Shame On You, Tim
I have written before about teams underperforming and outperforming their expected winning percentages, but a couple paragrahs in a recent Tim Kurkjian article has prompted me to update the data I have on the Cleveland Indians and the Oakland A's to again demonstrate the pitfalls associated with depending upon run differential alone to tell you how you expect a team will do. The Pythagorean winning percentage formula is informative, but its main problem is that it does not take into account how much a team's winning and losing margins; it treats all runs equally. Therefore, a team that wins in a blowout may receive 1.5 Pythagorean wins when they can only ever receive 1 actual win. The same principle applies in reverse to those teams who lose in blowouts so teams that win big and lose small will underperform and teams that win small and loss big will outperform their run differential.
The Indians and the A's are two teams at opposite ends of the spectrum on this issue. Cleveland has vastly underperformed their expected winning percentage by .079 because they have an average victory margin of 4.8 runs and an average defeat margin of 3.3 runs. That is why the Indians give off the impression of being a good team when they really are not. They are also wildly inconsistent.
The A's have been outperforming their expected winning percentage so far this season by .046. This is not as drastic as the Indians nor is there as much difference between their average victory margin of 3.1 and average defeat margin of 3.6, but there is enough of a difference for the Pythagorean formula to be too pessimistic in gauging their talent. Their consistency helps them in the same way Cleveland's inconsistency hurts.
If I had better mathematical prowess, I would fix the formula to correct for average victory and defeat margins, but for right now I can only say do not trust expected winning percentages entirely.
Done and done.
The Indians and the A's are two teams at opposite ends of the spectrum on this issue. Cleveland has vastly underperformed their expected winning percentage by .079 because they have an average victory margin of 4.8 runs and an average defeat margin of 3.3 runs. That is why the Indians give off the impression of being a good team when they really are not. They are also wildly inconsistent.
The A's have been outperforming their expected winning percentage so far this season by .046. This is not as drastic as the Indians nor is there as much difference between their average victory margin of 3.1 and average defeat margin of 3.6, but there is enough of a difference for the Pythagorean formula to be too pessimistic in gauging their talent. Their consistency helps them in the same way Cleveland's inconsistency hurts.
If I had better mathematical prowess, I would fix the formula to correct for average victory and defeat margins, but for right now I can only say do not trust expected winning percentages entirely.
"When the season is over,'' Beane said, "we should do a study for the Indians and for us, and see why this happened. There is data from both sides, and they don't make any sense.''
Done and done.
posted by David at
12:08 AM
0 Comments:
Post a Comment
<< Home