DARKO and LEBRON's Bogus Journey : NBA Statistics

RorschachsMask

Member
SoSH Member
Aug 23, 2011
3,364
Lynn
I haven’t checked today but last I saw Tatum ranked like this in the following

7th in LEBRON, 4th in wins added.
5th on overall RAPTOR, 2nd in wins added.
9th in DPM
Tied for 8th in EPM, 4th in wins added.
5th in RPM, 3rd in their wins added.

And that’s with him shooting 34% from three.


And Brown is the 6th best C by that measure.
Yeah for whatever reasons, advanced stats aren’t that high on Jaylen. Think he is only top 30-40 in one or two of them, the rest is lower.
 

Euclis20

Member
SoSH Member
Aug 3, 2004
5,385
Imaginationland
Interesting that Rudy Gobert is so high, given how little he does on offense.
What little he does, he does REALLY well. 15.4 ppg isn't nothing, and he leads the league in FG%, eFG%, TS%...It seems that there are a few advanced rating systems that are suckers for super high efficiency big men who cannot create their own shot (Bref has Rob Williams as #1 in offensive rating, by a lot).
 

RorschachsMask

Member
SoSH Member
Aug 23, 2011
3,364
Lynn
EPM, LEBRON, and DARKO are by far the three highest regarded free ones. Pretty similar results for all of them too, though obviously some variation.

50189
 
Last edited:

Jimbodandy

Member
SoSH Member
Jan 31, 2006
8,599
around the way
For those of us less versed in basketball stats, can you give a brief overview of why DARKO is your choice?
I'm not a statistician, but it seems to be the one that got the regression the rightest, if that makes sense. Secondly, it's an impact stat, not a box score based one. And finally, its DPM weighs both defense and offense, while also allowing you to separate them.

Tl;dr; I'm a skeptic, and DARKO laps the field in the eyeball test.
 

chilidawg

Member
SoSH Member
Jan 22, 2015
4,226
Cultural hub of the universe
I'm not a statistician, but it seems to be the one that got the regression the rightest, if that makes sense. Secondly, it's an impact stat, not a box score based one. And finally, its DPM weighs both defense and offense, while also allowing you to separate them.

Tl;dr; I'm a skeptic, and DARKO laps the field in the eyeball test.
Is there a Darko list version where we can see where all players? That'd be a good way to compare it to the different metrics.
 

RorschachsMask

Member
SoSH Member
Aug 23, 2011
3,364
Lynn
Last edited:
Jan 22, 2006
183
I'm not a statistician, but it seems to be the one that got the regression the rightest, if that makes sense. Secondly, it's an impact stat, not a box score based one. And finally, its DPM weighs both defense and offense, while also allowing you to separate them.

Tl;dr; I'm a skeptic, and DARKO laps the field in the eyeball test.
Reading up on DARKO, it seems like it is the closest analogue to the statcast suite of xERA, xAVE, xHR, in other words it looks at individual events versus historical data points to tell you how something should behave in the future. Is that a fair assessment?

Too many tools/stats get muddied in the descriptive vs predictive quagmire, makes it hard for a dummy like me to figure out what they're actually trying to communicate to you.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
39,756
I do not believe @bowiac said that teams are "moving away" from all-in-one metrics but the full quote is below. My interpretation is that the teams do use them but that the refinement process for all-in-one metrics is at its limits. In short, for now these are as good as it gets and teams then overlay their own proprietary process for fit etc which is kind of how I would expect them to use this data.


As far as I'm aware, most team proprietary "all in one" metrics are substantively similar to the public metrics - there's not that much action happening on that end. A few teams have reached out to me asking for help building their own DARKO-like model, which likewise suggests to me the public metrics are basically fine as far the entire concept goes. I discussed this on a podcast the other day, but essentially we've hit the diminishing returns point with the various all-in-one metrics.

Teams are rather focused much more on issues of fit, and understanding which specific actions various players excel or struggle at (e.g. pick and roll, help defense, on-ball defense), and which specific contexts players perform well in/struggle in (e.g., in the bonus, with the other team in foul trouble, playing opposite a big/small defender, etc...). A lot of this work depends on non-public raw tracking data, and the goal essentially is to find a guy who grades as subpar in his current role who can excel in another context. The distinction we usually draw when discussing these things is that RAPM, RAPTOR, DPM, etc... are all "impact" metrics. They're trying to understand (or project) how impactful a player will be in his current role/context. They're not really "talent" metrics, since ascertaining that requires pretty detailed understanding of context (talent + context = impact).

To tie this back to DeRozan, I would be surprised if there were teams with sophisticated stats departments which had DeRozan as a top 40-50 player coming into this year on any impact metric (the Bulls do not have a particularly sophisticated stats department, but their group likewise didn't like DeRozan). On the other hand, it's very possible that a bunch of teams thought DeRozan had a lot of talent and thought they could use him differently to maximize his skillset, or hide his flaws (on the defensive end in particular).
 

Cesar Crespo

79
SoSH Member
Dec 22, 2002
21,588
I do not believe @bowiac said that teams are "moving away" from all-in-one metrics but the full quote is below. My interpretation is that the teams do use them but that the refinement process for all-in-one metrics is at its limits. In short, for now these are as good as it gets and teams then overlay their own proprietary process for fit etc which is kind of how I would expect them to use this data.
https://hoopshype.com/lists/advanced-stats-nba-real-plus-minus-rapm-win-shares-analytics/

“I don’t really use any,” said one executive, who is the president of basketball operations for a team in the Eastern Conference. “They are all pretty bad.”

Others were less critical but felt that while all-in-one composite metrics are constantly getting better, the future of analytics is headed away from these measurements altogether.

“If I could add a wrinkle to your story, it would be that all-in-one stats are overused – that the next phase of basketball analytics is all about context-dependent numbers,” said another front office member from the Western Conference. “That would be the most honest quote I could give.”
Didn't say anything about Bowiac.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
39,756
I don't agree that the article is as definitive on this topic as some here might believe. From the piece:

This executive feels that analytics will move away from ridge-regression-based stats and instead attempt to answer questions about forecast future performance based on roles the player had for their team (e.g. BBall-Index.com and Backpicks.com have metrics on lineup spacing, playmaking value and defensive versatility).

The executive also added that “averaging across a few numbers you trust” (which is something Owen Phillips suggested in this newsletter) is “probably the way to go” to get the best evaluation.

However, the most common feedback to the survey we received was that most teams focus on their own custom-developed systems when evaluating players. But those measurements aren’t available to the public or to the media and can’t be readily cited. Ultimately, the goal of this project is to provide the most updated feedback on the evaluation tools that you can actually use.
An anonymous quote in a hoopshype piece about analytics where teams have little incentive to disclose their process seems thin but YRMV.
 

RorschachsMask

Member
SoSH Member
Aug 23, 2011
3,364
Lynn
But that’s just two quotes, when a lot more of them voted that they trust the stats. Yeah teams are starting to go more towards predicative stuff, but so are these websites that create the advanced/impact stats.

SURVEY SAYS: Among the 29 individuals who participated in our survey, four said that LEBRON was their preferred catch-all metric. Fourteen others said that they trust LEBRON as an all-in-one metric while two said that they did not.
SURVEY SAYS: Among the 29 individuals who participated in our survey, six said that EPM was their preferred catch-all metric. That was the second-most among all metrics. Eleven others said that they trust EPM as an all-in-one metric while only one said that they did not.
SURVEY SAYS: Among the 29 individuals who participated in our survey, eight (8) said that DPM was their preferred catch-all metric. That was the most among all metrics. Ten (10) others said that they trust DPM as an all-in-one metric while only one (1) said that they did not.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
39,756
But that’s just two quotes, when a lot more of them voted that they trust the stats. Yeah teams are starting to go more towards predicative stuff, but so are these websites that create the advanced/impact stats.
We are effectively relitigating this article six months after it came out. People can use the stats or not use them but the trend is clearly towards the use of advances analytics, including all-in-one measures as *part* of the roster building process. Its inescapable even if it threatens the role of gateway experts like sports columnists and "superfans".
 

RorschachsMask

Member
SoSH Member
Aug 23, 2011
3,364
Lynn
We are effectively relitigating this article six months after it came out. People can use the stats or not use them but the trend is clearly towards the use of advances analytics, including all-in-one measures as *part* of the roster building process. Its inescapable even if it threatens the role of gateway experts like sports columnists and "superfans".
So everybody has different ways to use advanced stats, including some refusing to use them at all. I look at 4 or 5 of them, and if they are mostly showing the same thing, I think it’s for a reason. Though I do tend to ignore RPM, and the bball reference ones.

I’ve said it here before, but I think advanced stats help bridge the gap between raw numbers, and the eye test. People tend to look at raw stats, and completely ignore context. This was big on Twitter when Tatum versus Jaylen was constantly being debated.
 

PedroKsBambino

Well-Known Member
Lifetime Member
SoSH Member
Apr 17, 2003
28,404
I do not believe @bowiac said that teams are "moving away" from all-in-one metrics but the full quote is below. My interpretation is that the teams do use them but that the refinement process for all-in-one metrics is at its limits. In short, for now these are as good as it gets and teams then overlay their own proprietary process for fit etc which is kind of how I would expect them to use this data.
This is pretty consistent with the pre-DARKO discussion around here of metrics. I recall discussing, in particular for defense, that you needed to be pretty aggressive in looking at role, fit, and rotations to really understand what someone can and will do for a team.

Put differently, it's great that we have the metrics and they are valuable in the overall discussion. And, teams have and continue to see the need for things beyond the metrics (at least, the all in one metrics)....because assessment of players true talent (in basketball) needs to be aligned to their fit with a given team/system too. DeRozan is a good example, I suspect.
 
Last edited:

Jimbodandy

Member
SoSH Member
Jan 31, 2006
8,599
around the way
Of course teams are focused on different processes and data for predictive reasons...they're trying to predict the future and forecast as they plan their roster.

It's nonsense to state that metrics that analyze current and past play are worthless because people who are trying to predict the future don't use them.
 

RorschachsMask

Member
SoSH Member
Aug 23, 2011
3,364
Lynn
Of course teams are focused on different processes and data for predictive reasons...they're trying to predict the future and forecast as they plan their roster.

It's nonsense to state that metrics that analyze current and past play are worthless because people who are trying to predict the future don't use them.
I just think it’s cutting off your nose to spite your face. Advanced stats have a lot of value, you just need to know that’s it’s not some outright ranking system.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
39,756
This is pretty consistent with the pre-DARKO discussion around here of metrics, and that in particular for defense you needed to be pretty aggressive in looking at role, fit, and rotations to really understand what someone can and will do for a team.

Put differently, it's great that we have the metrics and they are valuable in the overall discussion. And, teams have and continue to see the need for things beyond the metrics (at least, the all in one metrics)....because assessment of players true talent (in basketball) needs to be aligned to their fit with a given team/system.
Agreed. The idea behind all-in-one stats is a denomination one in my way of thinking. It provides a common basis for looking at how much an individual has contributed to winning (for lack of a better way of putting it) but it doesn't tell you *how* that production is achieved. That is where teams employ their own metrics as well as overlay their organizational philosophy etc in figuring out the fit. An all-in-one metric of X does not equal X in terms of skills. Just in terms of output.

Edit: needed to add tense so those as not to trigger those who are worried about predictive/non-predictive distinctions. Not referring to predictive stats at all for the purpose of this discussion.
 

Cesar Crespo

79
SoSH Member
Dec 22, 2002
21,588
I just think it’s cutting off your nose to spite your face. Advanced stats have a lot of value, you just need to know that’s it’s not some outright ranking system.
Advanced stats are more than just the all in ones though. I don't get the point in all in one's at all, really. They tell you less information.

And if it's not a "ranking system" or "comparison" system, I get the point even less.
 

RorschachsMask

Member
SoSH Member
Aug 23, 2011
3,364
Lynn
Advanced stats are more than just the all in ones though. I don't get the point in all in one's at all, really. They tell you less information.

And if it's not a "ranking system" or "comparison" system, I get the point even less.
I’m just asking out of curiosity, no snark intended. What do you like to use when evaluating players?

I think all in ones have value, and clearly a lot of nba people think the same. I think the mistake people make with them is they use it exclusively to rank players. Just like with raw stats, context needs to be used.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
39,756
There are no advanced stats police here. People can use them or ignore them but stating that "NBA teams don't, they are moving away from them." is clearly incorrect.
 

Cesar Crespo

79
SoSH Member
Dec 22, 2002
21,588
I’m just asking out of curiosity, no snark intended. What do you like to use when evaluating players?

I think all in ones have value, and clearly a lot of nba people think the same. I think the mistake people make with them is they use it exclusively to rank players. Just like with raw stats, context needs to be used.
Eye test, advanced rate stats that measure individual things rather than try to measure overall value, traditional box score stats.

I like advanced stats. I just don't like ones that decide to combine things like shooting and defense. I'm going to love context dependent numbers.

I just don't get what they offer in the grand scheme of things. I guess if you just use it as one of like 10 pieces of information it's fine, but on their own? Meh. There's no "fit."
 

RorschachsMask

Member
SoSH Member
Aug 23, 2011
3,364
Lynn
Eye test, advanced rate stats that measure individual things rather than try to measure overall value, traditional box score stats.

I like advanced stats. I just don't like ones that decide to combine things like shooting and defense. I'm going to love context dependent numbers.

I just don't get what they offer in the grand scheme of things. I guess if you just use it as one of like 10 pieces of information it's fine, but on their own? Meh. There's no "fit."
They all have offensive and defensive numbers though, or at least most of them do.

52DB041B-D5DD-41D7-98D0-B2C23D27C7E3.png16548D05-9C0A-4C9A-9C31-DF7215B93D5A.png
 

Cesar Crespo

79
SoSH Member
Dec 22, 2002
21,588
They all have offensive and defensive numbers though, or at least most of them do.
Yeah, I meant even more specific than offense/defense. I always look at sports from a team building perspective though. I think fit, role and skill set is so important in basketball that I'm not sure what an overall number really tells us. It works better in baseball.

Some of the all in ones break it down quite a bit and offer a lot of info. I'm not sure the need to put one number on it, but meh.
 

benhogan

Granite Truther
SoSH Member
Nov 2, 2007
15,843
Santa Monica
Yeah, I meant even more specific than offense/defense. I always look at sports from a team building perspective though. I think fit, role and skill set is so important in basketball that I'm not sure what an overall number really tells us. It works better in baseball.

Some of the all in ones break it down quite a bit and offer a lot of info. I'm not sure the need to put one number on it, but meh.
100% agree with the bolded
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,943
New York, NY
Teams absolutely use all-in-one metrics. They're not the end of the story or anything, but I'd be surprised if there's any analytics friendly org that doesn't maintain their own version of RAPM. I was just at Sloan last week, and had conversations with a half dozen teams about how DPM works, etc... I believe said executive may not, but his analytics team likely does.

Generally, expressing skepticism in these metrics is something of a shibboleth in analytics circles. It's a way of signaling you "get it" and value the importance of context, role, etc... And that's right - one should be skeptical of the completeness of these metrics. The equation is something like Talent + Context = Impact, and the all-in-one metrics are mostly just measuring Impact, while a lot of conversations people want to have are about "Talent". That's fine - talent is important and probably more interesting, but both are parts of the picture.
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,943
New York, NY
For those of us less versed in basketball stats, can you give a brief overview of why DARKO is your choice?
Cause you can bat signal me here I'll answer any questions you have!

The quick summary is that DARKO/DPM is the only one of the major public metrics mentioned in that Hoops Hype which is designed to be purely predictive in nature. So to give an example, at the start of last year, EPM had Mike Conley as like the 4th best player in basketball, while DPM probably had him 20th or something. But even Taylor wouldn't say he thought Mike Conley was the 4th best player in basketball going forward. He was just giving Conley credit for stellar play to start the season. EPM is thus trying to balance predictive and descriptive measures (sort of like xFIP or something), and is really useful for handing out the MVP that way.

DARKO/DPM is mostly useless for an MVP award meanwhile - the results it shows are purely predictive, so it's more akin to a daily-updating version of Steamer/PECOTA.
 

nighthob

Member
SoSH Member
Jul 15, 2005
11,606
I just think it’s cutting off your nose to spite your face. Advanced stats have a lot of value, you just need to know that’s it’s not some outright ranking system.
The most advanced stats are the ones derived from the Synergy data. They tell us how a player performs situationally which is a more valuable data set than the all-in-ones can provide. The AIO ratings might give us a rough and dirty estimate of a player’s impact, but it’s a lot less valuable to know that than to know how players defend the pick & roll, their offensive efficiency in a variety of plays, etc.. As NBA front offices all have the Synergy data they increasingly rely on it in evaluating players.
 

RorschachsMask

Member
SoSH Member
Aug 23, 2011
3,364
Lynn
The most advanced stats are the ones derived from the Synergy data. They tell us how a player performs situationally which is a more valuable data set than the all-in-ones can provide. The AIO ratings might give us a rough and dirty estimate of a player’s impact, but it’s a lot less valuable to know that than to know how players defend the pick & roll, their offensive efficiency in a variety of plays, etc.. As NBA front offices all have the Synergy data they increasingly rely on it in evaluating players.
Yeah I love the synergy/tracking stats, I just try soaking up as much as I possibly can. I spend far too much time doing so lol, but I really enjoy the digging.
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,943
New York, NY
I agree with the above. The other part of this is just that as the AIO metrics have become more and more similar, there's less value to front offices from them, since there's just less competitive advantage associated with using them.

It's like moving from ERA to pitch-grading information. It's not that ERA isn't useful, but there's no edge to it left. The smart front offices are putting their efforts into pitch-grading efforts since that's where they can differentiate.
 

PedroKsBambino

Well-Known Member
Lifetime Member
SoSH Member
Apr 17, 2003
28,404
Appreciate bowiac chiming in and sharing insights from the front lines on this one!
 

HomeRunBaker

bet squelcher
SoSH Member
Jan 15, 2004
25,521
Interesting that Rudy Gobert is so high, given how little he does on offense.
Gobert is actually crucial to what Utah does offensively. He occupies the big on the baseline to dissuade weak side help and is lethal in PNR as the screener. Even if he isn’t receiving the pass his presence rolling to the rim forces the defense to react which creates space for Mitchell, Clarkson, Conley, etc off the dribble.