Best stat to determine a players value to its team

[Borut]

What is the best stat that determines how important a player is for a teams success? I am going to study win shares for sure.

[OnePostLegend]

There is no single "best stat" that tells you who the most important and impactful players are. Win shares -- along with any other stat that simply takes and manipulates box score data in some arbitrary way -- are fine; there's no reason to completely eschew them, but the best way for fans like us to evaluate players is to take every stat/metric into consideration, find out exactly what each stat does and doesn't tell you, learn a little about the X's and O's of the game i.e., strategy (maybe you already have a great foundation -- I don't know your background), and watch as much basketball as you can! That way, you'll be more able to properly contextualize everything you see and avoid making hasty generalizations based on one dataset.

With all of that said, you should look into RAPM (regularized adjusted plus-minus). It's a little complicated, but once you understand the logic behind it you will see its appeal.

[Borut]

I agree with you. I am actually building a program that will predict results of NBA games. One of the attributes will be the injuries to players. I need a stat that I can use to evaluate the value of the injured players to their teams. I will definitly look into RAPM. After reading about win shares I don't think it is the stat that I am looking for.