rebelwithoutaclue wrote:Any stat is worthless if the sample size is tiny. One game means nothing, whether it be baseball or basketball.
Sasha. Pavlovic.
http://www.82games.com/ilardi2.htm
This column is by a couple of Ph.D's at 82games who came up with an "adjusted" +/- to try and compensate for the fact that mediocre players on good teams will have higher +/- numbers than good players on bad teams. I suggested dealing with this problem by using +/- only to compare players on the same team, but the authors came up with something a lot more sophisticated. Looking at the names of the players who came out on top of their adjusted rankings you'll see pretty much a Who's Who of NBA basketball in 2007-08. The exception is Jamario Moon, who somehow managed to get into the top ten as a rookie. The authors had a special observation about him, noting that their method of reducing statistical noise involves using five years of data, whereas with Moon they only had one year so the result isn't as valid. You'll notice that your buddy Sasha was dead last on the Cavs in adjusted +/- that year.
One thing I like about what they're doing is incorporating defense, which gets lost in a lot of other stats that are used to compare players. This gives an advantage to a player who contributes more or less equally on both ends like Kevin Garnett as opposed to a player whose contribution is heavily on the offensive side like Chris Paul or Kobe Bryant.
Edit:
If you're looking for the raw numbers for individual players and various player combinations, Reb, you can find them here. The numbers would be better for 2010-11 if they calculated them on a per/minute basis but unfortunately that column is blank. Check out the league's top 3-player combinations.
http://www.nba.com/statistics/plusminus/plusminus.jsp

