RealGM

yall been hittin mj with the "stacked team", "ppg+steals aint ****", "impact numbers arent that high", but kd be on the most loaded teams ever and rackin up hellla ppg and ts so im wonderin what his impact is

Not up to the standards of the premier players of his era impact-wise (LeBron, Curry, Paul), but he's usually been plenty good in every measure as far as I can remember.

Unlike for MJ, where we are operating on rough estimations and incomplete samples, we do have more granular data for Durant, so the criticisms are probably more valid here, I feel.

It's more like a commentary on even though KD has extremely high PPG and TS which are generally a good thing (And for Durant it is a good thing), there are major holes in the nuance of his game that makes him lag behind few other elites even if their PPG/TS combo isn't necessarily better.

Durant plays in the data ball era so this isn't just hearsay like when people try to slam great players because they dislike their playing style. This is an explanation for why Durant's impact stats might not be as good as one would think.



It's very easy to see a player with 30 PPG, 65 TS% and a bunch of tier 1 awards and go he's one of the GOATs without looking at any thing else. There are fans who think Durant is top ten all time, granted some of them probably just aren't taking into account older players but they also think Durant is like Lebron James good basically.

Sure if we take team-based +- and on/off stats without context as "pure impact" then he doesn't look as good but no wing scorer is going to come out on top in those type of stats. The top performers in those kinds of stats are overwhelmingly playmakers and defensive anchors on teams that put all their eggs in one basket and don't have capable back-ups for those roles so the team crumbles on offense or defense without them.

Depends what you mean by 'high' here. Is KD GOAT tier - no, probably not. But I think you can count on your fingers the guys with prime runs clearly better from the databall era - LeBron/Duncan/KG/Shaq/Steph. There's a few other guys I'd take, but KD isn't far behind them or anything.

He's an all universe scorer and not really bad at anything else on the court, he's obviously a very very good player and has been a top ~5 player for more than a decade when healthy.

I like to think of Durant as a weapon more than a player. I'm here to score points and chew bubblegum

henshao wrote:I like to think of Durant as a weapon more than a player. I'm here to score points and chew bubblegum

He is also a good playmaker for a Wing, provides defensive value as a rim protector from the wing position and is exceptional on or off-ball.

I think it can or may not be high depending on what the gauge is. Historically and even just within the data-ball era, Durant of-course has one of the better profiles amongst all players - but by no means is the the "champ" of the data-ball era and falls somewhat behind some other players from this century around the top 10-15 A.T. (most-so Duncan, Garnett, O'Neal, James, Curry).

Some examples:

On/Off Wise, Durant has three seasons >= 16.0 points / 100, three from 10-13 points, five from 6-10. Using this for single-year PS is more on the noisy end.. but FWIW Durant has two seasons over 10.0 and five more from 6-8.2.

PIPM grades him as a "MVP candidate" (value of 4.7-6.8) in six different seasons: 2010/13/14/16/17/19, but never at the "best player" threshold of a > 6.8 value.

I don't have full access to Augmented +/- or its per game variant at the moment, but FWIW - the metric isn't the fondest on his OKC playoff days where he tops out a little over 4 and hovers over 3 for many prime years.
(https://backpicks.com/2018/06/10/aupm-2-0-the-top-playoff-performers-of-the-databall-era/)

These are just metrics at the end of the day, where no one-number should tell the whole story. FWIW, I think that some sell him short on the offensive end with the consideration his game is more valuable serving as a ceiling raiser due to how well his scoring scales up /meshes up next to superstars and high-level talent. This has led to very stellar impact from a team results standpoint regardless of location. Impact that tries to separate his individual offensive value from the rest may under-sell his true value due to both systems somewhat being a better environment for Westbrook and Curry to floor-raise / indirectly pinball their values here. Defensively, I find one-numbers more closer to my view where Durant grades from a marginal to a slight positive value through his prime.

HeartBreakKid wrote:It's more like a commentary on even though KD has extremely high PPG and TS which are generally a good thing (And for Durant it is a good thing), there are major holes in the nuance of his game that makes him lag behind few other elites even if their PPG/TS combo isn't necessarily better.

Durant plays in the data ball era so this isn't just hearsay like when people try to slam great players because they dislike their playing style. This is an explanation for why Durant's impact stats might not be as good as one would think.



It's very easy to see a player with 30 PPG, 65 TS% and a bunch of tier 1 awards and go he's one of the GOATs without looking at any thing else. There are fans who think Durant is top ten all time, granted some of them probably just aren't taking into account older players but they also think Durant is like Lebron James good basically.

What are the major holes in the nuance of his game?

Yes, but overrated at times by casual fans. Echoing what others have said here, I've always felt that the idea of KD as a player is greater than the reality (but the reality is still pretty incredible of course).

The Explorer wrote:

HeartBreakKid wrote:It's more like a commentary on even though KD has extremely high PPG and TS which are generally a good thing (And for Durant it is a good thing), there are major holes in the nuance of his game that makes him lag behind few other elites even if their PPG/TS combo isn't necessarily better.

Durant plays in the data ball era so this isn't just hearsay like when people try to slam great players because they dislike their playing style. This is an explanation for why Durant's impact stats might not be as good as one would think.



It's very easy to see a player with 30 PPG, 65 TS% and a bunch of tier 1 awards and go he's one of the GOATs without looking at any thing else. There are fans who think Durant is top ten all time, granted some of them probably just aren't taking into account older players but they also think Durant is like Lebron James good basically.

What are the major holes in the nuance of his game?

Basically, he’s really good at getting good shots for himself, but not as good at getting shots for others. His passing has gotten better over the years, but is still not at the level of the top guys and a lot of his scoring is just from shooting over guys, not from driving to the basket where the defense will collapse and leave someone else open if he doesn’t shoot. Also, while KD’s a good defender, his impact defensively has consistently been only a little above league average, unlike guys like Kawhi and LeBron who have shown really elite impact defensively in their best years.

Chris Paul had better impact stats than KD 10 of their first 11 years in the league together. Steph has had better impact stats I believe every year since 2015. Jokic and Giannis’s best years are much better than anything KD ever did as well. Kawhi’s peak years are better than KD’s too. And of course LeBron crushes him. So impact stats would see KD more of a “top ten of his era” guy than a “top ten all time” guy.

The Explorer wrote:

HeartBreakKid wrote:It's more like a commentary on even though KD has extremely high PPG and TS which are generally a good thing (And for Durant it is a good thing), there are major holes in the nuance of his game that makes him lag behind few other elites even if their PPG/TS combo isn't necessarily better.

Durant plays in the data ball era so this isn't just hearsay like when people try to slam great players because they dislike their playing style. This is an explanation for why Durant's impact stats might not be as good as one would think.



It's very easy to see a player with 30 PPG, 65 TS% and a bunch of tier 1 awards and go he's one of the GOATs without looking at any thing else. There are fans who think Durant is top ten all time, granted some of them probably just aren't taking into account older players but they also think Durant is like Lebron James good basically.

What are the major holes in the nuance of his game?

For a perimeter oriented wing player, Durant's ball handling is very poor; his ball security is notably poor. Durant has far too many playoff series where he has more Turnovers than Assists. His Turnover Rate jumps far too high without a lead creator on the floor with him.

Also throughout Durant's prime he's shown to have limited court vision and court mapping skills. For a perimeter oriented wing he was a mediocre a passer throughout his prime.

He has shown in many playoff series too be far too thirsty of a isolationist scorer, playing a game of 'your turn, my turn with Westbrook.

Without a lead creator next to him his scoring effficiency craters substantially lower than other ATG great wing players meaning his overall scoring value is a bit too teammate dependent.

He’s put a pretty great career together obviously, but the case for him as a top 10 player of all time is pretty damn weak if we’re being objective about it.

Dutchball97 wrote:Sure if we take team-based +- and on/off stats without context as "pure impact" then he doesn't look as good but no wing scorer is going to come out on top in those type of stats. The top performers in those kinds of stats are overwhelmingly playmakers and defensive anchors on teams that put all their eggs in one basket and don't have capable back-ups for those roles so the team crumbles on offense or defense without them.

Or maybe raw scoring volume is a lot more replaceable and less essential than you think it is. Iverson and Kobe were their team’s lead playmakers yet did not exactly generate incredible impact.

I think raw impact signals mildly undersell Durant’s real value in the postseason in perhaps a similar way as it can for Kobe, but Durant’s typical inability to support a high-level team that way (I suppose outside of that laudable 26-game stretch in the 2014 regular season) is because of his own limitations. If he were a better defender or playmaker, he could contribute to his team beyond raw scoring or “gravity”, even without being a true lead creator or defensive anchor.

"impact" stats do not convert to winning or championships
Is he as good as Steph or Kobe? No, at least not yet
Is he better than KG or CP3? Yes
Strangely, many "impact" stats will suggest to you that KG and CP3 were/are better than Curry and Kobe. I say strange because those do not agree with my eye test and the actual results.

True, impact does not guarantee championships; being on championship teams does that.

dygaction wrote:"impact" stats do not convert to winning or championships
Is he as good as Steph or Kobe? No, at least not yet
Is he better than KG or CP3? Yes
Strangely, many "impact" stats will suggest to you that KG and CP3 were/are better than Curry and Kobe. I say strange because those do not agree with my eye test and the actual results.

Actually, winning is literally EXACTLY what impact stats measure. It's all about team performance. Curry has fantastic impact stats that are just as good as KG's and CP3's. I just looked up 25 year RAPM and the three of them rank as the 3rd, 4th, and 5th best players overall for that sample. Curry's slightly behind them, but a different counting method would just as easily put him on top. In fact, if he rates slightly lower cumulatively, it's probably just it took him a few years to really find his footing in the league where KG and CP3 were superstars from the jump. Kobe's a different animal. Kobe does rate much lower than those guys because his team didn't perform as well with him on the floor. He's gonna rank lower since despite having the best teammates, his team performed the worst with him on the floor. Here's the career point differential for their teams per 100 possessions:

Team performance with Curry: +8.4
Team performance with CP3: +6.9
Team performance with KG: +5.2
Team performance with Kobe: +4.2

How Kobe's teammates performed with him on the bench: -0.4
How CP3's teammates performed with him on the bench: -2.8
How Currys teammates performed with him on the bench: -3.1
How KG's teammates performed with him on the bench: -5.1

Now since the team played well with him on the bench, Kobe still managed to be on a lot of very successful teams. For example, the 3-peat Lakers from 2000-2002 were a very successful team that's often remembered as equally Shaq and Kobe's team since both guys shot the ball about the same amount. A more granular look at the data though would show that it was actually Shaq responsible for the winning with Kobe playing a much smaller role. Over the 3 year span, here's how the team performed with just Shaq, with just Kobe, and with both:

Just Shaq, no Kobe: 25-6 (.781 winning percentage)
Kobe and Shaq together: 144-47 (.754 winning percentage)
Just Kobe, no Shaq: 12-11 (.522 winning percentage)
No Shaq or Kobe: 0-1 (.000 winning percentage)

As you can see, when Kobe fell out of the lineup, there was no negative effect. The winning percentage actually went up a little bit. Whereas when Shaq left the lineup, the Lakers fell to approximately the level of the 2005-2007 teams that struggled to make the playoffs and never got out of the first round. Someone just paying attention to how many points each guy puts in the basket might say that Kobe and Shaq each won half of three championships together, but impact stats are going to look at this kind of scenario and see Shaq as a dominant force who carried teams to championships and Kobe as a good player who helped him. This is borne out, not just game by game, but possession by possession as well.

In 2000, PI RAPM ranked Shaq as the #1 player in the league with an RAPM of +8.52 (+6.21 offense, +2.31 defense), KG as the #2 player in the league with an RAPM of +7.19 (+3.48 offense, +3.71 defense), and Kobe as the #125 player with an RAPM of +0.11 (+1.81 offense, -1.70 defense). Someone watching casually could easily miss that Kobe was a negative defender that year only focusing on what happens when he has the ball, but impact stats catch everything by measuring the team performance on every possession with and without that player. Kevin Garnett might not seem like as much of a winner that season since his team lost in the first round, but the way that he elevated an absolutely nothing cast to 50 wins is actually a much more impressive feat than Kobe would be able to manage with better weak casts from 2005-2007.

If you were to look at Kevin Durant's stats playing with Stephen Curry, you'd find a very similar situation to what happened with Shaq and Kobe. When Durant missed games, the team actually got BETTER going 27-4 over a 3-year window. Whereas when Curry missed games and Durant played, the team dropped all the way to 23-17 barely even good enough to make the playoffs. So while a casual fan will watch the team, see that Durant and Curry score about the same number of points and conclude that they're equally responsible for the Warriors success, impact stats which measure winning will give the lion's share of the credit to Curry since he's the key factor in the team's performance. That's because Curry is a much better passer, playmaker, and off-ball cutter, has greater gravity, and is much better at getting open looks for his teammates while Kevin Durant's skills mostly just get open looks for Kevin Durant.

iggymcfrack wrote:

dygaction wrote:"impact" stats do not convert to winning or championships
Is he as good as Steph or Kobe? No, at least not yet
Is he better than KG or CP3? Yes
Strangely, many "impact" stats will suggest to you that KG and CP3 were/are better than Curry and Kobe. I say strange because those do not agree with my eye test and the actual results.

Actually, winning is literally EXACTLY what impact stats measure. It's all about team performance. Curry has fantastic impact stats that are just as good as KG's and CP3's. I just looked up 25 year RAPM and the three of them rank as the 3rd, 4th, and 5th best players overall for that sample. Curry's slightly behind them, but a different counting method would just as easily put him on top. In fact, if he rates slightly lower cumulatively, it's probably just it took him a few years to really find his footing in the league where KG and CP3 were superstars from the jump. Kobe's a different animal. Kobe does rate much lower than those guys because his team didn't perform as well with him on the floor. He's gonna rank lower since despite having the best teammates, his team performed the worst with him on the floor. Here's the career point differential for their teams per 100 possessions:

Team performance with Curry: +8.4
Team performance with CP3: +6.9
Team performance with KG: +5.2
Team performance with Kobe: +4.2

How Kobe's teammates performed with him on the bench: -0.4
How CP3's teammates performed with him on the bench: -2.8
How Currys teammates performed with him on the bench: -3.1
How KG's teammates performed with him on the bench: -5.1

Now since the team played well with him on the bench, Kobe still managed to be on a lot of very successful teams. For example, the 3-peat Lakers from 2000-2002 were a very successful team that's often remembered as equally Shaq and Kobe's team since both guys shot the ball about the same amount. A more granular look at the data though would show that it was actually Shaq responsible for the winning with Kobe playing a much smaller role. Over the 3 year span, here's how the team performed with just Shaq, with just Kobe, and with both:

Just Shaq, no Kobe: 25-6 (.781 winning percentage)
Kobe and Shaq together: 144-47 (.754 winning percentage)
Just Kobe, no Shaq: 12-11 (.522 winning percentage)
No Shaq or Kobe: 0-1 (.000 winning percentage)

As you can see, when Kobe fell out of the lineup, there was no negative effect. The winning percentage actually went up a little bit. Whereas when Shaq left the lineup, the Lakers fell to approximately the level of the 2005-2007 teams that struggled to make the playoffs and never got out of the first round. Someone just paying attention to how many points each guy puts in the basket might say that Kobe and Shaq each won half of three championships together, but impact stats are going to look at this kind of scenario and see Shaq as a dominant force who carried teams to championships and Kobe as a good player who helped him. This is borne out, not just game by game, but possession by possession as well.

In 2000, PI RAPM ranked Shaq as the #1 player in the league with an RAPM of +8.52 (+6.21 offense, +2.31 defense), KG as the #2 player in the league with an RAPM of +7.19 (+3.48 offense, +3.71 defense), and Kobe as the #125 player with an RAPM of +0.11 (+1.81 offense, -1.70 defense). Someone watching casually could easily miss that Kobe was a negative defender that year only focusing on what happens when he has the ball, but impact stats catch everything by measuring the team performance on every possession with and without that player. Kevin Garnett might not seem like as much of a winner that season since his team lost in the first round, but the way that he elevated an absolutely nothing cast to 50 wins is actually a much more impressive feat than Kobe would be able to manage with better weak casts from 2005-2007.

If you were to look at Kevin Durant's stats playing with Stephen Curry, you'd find a very similar situation to what happened with Shaq and Kobe. When Durant missed games, the team actually got BETTER going 27-4 over a 3-year window. Whereas when Curry missed games and Durant played, the team dropped all the way to 23-17 barely even good enough to make the playoffs. So while a casual fan will watch the team, see that Durant and Curry score about the same number of points and conclude that they're equally responsible for the Warriors success, impact stats which measure winning will give the lion's share of the credit to Curry since he's the key factor in the team's performance. That's because Curry is a much better passer, playmaker, and off-ball cutter, has greater gravity, and is much better at getting open looks for his teammates while Kevin Durant's skills mostly just get open looks for Kevin Durant.

Those averaged numbers can be very deceiving and rotation dependent. Similar to soccer, game by game wins are much more relevant. Three 1:0 games and one 0:5 have higher values than two 5:0 games and two 0:1 losses.

dygaction wrote:

iggymcfrack wrote:

dygaction wrote:"impact" stats do not convert to winning or championships
Is he as good as Steph or Kobe? No, at least not yet
Is he better than KG or CP3? Yes
Strangely, many "impact" stats will suggest to you that KG and CP3 were/are better than Curry and Kobe. I say strange because those do not agree with my eye test and the actual results.

Actually, winning is literally EXACTLY what impact stats measure. It's all about team performance. Curry has fantastic impact stats that are just as good as KG's and CP3's. I just looked up 25 year RAPM and the three of them rank as the 3rd, 4th, and 5th best players overall for that sample. Curry's slightly behind them, but a different counting method would just as easily put him on top. In fact, if he rates slightly lower cumulatively, it's probably just it took him a few years to really find his footing in the league where KG and CP3 were superstars from the jump. Kobe's a different animal. Kobe does rate much lower than those guys because his team didn't perform as well with him on the floor. He's gonna rank lower since despite having the best teammates, his team performed the worst with him on the floor. Here's the career point differential for their teams per 100 possessions:

Team performance with Curry: +8.4
Team performance with CP3: +6.9
Team performance with KG: +5.2
Team performance with Kobe: +4.2

How Kobe's teammates performed with him on the bench: -0.4
How CP3's teammates performed with him on the bench: -2.8
How Currys teammates performed with him on the bench: -3.1
How KG's teammates performed with him on the bench: -5.1

Now since the team played well with him on the bench, Kobe still managed to be on a lot of very successful teams. For example, the 3-peat Lakers from 2000-2002 were a very successful team that's often remembered as equally Shaq and Kobe's team since both guys shot the ball about the same amount. A more granular look at the data though would show that it was actually Shaq responsible for the winning with Kobe playing a much smaller role. Over the 3 year span, here's how the team performed with just Shaq, with just Kobe, and with both:

Just Shaq, no Kobe: 25-6 (.781 winning percentage)
Kobe and Shaq together: 144-47 (.754 winning percentage)
Just Kobe, no Shaq: 12-11 (.522 winning percentage)
No Shaq or Kobe: 0-1 (.000 winning percentage)

As you can see, when Kobe fell out of the lineup, there was no negative effect. The winning percentage actually went up a little bit. Whereas when Shaq left the lineup, the Lakers fell to approximately the level of the 2005-2007 teams that struggled to make the playoffs and never got out of the first round. Someone just paying attention to how many points each guy puts in the basket might say that Kobe and Shaq each won half of three championships together, but impact stats are going to look at this kind of scenario and see Shaq as a dominant force who carried teams to championships and Kobe as a good player who helped him. This is borne out, not just game by game, but possession by possession as well.

In 2000, PI RAPM ranked Shaq as the #1 player in the league with an RAPM of +8.52 (+6.21 offense, +2.31 defense), KG as the #2 player in the league with an RAPM of +7.19 (+3.48 offense, +3.71 defense), and Kobe as the #125 player with an RAPM of +0.11 (+1.81 offense, -1.70 defense). Someone watching casually could easily miss that Kobe was a negative defender that year only focusing on what happens when he has the ball, but impact stats catch everything by measuring the team performance on every possession with and without that player. Kevin Garnett might not seem like as much of a winner that season since his team lost in the first round, but the way that he elevated an absolutely nothing cast to 50 wins is actually a much more impressive feat than Kobe would be able to manage with better weak casts from 2005-2007.

If you were to look at Kevin Durant's stats playing with Stephen Curry, you'd find a very similar situation to what happened with Shaq and Kobe. When Durant missed games, the team actually got BETTER going 27-4 over a 3-year window. Whereas when Curry missed games and Durant played, the team dropped all the way to 23-17 barely even good enough to make the playoffs. So while a casual fan will watch the team, see that Durant and Curry score about the same number of points and conclude that they're equally responsible for the Warriors success, impact stats which measure winning will give the lion's share of the credit to Curry since he's the key factor in the team's performance. That's because Curry is a much better passer, playmaker, and off-ball cutter, has greater gravity, and is much better at getting open looks for his teammates while Kevin Durant's skills mostly just get open looks for Kevin Durant.

Those averaged numbers can be very deceiving and rotation dependent. Similar to soccer, game by game wins are much more relevant. Three 1:0 games and one 0:5 have higher values than two 5:0 games and two 0:1 losses.

-The very most deceiving is just looking at how the team that the player is on does (i.e. counting rings for Kobe and Durant).

-The next level from that is to look at how the team does with the player on the floor. (i.e. looking at the point differentials with the player on the floor I listed earlier)

-The level after that is to look at on/off, seeing how the team does with the player on the floor compared to how they do when he’s off the floor (i.e. subtracting the 2 numbers I listed earlier from each other)

All of these are very simple and easy to understand. However as you say, these can be rotation dependent and really hurt someone who plays a lot of minutes with a bad bench unit for example. That’s why people started using more complex formulas to make more true impact stats. These teammate adjustments are where we get adjusted plus-minus (APM) and regularized adjusted plus minus (RAPM). These stats try to account for the teammates that you’re on the floor with each possession to get a more accurate picture of how the team performs relative to expectations with you in the lineup.

This can still be pretty noisy for a season though, especially in the past when iron man players like Kevin Garnett might be on the floor for as much as 80% of a team’s minutes or also when 2 teammates are closely linked and play most of their minutes together. That’s why instead of just trying to run a regression on a single seasons on/off data, many formulas started using a prior. Simple prior informed RAPM (PI RAPM) uses the previous seasons numbers as a baseline to work from to get a better idea just how good those teammates really are. The extra season of data tends to give much more accurate results to the point that it’s shocking how often they get the best player right.

However, many modern impact stats want to contain themselves to one season so they use a box score prior instead. First they run regressions to calculate how each stat correlates to winning and then they use the given season’s numbers to set a baseline for the teammate adjustments before they regress those with on/off data to come up with a score.

In every case though, they’re taking the same basic thing (winning) and fine tuning the adjustments to more accurately show how well a player contributes to increasing that vs. what a similar player would do on their place.

As for your point about game wins being more relevant than point differential, this is also something that has been highly tested. If you look at how likely a team is to win in the future, in every case their point differential is more predictive of future results than their record. A 1-3 soccer team who’s scored 5 goals and given up 3 is on average going to win more games in the future than a 3-1 soccer team who’s scored 3 goals and given up 5. In fact, point differential is so much more predictive that professional sports bettors and bookmakers basically ignore game results in their formulas. If a team is favored by 7 points who has won all their games by close margins, their odds of winning on the moneyline will be the exact same as another 8 point favorite who has won a few games by huge margins and lost some close ones.

However, clutch stats for individual basketball players do exist and it’s true that there are some half-court creators who are especially valuable in close games down the stretch. If you want to look these up and give a benefit to the players who perform well in these scenarios, that’s very valid. Just know that since the advent of play-by-play data around the turn of the century, the 2 players who have performed the absolute best in clutch situations are Dirk Nowitzki….. and Chris Paul. So if anything these distribution type issues should make Chris Paul one of the most underrated players by impact stats due to his incredible clutch ability in the half court.

AEnigma wrote:

Dutchball97 wrote:Sure if we take team-based +- and on/off stats without context as "pure impact" then he doesn't look as good but no wing scorer is going to come out on top in those type of stats. The top performers in those kinds of stats are overwhelmingly playmakers and defensive anchors on teams that put all their eggs in one basket and don't have capable back-ups for those roles so the team crumbles on offense or defense without them.

Or maybe raw scoring volume is a lot more replaceable and less essential than you think it is. Iverson and Kobe were their team’s lead playmakers yet did not exactly generate incredible impact.

I think raw impact signals mildly undersell Durant’s real value in the postseason in perhaps a similar way as it can for Kobe, but Durant’s typical inability to support a high-level team that way (I suppose outside of that laudable 26-game stretch in the 2014 regular season) is because of his own limitations. If he were a better defender or playmaker, he could contribute to his team beyond raw scoring or “gravity”, even without being a true lead creator or defensive anchor.

Funnily enough with how far apart our approaches seem to be sometimes, I saw what you value in modern times
in a recent post and it's actually very similar as what I do with stats like EPM and LEBRON being the gold standard.

+- stats look at what happens when you insert a player into a line-up. How much direct "impact" adding or subtracting a certain player has on winning. Box stats look at how much someone is producing and how efficiently they are doing so.

I'm of the opinion you need both to even come close to approximating how good someone is and I simply don't agree that +- stats alone are enough (or somehow better/more pure without adding a box component) when they're so incredibly situational.