http://www.jimbeam.com/ourbourbon.aspx

has write ups on the different flavors and styles with some side by side notes.

I think +/- is an extemely useful stat. Thus said, it can have high variables. It needs a larger sample size to be more useful. If you notice on the box scores that includes +/-, it tends not to mean much for one game. However, the longer the duration that it indicates, the more accurate of an indicator that it is.

Still their are rotation variables. One of the best examples is a player that you site, Kevin Durant. Durnat has a negative +/-, yet the two other good players on his teams, Westbrrok and Green, have postive +/-. In fact, Westbrook has an excellent +/-. I’ve watched all three games vs. the Warriors and Scott Brooks has unusual substitute patterns with those three. Besides the opening and the closing of the halves, those three are rarely in the game at the same time. Brooks always has at least one of them in and usually two, but the patterns seem to change.

The thing is, the other players on that team are awful, so the Thunder are worse when only two of the three are playing and terrbile when only one of the three are playing. So, I’m guessing that the differences of Durant, Westbrook and Green’s +/-, may be due to randomness. Let’s see how they look next season and there after.

There is nothing random about LeBron’s +/-, however. He has the best +/- in the league, because his team is sooo much better when he is in the game. His +/- is much, much better than Kobe’s, btw.

Am I missing something here in regards to +/- ?

Yes.

Raw +/- doesn’t do that much, but within the context of a team, it can tell you a bunch. If a player averages 3 per 48, he’s negative. His team has been outscored with him on the court. If, when he’s on the bench, the team is -6 though, it suggests that he’s a net positive to the team, that without him, they do even worse. If those numbers were reversed (-6 on the, court, -3 off) it suggests that the player plays for a bad team overall, and he’s not one of the better players on that team. The ’net +/’ is far, far more useful than the raw total.

It is difficult to compare +/- between teams however. Some attempts (‘adjusted +/-’, which takes into consideration the quality of teammates in various manners) have been made, but I don’t find them all that reliable because they swing wildly from season to season for some players and the standard errors are huge. I think they change far more than the actual quality of a player’s play does and if they aren’t good predictors of future events, then they don’t serve that much function for me in isolation.

However, in conjunction with other measures, they can be rather useful. Generally, they tend to agree rating the players known to be ‘good’ as such, but there’s some times when different measures say different things. If someone has (for example) a low “PER” (not that I’m a big fan of PER, but just for example - it does say something about players) but a consistently high +/ on good teams (ala Battier) it suggests that he may be doing something that isn’t being record, something positive. If someone has a consistently high PER but lousy +/-, it could mean that one measure is unreliable (e.g. padded with ‘garbage time’ stats or defense is so bad that it cancels out measured stats). Last year, for example, Durant was far more negative than the Sonics overall. Given that beyond the unreliable PPG, his stats were lousy, it suggests that he wasn’t really an effective player. This year, the box score stats (FG%, rebounding) have turned around, but the team is still far more in the negative when he’s on. I suspect this means that while he’s improved quite a bit and is turning into a fine player, he’s still defensively ineffective. It’s also probably swayed something by the small sample of when he’s off court, largely in garbage time when the “Thunder” are behind and the other team doesn’t care about letting them back in. This should be taken into consideration (and analyzed), but there’s still evidence that Durant has some on court issues he’ll have to solve (including their coach, who, though better than PJ, still puts in some strange lineups that don’t seem to bring out Durant’s best).

Again, most of the time this sort of stuff doesn’t surprise. Duncan good by all measures, Lebron as well. Where this sort of stuff can get useful is in looking for bargains, guys who don’t impress the yahoo fans who want to see monster dunks and equate scoring 30 points as a great performance even if it took 33 shots to do it. (This, I’ve noticed, characterizes most basketball fans AND ‘journalists’ AND a surprising number of GMs who give out big contracts to inefficient big time scorers who consistently play for losing squads yet the GMs never seem to see the connection). You can actually add quite a few wins to a team’s total without adding a ‘star’ at a low price if you look for the bargains who don’t necessarily light up the scoring column and won’t make Youtube clips. Inevitably, with only 5 guys on the court, to contend you do need some acknowledged stars who really are the top players in the league, but the supporting cast do matter. Not all “role players” are created equal.

Closest so far. I have taught courses at the college level, but never hired as a prof.