Sabermetrics and Softball?

by sas723 on May 29, 2005 4:31 AM EDT Comment 4 comments

Hey, I'm a big fan of this site, and hope you guys keep up the good work. I have a question that hopefully isn't too far out there:

As a student of sabermetrics and a sports columnist at a D3 university, I have done some recreational research into my university's athletic teams. I have recently taken a close look at my university's softball team (admittedly for reasons beyond statistics).

I plugged in the softball team's past three seasons into the Pythagenport Formula, using the ((r+ra)/g)^.287 exponent developed by David Smyth. The sample size may have been small (approximately 50 games each season), but the correlation between the Pythagenport win percentage and the team's actual win total was exactly the same two out of the three years, and deviated by only three wins the other year.

I also used Bill James' run created formula to calculate how many runs each player was worth to the team this past year. When I added each player's total, it came within 10 runs of the total runs scored by the team (the difference can probably be accounted for by luck and small ball).

I am therefore wondering if any research has been done with softball and sabermetrics? I did not see any such research on the internet, and am curious if you believe the same baseball ideas that have arisen out of sabermetrics may apply to softball. Certain difficulties stand out in my mind:
-The small sample size.
-The fact that games only last seven innings and that the rules for extra innings are very different than those of baseball.
-The fact that the bases are much closer together, and that the different ball and pitching motion may produce different outcomes.

Nonetheless, I believe much like John Hollinger at espn.com has done with basketball many of the ideas of sabermetrics could be applied to softball, and I am curious to hear what you guys think. Thanks!

-Steven

0 recs | Comment 4 comments

Story-email Email Printer Print

Comments

The problem I think....

Is that many sabermetric policies and ideas only translate to major league baseball. Certain principles don't work even in the minors because the defense, hitting and pitching are not talented enough (yet). I am sure a statistical science could be created for softball but it would have to be independent from baseball.

You should try to find softball records for other schools and try the pythagorean record to see if that works. Since the run principle is the same in softball I think things like that could work, but other stats would need to be developed specifically for softball.

I think pythagorean does work for many different sports because it is basically a ratio for outscoring or underscoring your opponents, and the principle in head to head sports is to outscore. With that said, I'm not sure how much further you could take this. Do steals occur often in softball? Is there alot of power shown? If so, stats like SecAvg and Iso might work, or atleast work enough that you could get a good idea of someones contribution.

"I don't set the rosters, I just make fun of the guy who does" - Rob Neyer

by Marc Normandin on May 29, 2005 8:37 AM EDT reply actions 0 recs

Softball

A lot of the formulas are calibrated to major league norms... like any run estimation formula. The science of run scoring in a softball game could be slightly different than in baseball, although I'm sure that some of the same principles underlie the game, like "run scoring comes from getting players on base."

I've never seen any research, but I'm sure it's doable to make some formulas. Softball doesn't have the same data-keeping as baseball, but I'm sure that if a large enough database were out there, you'd see a lot of similarities between the two.

by Dan Scotto on May 29, 2005 10:39 AM EDT reply actions 0 recs

Thanks

Thanks guys, I appreciate you taking out your time to comment, and you guys are probably right. I just thought that since the runs created and pythagorean formulas produced such interesting results, that there may be some correlation. I may tinker with some formulas for fun over summer (I have way too much free time),znc see what I come up with. Regardless, I really enjoy this site, and keep up the good work.

by sas723 on May 29, 2005 2:57 PM EDT reply actions 0 recs