Pythagenpat

Pythagenpat is a baseball statistic created by David Smyth and "U.S. Patriot" that is used in the calculation of adjusted standings. It attempts to find the optimal exponent for the Pythagorean expectation formula, which compares a team's actual win-loss record to the expected win-loss record based on the team's number of runs scored versus number of runs allowed. There are two versions of the formula, each developed independently. One version is rpg^0.29, developed by Patriot, and the other is rpg^0.287, developed by David Smyth. It has been suggested that .28 might work better as an exponent, but it is unclear whether this is actually so.

The formula's name is derived from Baseball Prospectus' Clay Davenport's Pythagenport formula, which attempts to perform the same function. Davenport has endorsed the Smyth/Patriot or pythagenpat formula as "a better fit to the data."

One advantage that the Pythagenpat formula has over the Pythagenport formula is that, when rpg equals 1, the exponent given by Pythagenpat is also 1, which is not the case with pythagenport. One must be the only correct exponent in this situation because "if a team played 162 games at 1 RPG, they would win each game they scored a run and lose each time they allowed a run. Therefore, to make W/(W + L) = R^X/(R^X + RA^X), X must be set equal to 1.

...
Wikipedia