On/off numbers and their importance for remainder of the season

[FNG]

Daily +/- numbers are regularly mocked in the GDT, and I agree they often have little meaning with small sample sizes. Case in point: Josh Okogie was a -16 last night, but his defense on Fox when our PGs were struggling to contain him was in my opinion a major factor in last night's win. But over the course of a season, and especially over a career, you can learn a lot about a particular player's value by looking at their on/off stats. Kekgeek has referred to Cleaning the Glass as a great source for in-depth analysis, and I finally got myself a subscription last month...it's been eye-opening and confirming in most cases. If you look at the best and worst on/off numbers on each team, you will generally find the results confirming. For instance, the best on/off number on the Bucks is Giannis, and the worst is Bryn Forbes. Best on the Lakers is LeBron, and the worst is Markeif Morris. Best on GS is Curry, and worst is Wiseman. Best on the Heat is Butler, worst is Achiuwa. And we can go on and on. Great two-way players (well, perhaps not Curry) lead their teams, and rookies and poor defenders are generally at the bottom.

Why is this important for the Wolves? Because DLO is back and looked very impressive on offense last night...we need our second highest paid player to be a star for us to be successful. KAT is...he leads the team in on/off. But DLO is dead last...just a point behind Jarrett Culver. A few of us (including me) have advanced the narrative that DLO hasn't played many minutes with our top 3 on/off contributors (KAT, Jaden and Ricky by a significant margin), so his numbers are punished by playing with lesser players (a DLO-cynic might argue the reverse...that our three leaders' numbers are so good because they haven't played with DLO!). Frankly, we need this narrative to be true if this team is going to be successful, because DLO is going play a lot of minutes and hopefully a lot of minutes with those three guys. Last night was encouraging because DLO came out of the box with an impressive +8...meaningless because it's just one game, but meaningful in that he played a lot of his minutes with KAT and Ricky. I'm going to update the on/offs periodically to see if the pro-DLO narrative proves to be true...that he becomes a more effective player the more he plays with KAT/Jaden/Ricky. And if he does, I predict this team will be much better as the season progresses.

Here are the on/offs for the season after last night (players with 600 or more minutes only)...total points per 100 possessions (positive is good)/offense points per 100 (positive is good)/defense per 100 (negative is good);

KAT: 10.7/10.7/-.1
Jaden: 7.2/4.3/-2.8
Ricky: 4.9/3.7/-1.2
Nowell: .9/.1/-.8
Edwards: .9/3.3/+2.5
Beasley: .8/3.3/+2.5
Vando: -1.9/-7.3/-5.3
Josh: -2.3/2.4/+4.7
Naz: -4.7/-6.5/-1.8
DLO: -8.4/-1.8/+6.6

If DLO's -8.4 moves closer to 0 as the season progresses, this team could have a very good finish.

[Camden [enjin:6601484]]

Just to be clear, that's not the "pro D-Lo" narrative, at least from anything that I've said or read. Players playing better with good teammates isn't anything new and can be applied to everyone who has ever stepped on a basketball court.

The belief or narrative from my view is that D'Angelo Russell is an above average ball-handler and shot-creator who also possesses perimeter shooting skills that rank him among the league's best, especially in terms of pulling up off the dribble. Players with those skills are very attractive in today's NBA and difficult to find in the same player. With that said, Russell is not a standalone star player that can simply put a team on his back on a nightly basis. He requires additional pieces to be put around him in order to be successful. Frankly, that describes Karl-Anthony Towns as well as every other player in today's NBA.

Additionally, plus-minus or on/off metrics are merely tools in an entire toolbox. They should be used in combination with other stats and formulas to try and determine a player's value. If we find that a stat can't be trusted for individual games, then why would we believe that it could be entirely trusted when it is those same games that are then grouped together into bigger sample sizes? My point isn't meant to entirely discredit these metrics, but rather to place less importance on them by themselves.

FNG, kudos to you for digging into the numbers and I'll be checking on this thread as the season continues. But like you did with Josh Okogie last night, I hope you place more significance on what you see actually happening on the court than what these numbers tell you in a vacuum.

[Jester1534]

Just to be clear, that's not the "pro D-Lo" narrative, at least from anything that I've said or read. Players playing better with good teammates isn't anything new and can be applied to everyone who has ever stepped on a basketball court.

The belief or narrative from my view is that D'Angelo Russell is an above average ball-handler and shot-creator who also possesses perimeter shooting skills that rank him among the league's best, especially in terms of pulling up off the dribble. Players with those skills are very attractive in today's NBA and difficult to find in the same player. With that said, Russell is not a standalone star player that can simply put a team on his back on a nightly basis. He requires additional pieces to be put around him in order to be successful. Frankly, that describes Karl-Anthony Towns as well as every other player in today's NBA.

Additionally, plus-minus or on/off metrics are merely tools in an entire toolbox. They should be used in combination with other stats and formulas to try and determine a player's value. If we find that a stat can't be trusted for individual games, then why would we believe that it could be entirely trusted when it is those same games that are then grouped together into bigger sample sizes? My point isn't meant to entirely discredit these metrics, but rather to place less importance on them by themselves.

FNG, kudos to you for digging into the numbers and I'll be checking on this thread as the season continues. But like you did with Josh Okogie last night, I hope you place more significance on what you see actually happening on the court than what these numbers tell you in a vacuum.

Sorry as stat guy for baseball I will never understand this stats importance. Its why I make fun of the stat. For instance Rubio -9 last night then russel hit two quick 3s and free throw to make him -2. How can a stat be so reliable when involves what your teamates do and can change so drastically? I mean in example above Rubio went from shit +- to almost even in 2 mins and he didnt do a damn thing in those 2 mins.

[Camden [enjin:6601484]]

Sorry as stat guy for baseball I will never understand this stats importance. Its why I make fun of the stat. For instance Rubio -9 last night then russel hit two quick 3s and free throw to make him -2. How can a stat be so reliable when involves what your teamates do and can change so drastically? I mean in example above Rubio went from shit +- to almost even in 2 mins and he didnt do a damn thing in those 2 mins.

Nothing to be sorry about, jester. You're understanding the stat quite well from what I can tell. It's notorious for being noisy or misleading in that there are a lot of factors that can affect the numerical value that the player in question has no or little control over -- good or bad. Your example involving Ricky Rubio is a fitting one.

[Camden [enjin:6601484]]

Here's a baseball analogy. Plus-minus is like if all fielders and the pitcher were given an Earned Run Average, or ERA, while they were on the field. They can't control the opposing lineup and what their batters do. They can't control what the pitcher throws. They can't control what the fielders to the left and right of them do. They can only make the plays that come to them. But their individual ERA would reflect everything over the course of the game so long as they were still in the lineup. That means that a right fielder could theoretically have nothing hit to him during his entire game and yet he could have a disastrous ERA. As baseball fans and analysts we would discard this stat almost immediately.

In contrast, Major League Baseball went a different route creating Fielding Independent Pitching that focuses more on what the pitcher actually has control over and arguably gives a much better picture of their ability. The idea is to remove external factors so that an individual's performance can accurately be assessed. Major League Baseball does a decent job of this -- much better than any other sport at this time.

[thedoper]

I think + - is best applied to comparing 5 man lineups. It doesn't make sense to distill it down as an assessment of individual contributions. Brook Lopez was exclusively a negative + - player (except his first year in Brooklyn) until he started playing with peak Giannis. Now he is consistently top 40 to #6 in the league last year. There is no way to isolate Brook Lopez on this one stat and say that he is anything other than a guy that plays well off of Giannis, and even that is debatable. This is not a ranking stat of individual contribution as I see it, otherwise the margin of error is far to great.

All of these names were in the top 20 of last year's + - leaders corrected for playing a minimum of 35 games.

Brooks Lopez
Eric Bledsoe
George Hill
Wesley Mathews
Donte Divincenso
Gordon Hayward
Mark Gasol
Dwight Powell
Duncan Robinson

These are only the names that debatably wouldn't be on a top 20 list of any ranking category in basketball last season other than being decent complimentary players. That is a 45% margin of error in just the top 20. Plus minus has it's uses as a stat, ranking players is not one of them.

https://www.nba.com/stats/players/traditional/?sort=PLUS_MINUS&dir=-1&Season=2019-20&SeasonType=Regular%20Season&CF=GP*GE*35

If you start with Plus Minus as your mathematical foundation for any equation you are building your house in quicksand.

[AbeVigodaLive]

I think + - is best applied to comparing 5 man lineups. It doesn't make sense to distill it down as an assessment of individual contributions. Brook Lopez was exclusively a negative + - player (except his first year in Brooklyn) until he started playing with peak Giannis. Now he is consistently top 40 to #6 in the league last year. There is no way to isolate Brook Lopez on this one stat and say that he is anything other than a guy that plays well off of Giannis, and even that is debatable. This is not a ranking stat of individual contribution as I see it, otherwise the margin of error is far to great.

All of these names were in the top 20 of last year's + - leaders corrected for playing a minimum of 35 games.

Brooks Lopez
Eric Bledsoe
George Hill
Wesley Mathews
Donte Divincenso
Gordon Hayward
Mark Gasol
Dwight Powell
Duncan Robinson

These are only the names that debatably wouldn't be on a top 20 list of any ranking category in basketball last season other than being decent complimentary players. That is a 45% margin of error in just the top 20. Plus minus has it's uses as a stat, ranking players is not one of them.

https://www.nba.com/stats/players/traditional/?sort=PLUS_MINUS&dir=-1&Season=2019-20&SeasonType=Regular%20Season&CF=GP*GE*35

As a side note... maybe there's juice to Gordon Hayward being a really good basketball player.

His departure from Boston has the Celtics at only .500. His arrival in Charlotte has led to a resurgence where the Hornets are above the Celtics in the playoff hunt.

Although... Hayward is injured again. And the Hornets lost the first game without him by 30... to Boston.

[thedoper]

I think + - is best applied to comparing 5 man lineups. It doesn't make sense to distill it down as an assessment of individual contributions. Brook Lopez was exclusively a negative + - player (except his first year in Brooklyn) until he started playing with peak Giannis. Now he is consistently top 40 to #6 in the league last year. There is no way to isolate Brook Lopez on this one stat and say that he is anything other than a guy that plays well off of Giannis, and even that is debatable. This is not a ranking stat of individual contribution as I see it, otherwise the margin of error is far to great.

All of these names were in the top 20 of last year's + - leaders corrected for playing a minimum of 35 games.

Brooks Lopez
Eric Bledsoe
George Hill
Wesley Mathews
Donte Divincenso
Gordon Hayward
Mark Gasol
Dwight Powell
Duncan Robinson

These are only the names that debatably wouldn't be on a top 20 list of any ranking category in basketball last season other than being decent complimentary players. That is a 45% margin of error in just the top 20. Plus minus has it's uses as a stat, ranking players is not one of them.

https://www.nba.com/stats/players/traditional/?sort=PLUS_MINUS&dir=-1&Season=2019-20&SeasonType=Regular%20Season&CF=GP*GE*35

As a side note... maybe there's juice to Gordon Hayward being a really good basketball player.

His departure from Boston has the Celtics at only .500. His arrival in Charlotte has led to a resurgence where the Hornets are above the Celtics in the playoff hunt.

Although... Hayward is injured again. And the Hornets lost the first game without him by 30... to Boston.

Hayward is a nice complimentary player, and + - probably does account for some of that. I just don't think that made him a top 20 player last year.

[AbeVigodaLive]

I think + - is best applied to comparing 5 man lineups. It doesn't make sense to distill it down as an assessment of individual contributions. Brook Lopez was exclusively a negative + - player (except his first year in Brooklyn) until he started playing with peak Giannis. Now he is consistently top 40 to #6 in the league last year. There is no way to isolate Brook Lopez on this one stat and say that he is anything other than a guy that plays well off of Giannis, and even that is debatable. This is not a ranking stat of individual contribution as I see it, otherwise the margin of error is far to great.

All of these names were in the top 20 of last year's + - leaders corrected for playing a minimum of 35 games.

Brooks Lopez
Eric Bledsoe
George Hill
Wesley Mathews
Donte Divincenso
Gordon Hayward
Mark Gasol
Dwight Powell
Duncan Robinson

These are only the names that debatably wouldn't be on a top 20 list of any ranking category in basketball last season other than being decent complimentary players. That is a 45% margin of error in just the top 20. Plus minus has it's uses as a stat, ranking players is not one of them.

https://www.nba.com/stats/players/traditional/?sort=PLUS_MINUS&dir=-1&Season=2019-20&SeasonType=Regular%20Season&CF=GP*GE*35

As a side note... maybe there's juice to Gordon Hayward being a really good basketball player.

His departure from Boston has the Celtics at only .500. His arrival in Charlotte has led to a resurgence where the Hornets are above the Celtics in the playoff hunt.

Although... Hayward is injured again. And the Hornets lost the first game without him by 30... to Boston.

Hayward is a nice complimentary player, and + - probably does account for some of that. I just don't think that made him a top 20 player last year.

Oh. I wouldn't put him in the top 20... nor do I think "nice complementary player" really does him justice.

[thedoper]

I think + - is best applied to comparing 5 man lineups. It doesn't make sense to distill it down as an assessment of individual contributions. Brook Lopez was exclusively a negative + - player (except his first year in Brooklyn) until he started playing with peak Giannis. Now he is consistently top 40 to #6 in the league last year. There is no way to isolate Brook Lopez on this one stat and say that he is anything other than a guy that plays well off of Giannis, and even that is debatable. This is not a ranking stat of individual contribution as I see it, otherwise the margin of error is far to great.

All of these names were in the top 20 of last year's + - leaders corrected for playing a minimum of 35 games.

Brooks Lopez
Eric Bledsoe
George Hill
Wesley Mathews
Donte Divincenso
Gordon Hayward
Mark Gasol
Dwight Powell
Duncan Robinson

These are only the names that debatably wouldn't be on a top 20 list of any ranking category in basketball last season other than being decent complimentary players. That is a 45% margin of error in just the top 20. Plus minus has it's uses as a stat, ranking players is not one of them.

https://www.nba.com/stats/players/traditional/?sort=PLUS_MINUS&dir=-1&Season=2019-20&SeasonType=Regular%20Season&CF=GP*GE*35

As a side note... maybe there's juice to Gordon Hayward being a really good basketball player.

His departure from Boston has the Celtics at only .500. His arrival in Charlotte has led to a resurgence where the Hornets are above the Celtics in the playoff hunt.

Although... Hayward is injured again. And the Hornets lost the first game without him by 30... to Boston.

Hayward is a nice complimentary player, and + - probably does account for some of that. I just don't think that made him a top 20 player last year.

Oh. I wouldn't put him in the top 20... nor do I think "nice complementary player" really does him justice.

Fair enough. His role ended up being that on Boston. He definitely has the chops to be more of a lead dog. His injuries are a shame. Hope it works out for him one day.