Here's an old old summary of some stats from Kevin Pelton who used to work for the Sonics, now BasketballProspectus and ESPN Insider:
http://www.nba.com/thunder/news/stats101.htmlMandark's PER explanation is good enough for conceptually understanding it. Although I should note that it's calculated on a per-minute basis not per-possession, and because of this, as long as you shoot something like 36% or higher, the more you shoot, the more your PER goes up. (And because it uses team estimates it's "less accurate" than it could be, which is easy to fix like I have by using 82games' data.)
Before he went to ESPN Hollinger used to stress it being a "quick rough summary" more than an end-all measure and that a lot of its value was in being able to see players who score higher than you expect (since 15 is always the league average) or come off the bench (Michael Redd, Andrei Kirilenko, Zach Randolph, etc.) vs. guys who play gobs of minutes (Latrell Sprewell, Steve Francis, Cuttino Mobley, etc.) and then it was upon you to investigate further about the WHY this is. ESPN has been kinda promoting it more as a definitive score for a player.
If you like the PER formula, you'll love Oliver's offensive/defensive ratings or
Rosenbaum's adjusted plus/minus.
A lot of the percentage stats like rebound percentage, steal percentage, etc. are attempts to adjust for the amount of possible attempts. For example, if there's a hundred missed shots in a game and I grab 15 rebounds that's different than if there's only eighty and I grab 15. This is one reason you'll never see anyone in the modern game approach Wilt and Russell's 20+ rebounds per game because there were like 120+ possessions in a game, whereas today there's generally 90-100. Which is why despite never cracking the 20 rebound per game barrier, Rodman has something like seven of the nine best rebounding seasons in history because he was grabbing far more of the available rebounds, even if the total number was less.
The above mentioned TS% is similar, like eFG% in that it acknowledges that all shots are not equal, some are worth two, some three, and in TS%' case there's FT's.
Most of this is an attempt to strip out the "noise" to get closer to the actual "value." If you look at just the scoring totals for example that tells you more about how many games* a player played. So you go down to per game, which really tells you more about how many minutes they played. So you go down to per minute. (The 36 or 40 or 48 is just to make it look "nicer" and be more understandable...when you say "if someone played starter minutes they'd get 20 points" vs. "they get .556 points per minute!") And then the possession stuff is to go down to another level. Kobe might get 30 points but use 40 possessions, while Kyle Korver gets 10 but uses 5. (The value of usage argument is a whole separate thing.)
ORtg and DRtg are basically just saying this is how many points a player produces per possession and how many he "gives up" per possession.
Usage Rate (Usg%) is what % of a teams possessions the player uses while he's on the court. So you have the "stars" like LeBron, Kobe, Melo, etc. using 30+% and guys like Reggie Evans using something closer to 5%.
Win Shares (WS on B-R, there are other "versions" of this same concept like WARP, Wins Produced, etc.) are just attempts to distribute wins to the players based on how much "credit" they deserve. WS/48 is just divided by minutes times 48 so to see who contributes the most per minute.
There's others but I think those are most everything on B-R and I probably shouldn't start pulling out my files.
