It's been a long time; I shouldn't have left you, without a dope beat to step to...
Why are statistics quoted in "per game" format? Obviously this is a meaningless metric because it doesn't account for how many minutes per game a player actually plays, or who he plays against, or much of anything other than raw production. Some of you nerdier types use the "per 36" format, which is better but still inherently limited because it doesn't tell you what kind of system the player plays in. For example, a guy playing for Don Nelson or Mike D'Antoni is obviously going to have higher numbers (good and bad) than the same guy were he playing for Greg Popovich or Mike Brown (see Jackson, Stephen as a useful test bunny for this scenario).
So my question: why hasn't a metric been developed (or if it has been, why is it never quoted) that shows what a player is likely to do PER POSESSION? This could easily be reflected as a probabilty, and it would allow you to combine statistics (e.g: what is the probabilty of Player A taking AND making a shot each time down the floor?). It also seems that if you wanted to get super geeky on the scene you could overlay this metric with others like "minutes against starters," etc. This would give a far more complete picture than "per anything" can, because it would eliminate more variables.
This seems really simple, even to a layman like me. If better statistics were more readily available, fewer people would doubt their intrinsic value. Thoughts?