og15 wrote:esqtvd wrote:
I don't really care what Popovich says. The rest of the NBA uses the stat. It's more helpful than most, because defense figures in.
The question isn't whether the rest of the NBA uses the stat, it is HOW the rest of the NBA uses the stat. I have nothing against plus/minus, in fact, but the question to ask is whether how you use plus/minus from a single game is how teams around the NBA use it.
esqtvd wrote:I think you're taking its use too literally. I'll explain for the tenth time that if Jim is minus-3 and Joe is plus+3, of course it's probably noise. But once Jim is minus-7 and Joe is plus+7--or 14--it starts to affect a coach's decisions--or at least confirms them. We're leaking points with Jim out there--maybe let's give Joe some more burn and see what happens.
I disagree here too, and this is probably where we're going to agree to disagree. My question is this, are you sure that's how coaches are using it or is that simply what you think they are doing? Consistently what I read, see, hear NBA coaches doing is using +/- to measure LINEUP effectiveness, and over large samples to analyze effectiveness of player combinations.
The problem is that you are still isolating it to one player, but in a single game, it's generally impossible to compare one players plus/minus to another player like that. Okay, if Jim is minus 7:
Who was he on the court with?
What opponents were on the court along with them?
In how many minutes was he -7?
Was someone uncharacteristically hot on the opposing team?
Did his team miss a lot of open shots which caused us to be -7?
If we compare to Joe, and Joe is +7:
Who was Joe on with?
Was he on with a similar strength lineup to Jim or on with a stronger lineup while Jim was with a weak lineup?
Was our team hot while he was on and the other team missing open shots?
Then you have to look at what are Jim and Joe's role? Are they spot up shooters who just wait for open shots, and therefore aren't driving the offensive results? Was Jim poor on defense and getting picked on and that led to opposing teams open shots, or were both Jim and Joe just there? If Jim is on the court for 3 minutes and we go -3, then he's on the court for another 5 minutes and the team is -3 again, he's -6, but is that because of Jim? If Joe replaces Jim, and Paul George comes on and hits two three's in a row, and Zubac makes two stops at the basket, and Joe is +6, then gets subbed out two minutes later, the box score says Joe really impacted more than Jim, but did he? Or was he just along for the ride?
esqtvd wrote:I think plus/minus is more useful during a game than after--things can swing wildly in your favor once you get a combination that works. Plus/minus, like all stats, is seeing through the glass darkly to try to get there. But it's stupid to ignore any information that might help. And of course, examining the plus/minus of certain lineups and combinations over the longer term can also steer you in the right direction.
And when it comes to amateurs giving their opinions why Joe should be playing and Jim sucks, I'd like to see some facts that support that. Plus/Minus doesn't define reality, it reflects it.
"It's stupid to ignore any information that might help", this is the extreme example, I'm not and have never suggested ignoring anything, what I am against is the specific use case, single game plus/minus and saying that plus/minus, "is production". If you can find an NBA coach that would say that or die on that hill I would be very impressed.
I'll give you some quotes from teams that actually like +/- a lot:
“The individual game plus-minus stat is worthless,” Ainge said. “I wish they wouldn’t even put that on the individual game. We have adjusted plus-minuses that are much more complex, that we put a lot more stock into than just the raw plus-minus.”
https://celticswire.usatoday.com/2017/12/07/on-boston-celtics-gm-danny-ainge-calling-plus-minus-a-worthless-stat/Do you remember Evan Eschmeyer? Dallas signed him to a contract he shouldn't have gotten. Cuban later said that it was a mistake signing based on over-valuing small sample plus-minus, and that small sample was still in the 1000's, but not enough to make large conclusions like they did.
The Mavs were one if the innovators in using plus/minus in the league, but what they didn't do was use single game plus/minus, and what they didn't do was use unadjusted plus/minus, why? Because they were aware of the dangers. If you look around the league, the HOW teams use plus/minus is that they use adjusted plus/minus and try to remove as much noise as possible from the data.
To give you a basic understanding of the system, at its most basic its a plus minus system. Then we adjust it to take into account who the opponent is, is it home or away, are you playing against the other teams good lineup or bad lineup, what the score and game clock are (scoring the game winner is worth more than the 1st basket of the game. Scoring when up by 30 is worth nothing). If the team scores or gets a stop when the game is on the line, then your impact percentage goes up. We reward for getting the job done when it matters.
https://blogmaverick.com/2009/02/08/nba-all-stars-by-the-numbers/Almost all teams use plus/minus for patterns and for long term data. For example, when Lue earlier in the season said we gave it 10 games (sample size), and we saw that the three guard lineups were not working (player combinations / lineups, not individual).
When we say single game plus/minus is noisy, this is what we mean,
it tells you WHAT happened, but it does not tell you WHY it happened. So all single game plus minus for an individual player can tell me is that the team was positive or negative when this player was on, it can't tell me why. If I watch the game, I can sometimes see reasons why, though not always. The rest of the box score could show me why if a guy drops 25 points on 10/11 shooting, I can conclude that a good reason for the high plus/minus when he was on the floor was that he wasn't missing.
What you seem to be consistently trying to do is cite plus minutes and not just say WHAT, but say WHY, and that's what happens when you start saying, "plus-minus (single game) is production". The problem that happens here is that it can be used as fake analysis, it becomes a confirmation bias, and we see a bad plus/minus of a single player, and we narrate what happened in the game to conclude that it was because of that player, when that is not necessarily the case.
Plus/minus is production for a LINEUP. Individual plus/minus is not production for a single player.
