Marginal Win Score correlation numbers
In an attempt to test the legitimacy of this Marginal Win Score metric I use to allocate professional basketball wins and losses, I have been calculating its correlation coefficient versus a number of variables.
1. How well do a player’s Marginal Win Score from last season explain his Marginal Win Score this season?
Correlation: .68
r2: 45%
Sample Size: 299 players from the 2010-11 season who also played in the 2009-10 season, and 179 players from the 2009-10 season who played in the 2008-09 season.
2. How well do Marginal Win Score wins and losses correlate with actual wins?
Correlation: .971
r2: 94%
Sample Size: All NBA teams for seasons: 2010-11, 2009-10
3. How well do Marginal Win Score wins and losses correlate with Pythagorean wins and losses ?
Correlation: .988
r2: 98%
Sample Size: All NBA teams for 2010-11 season and 2009-10 season.
4. How well does the Marginal Win Score of an NBA player who changes teams in the off-season correlate with his Marginal Win Score in the previous season with his old team?
Correlation: .61
r2: 37%
Sample Size: 70 players who changed teams between 2009-10 and 2010-11 seasons.
