players whose impact metrics dont lineup with their reputations

[falcolombardi]

so i recently found this , a thread about the best 5-year stretches of

https://www.thespax.com/nba/quantifying-the-nbas-greatest-five-year-peaks-since-1997/#easy-footnote-1-4320

very useful although of course other sources may find slightly different results, what was more interesting was not the lebrons, curry, duncans, who were expected at the top, but the surprising appereances and -absences- from the best results

LeBron James 2012-16 6.46
LeBron James 2013-17 6.27
Kevin Garnett 2003-07 6.17
LeBron James 2006-10 6.15
Kevin Garnett 2000-04 6.01
Tim Duncan 2001-05 6.00
Tim Duncan 1999-03 6.00
Kevin Garnett 2002-06 5.98
Stephen Curry 2014-18 5.81
Kevin Garnett 2001-05 5.76

everythingh normal so far, later results include plus minus darlings as chris paul, wade, stockton or dirk as well as jordan last two bull seasons* (how good would it have been in his prime?)

Dwyane Wade 2006-10 5.73
Chris Paul 2012-16 5.64
Dirk Nowitzki 2008-12 5.18
Michael Jordan 1997-01 2 5.26
THEN things get interesting



[b]Joel Embiid 2015-19 5.15[/b]
Manu Ginobili 2002-06 5.15 (multiple stretches close to this one so is far from a outlier)
Shaquille O'Neal 2000-04 5.10
John Stockton 1998-02 50801 5.03

manu and embiid comparable to shaq is really interesting and old man stockton is just eyebrow raising


Draymond Green 2015-19 4.83
Nikola Jokic 2014-18 4.77
Kyle Lowry 2013-17 4.74
Westbrook 2013-17 4.55
Rasheed 2000-04 4.47
James Harden 2014-18 4.35
Kobe Bryant 2006-10 4.26

jokic, kobe and harden are expected but some players who were seen as too flawed stars (westbrook) or not real stars by many (lowry, green, rasheed) appear ahead of much bigger names like

Kawhi Leonard 2017-21 4.04
Kevin Durant 2012-16 3.87

what do you make of results like this that go so differently from a player actual perceived level?

was Ginóbili a secret superstar on the level of mvp contenders or are his results to be taken with a grain of salt as a sixth man?

was there somethingh about kawhi or durant making them less impactful than their tslent and best in the league level reputation would suggest or is it just noise?

are players like old man stockton truly that impactful? is young embiid really that good or was his team just too dependant on him?

[sp6r=underrated]

I'll just say the Spurs success during the mid-aughts supports the idea they had some secret superstar on their roster. In the span of 6 years

3 Championships
4 CF appearances
~59 Wins Average
6.86 SRS

Duncan from 04-06 had some health issues with his feet that impacted his play for stretches of the season and yet their success quite similar to showtime or Bird's Celtics (save 86) and basically all the non-GOAT level dynasties.

Manu looks like a god from a per minute perspective on either the plus/minus or box score stats.

[Texas Chuck]

Reminder that plus/minus stats are not a stand-in for actual impact. Just way too many factors coming into play to make that claim. Now I know the OP understands this already, but a reminder to those who read the thread.

But yeah I think this is valuable data that exposes some of our reliance on narrative(too old, doesn't play "right" way, team doesn't win enough) and highlights that stuffing box scores doesn't necessary signify great lift. Doc has been raising this alarm on Luka for instance.

but mostly I will say anyone who didn't already realize Stockton was an-all time great, probably needs to check some of their biases against small white dudes. And Westbrook getting turned into a punching bag in his prime was always unfair.

And I think most people here realize how good Sheed/Lowry/Draymond are in the non-volume scoring game, but the casual fans would definitely be startled by these numbers.

[jalengreen]

lol always fun / simultaneously bizarre to come across my site being linked places

i should note for anyone curious that while i used a prior for this RAPM, it was not based on box score stats and was instead an attempt at adjusting for age (so that we're not penalized a player for not succeeding with rookie lebron simply because of how good peak lebron was). so the box score isn't factored in at all other than minutes per game.

e: oh yeah, playoffs are included as well and are given 2x weight

[An Unbiased Fan]

RAPM is the problem. I mean if you value that stat go ahead, but there has never been correlation between RAPM and impact shown.

I look forward to when AI is used on old game footage and we get actually new metrics. Advanced stats use the same box scores we've had for 70 years.

[falcolombardi]

RAPM is the problem. I mean if you value that stat go ahead, but there has never been correlation between RAPM and impact shown.

I look forward to when AI is used on old game footage and we get actually new metrics. Advanced stats use the same box scores we've had for 70 years.

the uset jalengreen, who to my surprise mentioned making this specific version, explained he used plus minus rather thsn boxscore

plus-minus is not perfect but is undeniable correlated with impact

[An Unbiased Fan]

RAPM is the problem. I mean if you value that stat go ahead, but there has never been correlation between RAPM and impact shown.

I look forward to when AI is used on old game footage and we get actually new metrics. Advanced stats use the same box scores we've had for 70 years.

the uset jalengreen, who to my surprise mentioned making this specific version, explained he used plus minus rather thsn boxscore

plus-minus is not perfect but is undeniable correlated with impact

Show me the correlation. This is something that's been asked for 10+ years here. Draymond was a RAPM god, and we saw his "impact" once the splash brothers went down.

Those numbers are more about roster utility than anything else, not impact.

[LukaTheGOAT]

lol always fun / simultaneously bizarre to come across my site being linked places

i should note for anyone curious that while i used a prior for this RAPM, it was not based on box score stats and was instead an attempt at adjusting for age (so that we're not penalized a player for not succeeding with rookie lebron simply because of how good peak lebron was). so the box score isn't factored in at all other than minutes per game.

e: oh yeah, playoffs are included as well and are given 2x weight

Wow, I just wanted to say I love your work and it is incredible having a celebrity on the boards with us!

[jalengreen]

lol always fun / simultaneously bizarre to come across my site being linked places

i should note for anyone curious that while i used a prior for this RAPM, it was not based on box score stats and was instead an attempt at adjusting for age (so that we're not penalized a player for not succeeding with rookie lebron simply because of how good peak lebron was). so the box score isn't factored in at all other than minutes per game.

e: oh yeah, playoffs are included as well and are given 2x weight

Wow, I just wanted to say I love your work and it is incredible having a celebrity on the boards with us!

haha i don't know about that, but i appreciate it

[jalengreen]

RAPM is the problem. I mean if you value that stat go ahead, but there has never been correlation between RAPM and impact shown.

I look forward to when AI is used on old game footage and we get actually new metrics. Advanced stats use the same box scores we've had for 70 years.

the uset jalengreen, who to my surprise mentioned making this specific version, explained he used plus minus rather thsn boxscore

plus-minus is not perfect but is undeniable correlated with impact

Show me the correlation. This is something that's been asked for 10+ years here. Draymond was a RAPM god, and we saw his "impact" once the splash brothers went down.

Those numbers are more about roster utility than anything else, not impact.

there are certainly roster constructions and roles that can maximize the same player's impact. it is certainly possible for a player to be misused and have a RAPM that underrates the impact they could potentially offer a team. that's not the point of these metrics, though - it's to isolate the impact they *actually* had in their situation.

there are players (like batum, love, etc) who had large drops/jumps in their impact metrics upon joining a new team. was it because they got that much better/worse ? or did they join a role that was more/less fit to maximize the potential impact they could provide?

[falcolombardi]

RAPM is the problem. I mean if you value that stat go ahead, but there has never been correlation between RAPM and impact shown.

I look forward to when AI is used on old game footage and we get actually new metrics. Advanced stats use the same box scores we've had for 70 years.

the uset jalengreen, who to my surprise mentioned making this specific version, explained he used plus minus rather thsn boxscore

plus-minus is not perfect but is undeniable correlated with impact

Show me the correlation. This is something that's been asked for 10+ years here. Draymond was a RAPM god, and we saw his "impact" once the splash brothers went down.

Those numbers are more about roster utility than anything else, not impact.

roster utility and impact sounds like the same thingh with different words

[homecourtloss]

lol always fun / simultaneously bizarre to come across my site being linked places

i should note for anyone curious that while i used a prior for this RAPM, it was not based on box score stats and was instead an attempt at adjusting for age (so that we're not penalized a player for not succeeding with rookie lebron simply because of how good peak lebron was). so the box score isn't factored in at all other than minutes per game.

e: oh yeah, playoffs are included as well and are given 2x weight

Wow, I just wanted to say I love your work and it is incredible having a celebrity on the boards with us!

haha i don't know about that, but i appreciate it

No wonder you always make interesting posts. Great work!

[An Unbiased Fan]

the uset jalengreen, who to my surprise mentioned making this specific version, explained he used plus minus rather thsn boxscore

plus-minus is not perfect but is undeniable correlated with impact

Show me the correlation. This is something that's been asked for 10+ years here. Draymond was a RAPM god, and we saw his "impact" once the splash brothers went down.

Those numbers are more about roster utility than anything else, not impact.

roster utility and impact sounds like the same thingh with different words

They're quite different actually.

Roster utility is related to how the player's team in relation to player lineups, and depends on other players

Player impact is a player's actual value in relation to his impact on the floor.

A center who is giving a team 20 ppg and playing great defense, but has offcourt rotations where his team has subpar players at center, defending the paint, etc...will look better than a center who does the exact same, but has better offcourt rotations for his team. That's Roster Utility. A player like KG on Minny will post absurd +/- stats because those teams had bad personnel to substitute his roles on the team while offcourt. Does this mean KG was the best player impactwise? No, because his oncourt impact wasn't greater than many of his peers.

Ask yourself, why is Giannis's 2022 DRPM barely above Trae Young and Luka's? Hell, Curry's DRPM is 4.31 to Giannis's 1.67

The answer lies in roster utility not impact because Giannis is a FAR more impactful defender.

[Morb]

Meh.
Manu Ginobili 2001-05 34637 5.01
Ray Allen 2001-05 59629 4.15
Paul Pierce 2001-05 70297 4.00
Baron Davis 2001-05 55937 3.90
Steve Francis 2001-05 56474 3.60
Tracy McGrady 2001-05 62800 3.31
Kobe Bryant 2001-05 76224 3.16
Vince Carter 2001-05 51268 3.03
Jason Kidd 2001-05 72180 3.02
Richard Hamilton 2001-05 65346 2.51
Metta World Peace 2001-05 41667 2.45
Steve Nash 2001-05 67628 2.05
Gilbert Arenas 2001-05 40140 1.86
Michael Jordan 2001-05 18114 1.82
Mike Bibby 2001-05 65841 1.63
Allen Iverson 2001-05 68730 1.58
Peja Stojakovic 2001-05 68135 1.47
Stephon Marbury 2001-05 62231 1.09
Tayshaun Prince 2001-05 37726 1.02
Chauncey Billups 2001-05 61649 0.46
Rick Fox 2001-05 39525 0.45
Bruce Bowen 2001-05 60032 -0.01

Also, this author "forgot" about Holy T-Mac '01-'05 in this topic.
https://www.thespax.com/nba/the-nbas-most-prolific-playoff-scorers-since-1973/

Sent from my Nokia 5.1 Plus using RealGM mobile app

[An Unbiased Fan]

the uset jalengreen, who to my surprise mentioned making this specific version, explained he used plus minus rather thsn boxscore

plus-minus is not perfect but is undeniable correlated with impact

Show me the correlation. This is something that's been asked for 10+ years here. Draymond was a RAPM god, and we saw his "impact" once the splash brothers went down.

Those numbers are more about roster utility than anything else, not impact.

there are certainly roster constructions and roles that can maximize the same player's impact. it is certainly possible for a player to be misused and have a RAPM that underrates the impact they could potentially offer a team. that's not the point of these metrics, though - it's to isolate the impact they *actually* had in their situation.

there are players (like batum, love, etc) who had large drops/jumps in their impact metrics upon joining a new team. was it because they got that much better/worse ? or did they join a role that was more/less fit to maximize the potential impact they could provide?

But these numbers reflect the team rotations, not the player himself. Which is why the numbers fluctuate from season to season. And the style of coaching will weigh into the numbers too. If a coach has their star with the 2nd team quite a bit as not to give up leads, then that player's +/- numbers will be lesser than a player who's coach takes out all the starters at the same time. Does the coach utilize offensive or defensive heavy rotations? That too will muck up the waters. So many factors.

A good player will be positive, but the degree of which they are depends on the rotations used, not their direct impact.

[falcolombardi]

Show me the correlation. This is something that's been asked for 10+ years here. Draymond was a RAPM god, and we saw his "impact" once the splash brothers went down.

Those numbers are more about roster utility than anything else, not impact.

there are certainly roster constructions and roles that can maximize the same player's impact. it is certainly possible for a player to be misused and have a RAPM that underrates the impact they could potentially offer a team. that's not the point of these metrics, though - it's to isolate the impact they *actually* had in their situation.

there are players (like batum, love, etc) who had large drops/jumps in their impact metrics upon joining a new team. was it because they got that much better/worse ? or did they join a role that was more/less fit to maximize the potential impact they could provide?

But these numbers reflect the team rotations, not the player himself. Which is why the numbers fluctuate from season to season. And the style of coaching will weigh into the numbers too. If a coach has their star with the 2nd team quite a bit as not to give up leads, then that player's +/- numbers will be lesser than a player who's coach takes out all the starters at the same time. Does the coach utilize offensive or defensive heavy rotations? That too will muck up the waters. So many factors.

A good player will be positive, but the degree of which they are depends on the rotations used, not their direct impact.

sure but

a) these are multi year averages so the outliers even out
with the bigger sample
b) regular stats are lineup driven to a very significant degree too, a player boxcore or any other stats will depend on teammates and role too

even eye test will ve heavily affected by lineups, a player wont look as good without spacing as with it for example

basketball is a team sport, players dont play in a vacuum so to some degree everything they do (+/-, boxscore, other stats, eye test, accolades) will be affected by who is in the court with them

[An Unbiased Fan]

there are certainly roster constructions and roles that can maximize the same player's impact. it is certainly possible for a player to be misused and have a RAPM that underrates the impact they could potentially offer a team. that's not the point of these metrics, though - it's to isolate the impact they *actually* had in their situation.

there are players (like batum, love, etc) who had large drops/jumps in their impact metrics upon joining a new team. was it because they got that much better/worse ? or did they join a role that was more/less fit to maximize the potential impact they could provide?

But these numbers reflect the team rotations, not the player himself. Which is why the numbers fluctuate from season to season. And the style of coaching will weigh into the numbers too. If a coach has their star with the 2nd team quite a bit as not to give up leads, then that player's +/- numbers will be lesser than a player who's coach takes out all the starters at the same time. Does the coach utilize offensive or defensive heavy rotations? That too will muck up the waters. So many factors.

A good player will be positive, but the degree of which they are depends on the rotations used, not their direct impact.

sure but

a) these are multi year averages so the outliers even out
with the bigger sample
b) regular stats are lineup driven to a very significant degree too, a player boxcore or any other stats will depend on teammates and role too

even eye test will ve heavily affected by lineups, a player wont look as good without spacing as with it for example

basketball is a team sport, players dont play in a vacuum so to some degree everything they do (+/-, boxscore, other stats, eye test, accolades) will be affected by who is in the court with them

a) Oh I know RAPM uses multi-year averages. To me that's a whole other issue with the stat. You can't use a roster lineup stat based stat, and just casually used previous seasons. That's always been wonky

b) I agree about regular stats. Those too need lots of context, because sadly the NBA has used the same small dataset for the last 70 years. Only recently have we had more numbers to look at. Hopefully computer AI can be used to go back and scan old footage for newer things, though that has its limits in effectiveness too.

The problem I and many have had for 10+ years with RAPM and +/- stats, is that people attribute them to impact. Nothing wrong with looking at them, or comparing them. But I've seen them used to discredit some players, and elevate others. In ranking discussions, I see the numbers used defacto now, as if they directly correlate to who was better. Defensive +/- especially has been overvalued, which is why I pointed out the crazy 2022 Curry number in comparison to Giannis.

[jalengreen]

Meh.
Manu Ginobili 2001-05 34637 5.01
Ray Allen 2001-05 59629 4.15
Paul Pierce 2001-05 70297 4.00
Baron Davis 2001-05 55937 3.90
Steve Francis 2001-05 56474 3.60
Tracy McGrady 2001-05 62800 3.31
Kobe Bryant 2001-05 76224 3.16
Vince Carter 2001-05 51268 3.03
Jason Kidd 2001-05 72180 3.02
Richard Hamilton 2001-05 65346 2.51
Metta World Peace 2001-05 41667 2.45
Steve Nash 2001-05 67628 2.05
Gilbert Arenas 2001-05 40140 1.86
Michael Jordan 2001-05 18114 1.82
Mike Bibby 2001-05 65841 1.63
Allen Iverson 2001-05 68730 1.58
Peja Stojakovic 2001-05 68135 1.47
Stephon Marbury 2001-05 62231 1.09
Tayshaun Prince 2001-05 37726 1.02
Chauncey Billups 2001-05 61649 0.46
Rick Fox 2001-05 39525 0.45
Bruce Bowen 2001-05 60032 -0.01

Also, this author "forgot" about Holy T-Mac '01-'05 in this topic.
https://www.thespax.com/nba/the-nbas-most-prolific-playoff-scorers-since-1973/

Sent from my Nokia 5.1 Plus using RealGM mobile app

woah, i never thought that this guy:



was on realgm. i joked with my friends that this morb guy must be tracy mcgrady

but yeah this is what i get for tmac's statistical playoff multi-season scoring peaks:



IIRC i may have used a 30 GP requirement? if so 02-08 should've made the cut with 38 points scored per 100 possessions on aprx even relative efficiency

[jalengreen]

But these numbers reflect the team rotations, not the player himself. Which is why the numbers fluctuate from season to season. And the style of coaching will weigh into the numbers too. If a coach has their star with the 2nd team quite a bit as not to give up leads, then that player's +/- numbers will be lesser than a player who's coach takes out all the starters at the same time. Does the coach utilize offensive or defensive heavy rotations? That too will muck up the waters. So many factors.

A good player will be positive, but the degree of which they are depends on the rotations used, not their direct impact.

a) Oh I know RAPM uses multi-year averages. To me that's a whole other issue with the stat. You can't use a roster lineup stat based stat, and just casually used previous seasons. That's always been wonky

b) I agree about regular stats. Those too need lots of context, because sadly the NBA has used the same small dataset for the last 70 years. Only recently have we had more numbers to look at. Hopefully computer AI can be used to go back and scan old footage for newer things, though that has its limits in effectiveness too.

The problem I and many have had for 10+ years with RAPM and +/- stats, is that people attribute them to impact. Nothing wrong with looking at them, or comparing them. But I've seen them used to discredit some players, and elevate others. In ranking discussions, I see the numbers used defacto now, as if they directly correlate to who was better. Defensive +/- especially has been overvalued, which is why I pointed out the crazy 2022 Curry number in comparison to Giannis.

i'm not sure i follow

you have a problem with people attributing RAPM-based metric to impact and you cite factors such as coaching, rotations, etc that complicate things. but do these factors not surely affect impact? why would a metric that attempts to isolate impact *not* be affected by things like rotations?

if a team has poor rotations around its star player, the star player's ability to make an impact may be harmed. so why wouldn't we expect to see RAPM fluctuate?

i guess i don't really understand what you mean when you use the word "impact."

[Morb]

Meh.
Manu Ginobili 2001-05 34637 5.01
Ray Allen 2001-05 59629 4.15
Paul Pierce 2001-05 70297 4.00
Baron Davis 2001-05 55937 3.90
Steve Francis 2001-05 56474 3.60
Tracy McGrady 2001-05 62800 3.31
Kobe Bryant 2001-05 76224 3.16
Vince Carter 2001-05 51268 3.03
Jason Kidd 2001-05 72180 3.02
Richard Hamilton 2001-05 65346 2.51
Metta World Peace 2001-05 41667 2.45
Steve Nash 2001-05 67628 2.05
Gilbert Arenas 2001-05 40140 1.86
Michael Jordan 2001-05 18114 1.82
Mike Bibby 2001-05 65841 1.63
Allen Iverson 2001-05 68730 1.58
Peja Stojakovic 2001-05 68135 1.47
Stephon Marbury 2001-05 62231 1.09
Tayshaun Prince 2001-05 37726 1.02
Chauncey Billups 2001-05 61649 0.46
Rick Fox 2001-05 39525 0.45
Bruce Bowen 2001-05 60032 -0.01

Also, this author "forgot" about Holy T-Mac '01-'05 in this topic.
https://www.thespax.com/nba/the-nbas-most-prolific-playoff-scorers-since-1973/

Sent from my Nokia 5.1 Plus using RealGM mobile app

woah, i never thought that this guy:



was on realgm. i joked with my friends that this morb guy must be tracy mcgrady

but yeah this is what i get for tmac's statistical playoff multi-season scoring peaks:



IIRC i may have used a 30 GP requirement? if so 02-08 should've made the cut with 38 points scored per 100 possessions on aprx even relative efficiency

You did your job, thanks!

Jordan averaged 43.3 points per 100 possessions throughout his illustrious postseason career, by far the most in NBA history. Next up is Kevin Durant at 37.3.

So, next up is T-Mac? Wow, nice, how you "forgot" him?

Sent from my Nokia 5.1 Plus using RealGM Forums mobile app