While working on more serious NBA matters, I was struck by an idea. During many basketball based arguments, someone always throws out, "The stats say..." to support their position. Of course, such a statement usually only refers to one stat, which leads to other participants pulling out their respective stat, which of course agrees with them. This often leads to the non-stat group claiming "victory", since all the stats "disagree" with each other.
So, I decided to average the 6 major NBA metrics together, to see where each player ranks according to "all" the stats. The metrics used in this evaluation were Win Shares (WS), Wins Produced (WP), Player Efficiency Rating (PER), EzPM, ASPM, and RAPM. The rankings for the top 30 are posted under "Consensus Rankings".
Since I had every players score tabulated, I decided to go ahead and see which metrics had the highest correlation to point differential, for the 2012-2013 NBA season. (Point differential is a teams points scored minus the points given up across a season. Its the best indicator for how "good" a team is.)
Metric Correlation (R2)
ASPM 98%
Win Shares 98%
RAPM 86.5%
EzPM 84.6 %
PER 75.6%
Wins Produced 57.8%
Needless to say, these look a little funky. Wins Produced is usually around the 90-95% range, while RAPM is supposed to be a poor explanatory metric. I am fairly certain that the results are accurate for every metric except Wins Produced, as I had to re-format the data I took from theNBAGeek, which could have thrown off the team sums.
That's pretty much it. While their is probably some funny business going on with the Wins Produced numbers, I'm certain it only extends to the correlation issue. The posted average of all 6 metrics is accurate, and the same is probably true for most of the correlations.
Data was pulled from Godismyjudge.com and theNBAGeek.com.
So, I decided to average the 6 major NBA metrics together, to see where each player ranks according to "all" the stats. The metrics used in this evaluation were Win Shares (WS), Wins Produced (WP), Player Efficiency Rating (PER), EzPM, ASPM, and RAPM. The rankings for the top 30 are posted under "Consensus Rankings".
Since I had every players score tabulated, I decided to go ahead and see which metrics had the highest correlation to point differential, for the 2012-2013 NBA season. (Point differential is a teams points scored minus the points given up across a season. Its the best indicator for how "good" a team is.)
Metric Correlation (R2)
ASPM 98%
Win Shares 98%
RAPM 86.5%
EzPM 84.6 %
PER 75.6%
Wins Produced 57.8%
Needless to say, these look a little funky. Wins Produced is usually around the 90-95% range, while RAPM is supposed to be a poor explanatory metric. I am fairly certain that the results are accurate for every metric except Wins Produced, as I had to re-format the data I took from theNBAGeek, which could have thrown off the team sums.
That's pretty much it. While their is probably some funny business going on with the Wins Produced numbers, I'm certain it only extends to the correlation issue. The posted average of all 6 metrics is accurate, and the same is probably true for most of the correlations.
Data was pulled from Godismyjudge.com and theNBAGeek.com.