As many of you know, Bill James pioneered pythagorean expectation in baseball. His original formula was:
runs scored squared / (runs scored squared + runs allowed squared)
This ratio should equal your win percentage. Significant differences in actual win totals and projected win totals are attributed to luck.
Currently, instead of squaring runs scored and allowed, the exponent 1.83 is used in baseball. (Different exponents are used in football and basketball).
This year, the Coogs have scored 135 runs and yielded 68.
135^1.83 / (135^1.83 + 68^1.83) = 0.7782
We have played 22 games. This implies 17.1 wins. We have won 17 games.
Warren Nolan tracks this here:
http://warrennolan.com/baseball/2017/pythag
Here's a look at some prior seasons:
2016: We won 36 games. We "should have won" (were expected to win) 39 games based on our run totals.
2015: We won 43 games. We should have won 42.
2014: We won 48 games. We should have won 49.
2013: We won 36 games. We should have won 34.
Pretty freaking accurate.
Note the role of defense. Take two teams that win by an average of 3 runs per game. The team winning games by an average score of 8 to 5 is expected to win 64 / (64+25) or 71.9% of its games.
The team winning games 5 to 2 is expected to win 25 / (25 +4) or 86.2% of its games.
Stated another way, winning games 5 to 2 is just as good as winning games 10 to 4, in terms of expected wins. (Assuming a constant exponent.)
runs scored squared / (runs scored squared + runs allowed squared)
This ratio should equal your win percentage. Significant differences in actual win totals and projected win totals are attributed to luck.
Currently, instead of squaring runs scored and allowed, the exponent 1.83 is used in baseball. (Different exponents are used in football and basketball).
This year, the Coogs have scored 135 runs and yielded 68.
135^1.83 / (135^1.83 + 68^1.83) = 0.7782
We have played 22 games. This implies 17.1 wins. We have won 17 games.
Warren Nolan tracks this here:
http://warrennolan.com/baseball/2017/pythag
Here's a look at some prior seasons:
2016: We won 36 games. We "should have won" (were expected to win) 39 games based on our run totals.
2015: We won 43 games. We should have won 42.
2014: We won 48 games. We should have won 49.
2013: We won 36 games. We should have won 34.
Pretty freaking accurate.
Note the role of defense. Take two teams that win by an average of 3 runs per game. The team winning games by an average score of 8 to 5 is expected to win 64 / (64+25) or 71.9% of its games.
The team winning games 5 to 2 is expected to win 25 / (25 +4) or 86.2% of its games.
Stated another way, winning games 5 to 2 is just as good as winning games 10 to 4, in terms of expected wins. (Assuming a constant exponent.)
Last edited:
