
CIN       8.33     6.50     .524
SFG       5.00     7.50     .520
ARI       7.00     8.00     .514
SDP       6.17     7.83     .508
STL       7.00     7.17     .500
COL      11.17     8.00     .497
CHC      10.33    10.17     .460
FLA      12.33    10.50     .458
MIL      10.00    12.33     .456
MON      11.33    11.17     .453
PIT      10.83    11.17     .447
PHI       9.67    12.17     .441

One of the last things at which I want to look is whether the Walk Gap is an indicator of whether a team will outperform its Pythagorean record. It's not clear from the data Mike sent in, and it is something about which I've been wondering. I'll keep plugging in my own numbers and see if anything comes of it.

Joe Sheehan is an author of Baseball Prospectus. Contact him by clicking here.

Joe Sheehan is an author of Baseball Prospectus. 
Click here to see Joe's other articles. You can contact Joe by clicking here