+/- is just "how much more does you team score than the opponent while you're on the floor." If team A outscores team B by 5 in a span of 2 minutes, that means they played really well in that stretch. If they maintain that same level of dominance over 48 minutes, then the point differential should scale up.
Just my opinion, but I actually think it makes more sense to put +/- per 48 minutes. Individual box score stats make sense per-36, because it gives an idea of what the player would produce playing full starter minutes. But +/- is really referring to team point differential with the particular player on the floor. We usually think of team stats per-48 minutes rather than per-36.
Basic +/- just keeps track of the changes in the score (up or down) while a player is on the court. Minutes played aren't taken into account. Dividing by minutes just leads to some arbitrary number as far as I can tell.
This is my point. I think +/- is more like fg%, it takes into account the good and the bad. Neither is bounded by per game or per minute. A player who takes more shots but isn't efficient shooting will have a low fg%. A player who takes less shots but is more efficient shooting will have a higher fg%. A player who plays when more shots are taken but isn't efficient at helping his team win will have a low +/-. A player who plays when less shots are taken but is more efficient at helping his team win will have a high +/-. You can compare either by the minute but it isn't necessary, imo, as it is a ratio that isn't limited by per game like the other box score stats. I hope this clarifies why I see it as different from stats like APG, PPG, and RPG that can be misleading because they penalize players who play less minutes. Efficiency ratings like FG%, FT%, and +/- are independent of "PG" therefore there is no need to adjust them by minute.
+/- is not a ratio. It is dependent on how many minutes the player plays. If you take the absolute value of +/- for every player in the league, it will highly correlate to minutes played. The more you play, the more positive or more negative your +/- will likely be.
So if you have two players with a net +/- of +5 for the season, but one player has played 1000 min while the other has played 100, what conclusions can be drawn from that.
Ok, thanks haha. I wasn't thinking about that. It is more of a raw stat. You are correct. I left my brain at work this week. :grin: Maybe it should be a ratio! :grin: EDIT: That last comment was a joke but wouldn't it be a more useful stat if it was a ratio? If it was actually (+)/(-) or rather (-)/(+) instead of (+)-(-).
Is there any factual or statistical evidence to back up the assertion that per-minute +/- can be extrapolated accurately from total +/- stats? because from my point of view the +/- stat is highly situational already; a good player on a bad team will get penalized because his team does badly, while a bad player on a good team may have a decent +/- rating because he gets carried by his teammates. For example, Michael Finley is 4 years past when he should have retired, yet he has a +7 per game in +/-, good for 2nd on that team, despite being (IMO) a big liability to the spurs. Does that mean the Spurs should play him starter minutes? hell no.
Using the raw total + and - from nba.com and making a ratio here's what you get. How does that compare to the per minute numbers? Code: [B]plus minus ratio[/B] Lowry 1109 1050 1.056190476 Hayes 1094 1048 1.04389313 Budinger 836 801 1.043695381 Landry 1235 1189 1.038687973 Ariza 1827 1768 1.033371041 Andersen 615 601 1.023294509 Brooks 1528 1518 1.006587615 Scola 1324 1320 1.003030303 Battier 1452 1469 0.988427502 Taylor 40 49 0.816326531 Cook 31 40 0.775 Mensah-Bonsu 24 32 0.75
Pure +/- is a skewed and often unreliable stat. This is why adjusted +/- was created. The stat only measures the success of the team while you happen to be in the lineup. This goes back to what I was saying about you could be coming in against 2nd string players or at the end of the game and ramp up your +/-. On the contrary, adjusted +/- takes into consideration the players you played with, for example. Are either of them perfect? No.
Yeah, but what is that giving you? The ratio of "points they scored to points we scored while I was playing"?
Good point. That ratio would actually be a better way to represent it, now that you mention it. It incorporates two relevant pieces of information: X +/- per minute, and Y points allowed per minute: (+)/(-) = (Y + X) / Y A team that's +5 with a score of 75 to 70 is better than a team that's +5 with an average score of 115 to 110. In the first case, you could expect a win% of 0.725, while in the second case you'd expect a win% of 0.651 based on pythagorean W-L estimate (here). By taking it as a ratio (1.07 to 1.05) you get a better picture of which team performed better. Unfortunately, individual (+) and (-) is not presented in the boxscore, so I don't think Commodore could add such a ratio to his chart (unless he wants to parse the play by plays).
Yeah, I just think there must be a better way of representing +/- without introducing a new variable. What if it was like a points%? You compare the number of points your team scores by total points scored while you were on the floor. So, if you are >50% you're helping the team win. Some of you all focus much more on stats than me. Isn't there some way to do this? Code: plus minus points points% Lowry 1109 1050 2159 0.513663733 Hayes 1094 1048 2142 0.510737628 Budinger 836 801 1637 0.510690287 Landry 1235 1189 2424 0.509488449 Ariza 1827 1768 3595 0.508205841 Andersen 615 601 1216 0.505756579 Brooks 1528 1518 3046 0.501641497 Scola 1324 1320 2644 0.50075643 Battier 1452 1469 2921 0.497090038 Taylor 40 49 89 0.449438202 Cook 31 40 71 0.436619718 Mensah-Bonsu 24 32 56 0.428571429
There is none, but does it matter? But not being able to extrapolate precisely from data doesn't mean the data isn't worthwhile taking a look at.
Yeah, what's up with that? The nba DOES track them individually, you can find the totals, but they don't put them in the box score individually. That's odd.
You can approximate it, though. If the final score is 105 to 100, and Scola played 36 minutes with a +5 +/-, then we could say the (+) is about 79, and the (-) is about 74.
You could, and for players on the same team that's fine. If you're comparing players across different teams that play at a very different pace (like, one has high scoring games, while the other keeps it low scoring), then straight point differential (even per minute) would be inflated for the higher scoring team.
You could. Assuming those (+) and (-) numbers are updated frequently, you could just take the difference after each game is played. Of course, that doesn't help you with the games that have already been played.
