Why Is +/- A Good Idea For The NBA?

[OGMacgyver]

I don't think anyone who thinks it is understands how and why it works for the NHL. It's supposed to show you how good a player's performance was in a game by showing you his team's performance with him in the game vs without him in the game. It works great in the NHL because they only play around 20 minutes a game, which leaves 40 minutes without them.

In the NBA, If a guy plays 40+ minutes in a game, his +/- is gonna be in the neighborhood of the differential of the game's final score, whether he plays good or bad. So the Stat doesn't tell you anything about the player's performance in the game. It is a great idea for anyone who plays thirty minutes or less because there's plenty of minutes of game time WITHOUT the player in the game to make the stat more balanced.

[giberish]

There really aren't that many guys playing 40+ mpg anymore. Especially 40+ mpg along with 80+ games. While +/- is obviously going to rather useless for a Wilt 1962 season with the relatively big minute guys only going 35 mpg for 75 games there's a significant amount of off time to work from.

[blabla]

The irony of thinking it's better suited for NHL
+- is a stat that benefits greatly from high possession and high scoring numbers

All that said, single game +- is junk in any sport. You'll want 20+ game sample almost always

[ceoofkobefans]

The raw +/- isn’t what’s good it’s the data you can get from lineup +/- that makes APM (adjusted plus minus) which is the best impact metric there is in basketball

[JonFromVA]

The raw +/- isn’t what’s good it’s the data you can get from lineup +/- that makes APM (adjusted plus minus) which is the best impact metric there is in basketball

Raw +/- is a fantastic stat, but it's a contextual stat not an individual stat. APM alone just gives you a number, it's not going to tell you the key to getting the most out of player X is to put him in a lineup that provides Y.

[Colbinii]

The raw +/- isn’t what’s good it’s the data you can get from lineup +/- that makes APM (adjusted plus minus) which is the best impact metric there is in basketball

That isn't true. Raw +/- is great when there is a large sample size [5000-6000 minutes can really paint a strong picture].

There is a clear correlation between Positive Raw +/- and being a better player than those that are less.

There is no indication a player who scorer 18 PPG is better than a player who scores 12 PPG.

[chromeder]

So I'm a huge Plus/Minus guy. I think you can compare to any 'revolutionary' statistic in the sport. But like every that's come before it and that'll come after it, Plus/Minus is, at its core, a tool, which means it can be abused.

I'm very much a fan of doing player analysis from the ground-up. Arbitrarily weighing top-down PM and failing to connect it back to something that's happening on the court, especially for modern players, doesn't seem like a great idea to me. Where PM's value comes from, in my opinion, is its ability to map trends across archetypes and skills. Which kinds of players are generally providing the most value? And based on those patterns, we can infer which skills are really valuable.

PM is a benchmark at heart, something better than intuition but not so robust to the point that it's a substitute for analysis. That's why it's the perfect tool for balanced analysis. It keeps us on our feet by posing relevant questions with smart technique, but is ignorant enough to make us want to keep reading between the lines.

(Edit: By 'Plus/Minus,' I'm talking about the whole family of metrics, but particularly RAPM, WOWY)