SGA, Jokic and Giannis and the breaking advanced stats

[dhsilv2]

So you're telling me the old generation players were not so good at a stat which they were not even aware of?
That's surprising.

SGA, Jokic, LeBron might all be the greatest at EPM, but I'll tell you a secret: they pretty suck at SNSFYFN

Spoiler:

SNSFYFN
Some New Stat Fifty Years From Now

No one plays to maximize certain advanced stat, even if one tries, it's hard to do, players play to win or score/assist/rebound...

So maybe it's worth judging players based more on what they do try to maximize...

Is that not to help your team win? That's what EPM is trying to measure. That's it.

[CobraCommander]

Disclaimer:
SGA is having the best guard season in modern history other than Steph in 15-16, and Jokic is playing at his peak as top 10 ATG,they are having amazing ATG seasons, we are watching greatness, and we should be highly appreciative of this, this is more of a discussion of the breaking of advanced stats, what is leading to this, and why outlier seasons are becoming so common nowadays.

EPM combines volume stats and RAPM to come up with estimation of the the points certain player contributes per 100 possessions.
Highest EPM on record is LeBron James in 2009, he had 9.1 EPM, this is also the only 9+ EPM season, and one of the few that are 8+.
SGA is on pace to break the EPM record at 9.2
Jokic is on pace to have the 3rd best EPM season on record (since 2001) at 9.0
This is a list of all 8+ EPM seasons on record

Code: Select all

SGA     25   9.2
LBJ     09   9.1
Jokic   25   9
Jokic   24   8.3
Jokic   23   8.4
Steph   17   8.2
LBJ     10   8.1

In fact, despite the stat being adjusted to pace, since the offensive explosion of 2020s, 6+ EPM became an all NBA tier stat, when it was before exceptional MVP season stat.

In a way, EPM today is a much better indicator than EPM 20 years, if you look at the top 5 in EPM and their order, it actually follows the consensus and order of who the best players are in the league, this wasn't the case 20 years, but comparing great seasons from 00-10 season to great season in 20-25, give huge difference in favour of modern era.


For example, Dirk Nowitzki 2007 MVP season, Mavs started 0-4 and finished 67-15, Dirk dominated the league, he had no rivals, only one that was in contention for MVP was Kobe who carried a non-NBA roster to the PO, Dirk EPM that year 4.9, he ranked 4th, Kobe wasn't in top 5.
Kevin Garnett 2004 season, where he went to WCF and was the best player in the league, was 5.9, to wit, this is worse than Lillard in 2023 and Luka this year.

We are seeing EPM levels that are double absolutely great seasons, in addition to 07 Dirk and 04 KG,Tim Duncan is a top 5 all time player, they never approached levels all NBA players are posting regularly in the last 3-4 years.


BPM models volume stats + effeciency, the breaking of BPM is different than EPM and much less dramatic, it's more of a position thing combined with archetypes that weren't common before, double digit BPM used to mean amazing year of an all time player, now it's more of all NBA year.
Jokic reigns supreme in this stat because it gives massive weight to assists at center possession, and it punishes TOs, so and offensive hub centers who doesn't usually bring the ball up and PFs like Jokic and Giannis have inherent advantage, these are unicorns in terms of style and way of impact.
BPM kings are Jokic, Giannis and LeBron
Jokic repeatedly has 13+ BPM season in his 5 year peak, topping at 13.7 in the year he had no one else on roster in 22.
LBJ had 13.2, 11.7, 11, 11.8, 11
Giannis notable BPMs, 11.2, 11.5 and 10.4 in his 19-21 peak

Guards are typically lower
SGA has 11.5 BPM this year, peak Harden topped at 11 BPM, triple double MVP Westbrook 11.1, he is second only to unanimous MVP Curry in 2016 who topped at 11.9, highest guard ever at this stat.

Notable 10 BPM season -- Steph 15, Durant 14 (lower BPM due to lower assists and being marked as SF), CP3 08 + 09, TMAC 03, Wade 09, Luka 24.

To use same examples of marquee seasons from before, Dirk MVP was 8, KG MVP was 10.2, Shaq 9.3 at the peak of his dominance.

I think it's fair to say Jokic has better season than 07 Dirk or 04 KG, but it's also fair to say it's not 13 Vs 8 in BPM or 9 Vs 4.8 in EPM.
Damian Lillard had a very good season in 23, but it shouldn't be better than anything ATGs like Duncan, KG and Dirk ever posted in terms of EPM.

I think there are lots of factors at play:
Analytics and intensive video analysis moved the league light years ahead in terms of efficiency, elimination of less efficient shots systematically, allocating specific shots to specific players based on their efficiency profile, etc...
Coaching is able to identify game breaking players and utilize them more and more.
Archetypes that were foreign to previous eras, greatest passing big and one of the best passers of all time who is an elite scorer on his own, the best transition and rim pressure force in league history, freak in size, length and athleticism who has guard skills.

Thanks for the thoughtful thread on a board that is a “sea of sameness”. This was good post

[NZB2323]

You can't compare across eras. If Jordan's coach and front office were aware that a certain mix of shots would result in a higher advanced stat, they would have implemented it. All of the players' training has adjusted to advancing these new numbers, but none of them players in the 80's and 90's were training on the basis of this mix of shots. So your judgement too heavily involves front offices and coaching staff in the formula. They are going to keep discovering new stuff and the best offensive player in the NBA will keep pushing that EPM. We're not ever at a peak in raw numbers so it's kind of crazy to mark a point in time and say "wow we're higher than ever". Of course we're higher than ever.

The person who was ranked #1 in EPM in 1992 would be battling for #1 EPM in 2025 if they all knew the same stuff. We can't know how much lower in EPM Lebron/Curry would be in 1992. We can't know how much higher in EPM Michael Jordan would be in 2025. There are too many factors that are not related to the players' skill, talent and effort.

For that reason they keep saying over and over and over and over again: you can't compare across eras so for the love of god please try to stop your thoughts when it tries to go there. Just be happy for whoever is number 1 right now and who was number 1 back then. It's not that hard.

I believe EPM uses data from 2002-present, so no one was #1 in EPM in 1992.

[AleksandarN]

Espn just came out with an article helping explain Jokic unconventional ways of defensive impact. It’s a great read. Here is a little exert

https://www.espn.com/nba/story/_/id/44346405/nikola-jokic-quietly-league-analytics-defender

"A lot of times on out-of-bounds plays, he'll literally move you into the spot. He does it every single game. [Viewers] may not see it, and it might be subtle, but he moves you into your spot," wing Christian Braun told ESPN. "He's going to look at you and tell you, 'Hey, this is what's coming; this was their call.' You'll see him looking at the other coach to hear their call. He knows what's coming and knows the best way to guard it."

Added Nuggets forward Peyton Watson: "I haven't seen him be wrong [when calling out a play]," he told ESPN. "Dude's on another level intellectually with the game. ... So we'll be waiting on the day when he's wrong."

Jokic's memory bank as a defender is robust, according to Nuggets assistant coach Popeye Jones. "If a team tries to go back to [the exact same out-of-bounds play] the next year, he'll still remember it," Jones told ESPN.

[Doctor MJ]

Disclaimer:
SGA is having the best guard season in modern history other than Steph in 15-16, and Jokic is playing at his peak as top 10 ATG,they are having amazing ATG seasons, we are watching greatness, and we should be highly appreciative of this, this is more of a discussion of the breaking of advanced stats, what is leading to this, and why outlier seasons are becoming so common nowadays.

I think there are lots of factors at play:
Analytics and intensive video analysis moved the league light years ahead in terms of efficiency, elimination of less efficient shots systematically, allocating specific shots to specific players based on their efficiency profile, etc...
Coaching is able to identify game breaking players and utilize them more and more.
Archetypes that were foreign to previous eras, greatest passing big and one of the best passers of all time who is an elite scorer on his own, the best transition and rim pressure force in league history, freak in size, length and athleticism who has guard skills.

Good thoughts though I won't co-sign your title.

To me here's the key thing going on:

The tactical arms race is causing the players most capable of value to be more valuable per minute than they were in the past, and this isn't just a databall era thing. It's been a gradually process across generations.

One aspect of it is that players are playing less than they did in the past and are expected to not conserve energy while they play as a result. Some players are still conserving energy of course, and those players are showing up as less valuable by these per minute stats.

It's important imho to not assume that current top players are inherently more capable at basketball than those that came before though there is a trend in that direction. Players from the past would get utilized more like this today if they could be, and many of them (but not all of them) could be.

[Mavrelous]

Disclaimer:
SGA is having the best guard season in modern history other than Steph in 15-16, and Jokic is playing at his peak as top 10 ATG,they are having amazing ATG seasons, we are watching greatness, and we should be highly appreciative of this, this is more of a discussion of the breaking of advanced stats, what is leading to this, and why outlier seasons are becoming so common nowadays.

I think there are lots of factors at play:
Analytics and intensive video analysis moved the league light years ahead in terms of efficiency, elimination of less efficient shots systematically, allocating specific shots to specific players based on their efficiency profile, etc...
Coaching is able to identify game breaking players and utilize them more and more.
Archetypes that were foreign to previous eras, greatest passing big and one of the best passers of all time who is an elite scorer on his own, the best transition and rim pressure force in league history, freak in size, length and athleticism who has guard skills.

Good thoughts though I won't co-sign your title.

To me here's the key thing going on:

The tactical arms race is causing the players most capable of value to be more valuable per minute than they were in the past, and this isn't just a databall era thing. It's been a gradually process across generations.

One aspect of it is that players are playing less than they did in the past and are expected to not conserve energy while they play as a result. Some players are still conserving energy of course, and those players are showing up as less valuable by these per minute stats.

It's important imho to not assume that current top players are inherently more capable at basketball than those that came before though there is a trend in that direction. Players from the past would get utilized more like this today if they could be, and many of them (but not all of them) could be.

The numbers back the graduality of the process indeed.
I totally co-sign the bottom paragraph.

[canada_dry]

Umm...nash was dirks rival for mvp in 2007...

I wanted to just enjoy reading the post but i couldn't get past that historical inaccuracy

Sent from my SM-G960W using RealGM mobile app

[Saints14]

Jokic has a 13.9 BPM but it's always interesting how for whatever reason his DBPM is so high. This year 3.3 DBPM. Everyone knows Jokic is not THAT impactful on the defensive end.

I feel that the stat might overrate rebounding for centers when in reality a lot of rebounds can be uncontested or given up, and of course if a teammate is helping to box someone out that rebound is not actually the center.

Imo the bigger measure is how many rebounds are contested, how many are being given up, and that requires looking at the opponent stats and an interesting thing about Denver is that they are top 5 in opponent offensive rebounds which means that Jokic as a rebounder might be a little overrated.

BPM uses assists as a proxy for spacial awareness because...that's what the RAPM correlation data shows. It's kinda right, but obviously misses for Jokic. But remember you calculate BPM first. You don't calculate O and D separately and add them.

Oddly all the RAPM data has Jokic's defense dropping some this year and BPM followed. So while it's over stated, clearly, it's tracking reasonably well after you realize it's giving him about a 2 point bump. And likely those 2 points should just be added back to his OBPM.

Yeah, and I believe there is some variable in the formula that is AST x REB that was meant as a proxy for good defense because before 10 years ago only wings with good feel had both high AST and REB values, and that correlated to good defenders. Westbrook and now Jokic have broken that though

[JM00n69]

All three are having elite seasons all time.

But imo, trying to boil down an NBA players impact/contribution to their team and also across the leauge against changing lineups over an 82 game season to one nr seems impossible to me. And that's over one season. To think you can compare different eras just by that one nr is silly and clearly not quantifiable.

[dhsilv2]

Jokic has a 13.9 BPM but it's always interesting how for whatever reason his DBPM is so high. This year 3.3 DBPM. Everyone knows Jokic is not THAT impactful on the defensive end.

I feel that the stat might overrate rebounding for centers when in reality a lot of rebounds can be uncontested or given up, and of course if a teammate is helping to box someone out that rebound is not actually the center.

Imo the bigger measure is how many rebounds are contested, how many are being given up, and that requires looking at the opponent stats and an interesting thing about Denver is that they are top 5 in opponent offensive rebounds which means that Jokic as a rebounder might be a little overrated.

BPM uses assists as a proxy for spacial awareness because...that's what the RAPM correlation data shows. It's kinda right, but obviously misses for Jokic. But remember you calculate BPM first. You don't calculate O and D separately and add them.

Oddly all the RAPM data has Jokic's defense dropping some this year and BPM followed. So while it's over stated, clearly, it's tracking reasonably well after you realize it's giving him about a 2 point bump. And likely those 2 points should just be added back to his OBPM.

Yeah, and I believe there is some variable in the formula that is AST x REB that was meant as a proxy for good defense because before 10 years ago only wings with good feel had both high AST and REB values, and that correlated to good defenders. Westbrook and now Jokic have broken that though

That was the OLD VORP, which is when Westbrook broke things. The new one reduced that skew but I think added positional stuff which has it's own issues. 3.0 is in the works.

[Raps in 4]

FWIW

Read on Twitter

I'm not sure what the solution is when it comes to adjusting impact metrics in the NBA, but what I like about baseball stats is they allow easy comparisons across eras.

Someone like Babe Ruth wouldn't make a high school baseball team today, but he's the all-time WAR leader because he was head-and-shoulders the best player in his era. Duncan was a top-3 player throughout his prime, and arguably just as dominant, in his era, as guys like Jokic and Steph are today, but impact metrics don't necessarily reflect that.

Yes, players today are more skilled, but that should be irrelevant when assessing ATG impact.

[The Master]

Yes, players today are more skilled, but that should be irrelevant when assessing ATG impact.

Yeah, but it's always about:

a) impact in specific era,
b) level of competition in specific era,

when we speak about ATG discussions.

So, for example, we know that Bird and Magic individually dominated 80s and while they were still in their primes, Jordan clearly outplayed them in overall production. Do we really need stats for multi-era comparisons? Yeah, it would be helpful, but we can also quantitatively determine what is a value of Jordan on mediocre Bulls team outplaying prime Bird and Magic, and winning MVP in 1988. We can also scale statistical domination over player's era easily. LeBron having +100% of VORP production over prime Bryant or Dwight in regular season in his Cavs years is also nuts, that's why we can estimate him being much better than Giannis in 2019, for example, even if Giannis as well had +30 PER and +10 BPM MVP year.

[Doctor MJ]

I'm not sure what the solution is when it comes to adjusting impact metrics in the NBA, but what I like about baseball stats is they allow easy comparisons across eras.

Someone like Babe Ruth wouldn't make a high school baseball team today, but he's the all-time WAR leader because he was head-and-shoulders the best player in his era. Duncan was a top-3 player throughout his prime, and arguably just as dominant, in his era, as guys like Jokic and Steph are today, but impact metrics don't necessarily reflect that.

Yes, players today are more skilled, but that should be irrelevant when assessing ATG impact.

So, I would say that in general basketball analytics leans not less but more into rate & relative statistics (rather than absolutes) than baseball does. So, unless I'm misunderstanding you, the thing you like about baseball stats is already done better by basketball stats.

From PER onward basically every advanced stat has some kind of built into rate/relative aspect built in, and that absolutely includes impact stats.

Now, you can of course take an already relative-ized stat and normalize for variance, and this is oftentimes a good thing to do to make better comparisons across eras, but such normalization typically leads us away from understanding how impactful guys are within a given season.

[Sane]

You can't compare across eras. If Jordan's coach and front office were aware that a certain mix of shots would result in a higher advanced stat, they would have implemented it. All of the players' training has adjusted to advancing these new numbers, but none of them players in the 80's and 90's were training on the basis of this mix of shots. So your judgement too heavily involves front offices and coaching staff in the formula. They are going to keep discovering new stuff and the best offensive player in the NBA will keep pushing that EPM. We're not ever at a peak in raw numbers so it's kind of crazy to mark a point in time and say "wow we're higher than ever". Of course we're higher than ever.

The person who was ranked #1 in EPM in 1992 would be battling for #1 EPM in 2025 if they all knew the same stuff. We can't know how much lower in EPM Lebron/Curry would be in 1992. We can't know how much higher in EPM Michael Jordan would be in 2025. There are too many factors that are not related to the players' skill, talent and effort.

For that reason they keep saying over and over and over and over again: you can't compare across eras so for the love of god please try to stop your thoughts when it tries to go there. Just be happy for whoever is number 1 right now and who was number 1 back then. It's not that hard.

I believe EPM uses data from 2002-present, so no one was #1 in EPM in 1992.

Thank you for the technical point, but the point stands for whoever would've been #1 in EPM in 1992.