Those stats reflect ball-movement, lack of. Our eyes tell us that Luka is one of the best players in the league. Those stats tell us how the other guys are doing with Luka and without Luka. Of course the whole game is about putting the ball in the basket and how best to accomplish that. Are the guys getting stagnant waiting for their turn to handle the ball, or are they more involved if they get to touch the ball, themselves, every once in a while? Luka is fun to watch. Is Luka fun to play basketball with?
With Luka... (21 games, 12-9 record)
Off Rating: 108.4
Def Rating: 109.4
Net Rating: -1.0
IN those 21 games with Luka playing....
Luka ON-COURT: -5.5
Luka OFF-COURT: +3.7
Completely without Luka... (14 games, 5-9 record, and a more difficult strength of schedule and a more depleted roster overall)
Off Rating: 110.8
Def Rating: 107.5
Net Rating: +3.3
So even though the Mavs are 12-9 when Luka has played, they have been outscored -5.5 pts per 100 possessions when he has been on the court. They have outscored the opponent +3.7 when he hasn't been on the court in those games. THEN in the games he hasn't played at all, they have still outscored the opponents by +3.3 and had some real close losses (the biggest margin in those 9 losses was 8 points). Whereas with Luka seven of the nine losses have been by 10 or more points.
Don't know if anyone has read Seth Partnow's new book, The Mid-Range Theory. He is a stats guy and formerly worked in the Bucks front office. The book is an explanation of basketball metrics and how they are used by NBA teams.
He spends a fair amount of time discussing the various +/- stats, and how front offices use them. A few points relevant to the discussion --
1. Raw +/- figures are far too noisy to provide a basis for any meaningful analysis, even based on an entire season of data. Even with multiple seasons of data. Teams usually have proprietary figures of their own, with numerous adjustments to try to filter out some of the noise. Even with numerous adjustments, it is a noisy stat, and is never used by itself for player evaluation purposes.
2. On-off +/- can be a useful starting point, but it's very easily over-interpreted, and what you can glean from it is wholly dependent on context. For example, one of the team's starters might spend a stint or two every game anchoring an otherwise weak lineup. In such a case, the team's scoring margin could very easily be higher when the starter is off the court, due to the weak lineup dragging the starter's +/- down. Jumping to the conclusion that the team plays better without the starter, based on the on-off stat, would not be justified. Similarly, being on or off the floor in blowouts tends to affect cumulative +/- disproportionately, and it is not unusual for garbage time lineups to look much better or worse than the scoring margin of the team as a whole.
3. Even adjusted +/- stats must be considered with the known limitation that they rank zero in "explainability." That is, they tell you something about what happened on the court, but tell you absolutely nothing about why it happened. You have to do a contextual dive to learn anything at all from these stats. For example, which teammates were on the floor at the time being measured, which opponents were on the floor at the time being measured, what were the circumstances in the game at that point (clutch, garbage time, etc.) and much more.
4. In evaluating the usefulness of stats for various purposes, one must be conscious that they should pass the smell test. If particular stats lead you to a conclusion that seems very counterintuitive, maybe you've discovered a surprising nugget, or maybe your stat isn't really telling you what you think it is.
5. By far the most important stat is shown on the scoreboard, and NBA statisticians are aware that achieving wins is their North Star, not achieving some other statistical outcome.
__________________________
It's an interesting book, for people interested in that type of thing, as it provides an inside look.
Anyway, I think it points in the direction of a common-sense analysis. Basing an eyebrow-raising conclusion like "the Mavs are a better team without Luka" on a very oddball partial season of on-off stats = GIGO.
Of course, if team officials studied the context and determined what really happened on the court and why, then that would be a horse of a different color. Presumably, that's what they're doing, and afaik, it hasn't led them to keep Luka on the pine as much as possible in the interests of the team playing better. That's all we can know about their thinking at this point.
Except that I'm sure they have their eye on the most important stat. With Luka, the team is winning 57% of their games. Without Luka, they're losing 64%.
(01-01-2022, 10:40 AM)mavsluvr Wrote: [ -> ]Don't know if anyone has read Seth Partnow's new book, The Mid-Range Theory. He is a stats guy and formerly worked in the Bucks front office. The book is an explanation of basketball metrics and how they are used by NBA teams.
He spends a fair amount of time discussing the various +/- stats, and how front offices use them. A few points relevant to the discussion --
1. Raw +/- figures are far too noisy to provide a basis for any meaningful analysis, even based on an entire season of data. Even with multiple seasons of data. Teams usually have proprietary figures of their own, with numerous adjustments to try to filter out some of the noise. Even with numerous adjustments, it is a noisy stat, and is never used by itself for player evaluation purposes.
2. On-off +/- can be a useful starting point, but it's very easily over-interpreted, and what you can glean from it is wholly dependent on context. For example, one of the team's starters might spend a stint or two every game anchoring an otherwise weak lineup. In such a case, the team's scoring margin could very easily be higher when the starter is off the court, due to the weak lineup dragging the starter's +/- down. Jumping to the conclusion that the team plays better without the starter, based on the on-off stat, would not be justified. Similarly, being on or off the floor in blowouts tends to affect cumulative +/- disproportionately, and it is not unusual for garbage time lineups to look much better or worse than the scoring margin of the team as a whole.
3. Adjusted +/- stats must be considered with the known limitation that they rank zero in "explainability." That is, they tell you something about what happened on the court, but tell you absolutely nothing about why it happened. You have to do a contextual dive to learn anything at all from these stats. For example, which teammates were on the floor at the time being measured, which opponents were on the floor at the time being measured, what were the circumstances in the game at that point (clutch, garbage time, etc.) and much more.
4. In evaluating the usefulness of stats for various purposes, one must be conscious that they should pass the smell test. If particular stats lead you to a conclusion that seems very counterintuitive, maybe you've discovered a surprising nugget, or maybe your stat isn't really telling you what you think it is.
5. By far the most important stat is shown on the scoreboard, and NBA statisticians are aware that achieving wins is their North Star, not achieving some other statistical outcome.
__________________________
It's an interesting book, for people interested in that type of thing, as it provides an inside look.
Anyway, I think it points in the direction of a common-sense analysis.
Once called the eye-test.
(01-01-2022, 10:40 AM)mavsluvr Wrote: [ -> ]By far the most important stat is shown on the scoreboard
Yep 100%!!!!
Been preaching this and beating this drum for years and years. Outscoring the opponent as a TEAM is the only thing that matters at the end of the day. Mavs have been better at doing that as of late (+3.3 in the last 10 games) and as the statistics show, this will eventually translate into more wins. Eyeball and stats both say that the Mavs have been playing their best basketball as of late even with the crazy COVID and roster situations.
I think line-up data, 5-man, or even 3-man or pairing data is useful. Seems like the closer you get to 5-man, the more noise you remove. But even with 5-man, there is noise, because you don't know, just by the numbers, who that lineup has been playing against.
But individual +/- is impacted by so many things, and the number doesn't give you the context.
Regarding Luka, he hasn't been LUKA, but he's still been good compared to an average NBA player. You can point to things that Luka has done that could negatively impact a lineup (not getting back on defense because he'd rather chat with the ref), AND you can point to lineup issues that would impact Luka's +/- number. He has spent lots of minutes in the dreaded 2-big starting lineups that have tanked. And Luka has spent a lot of time in 4th quarter lineups that have tanked. I'd say Luka not being LUKA has contributed a little to the tanking of the 1st and 4th quarter lineups, but the lineups themselves have been a huge problem because of non-Luka factors.
(01-01-2022, 10:40 AM)mavsluvr Wrote: [ -> ]Except that I'm sure they have their eye on the most important stat. With Luka, the team is winning 57% of their games. Without Luka, they're losing 64%.
With Luka the average of opponents winning percentages (discounting games against the Mavs) is 46.9%.
Without Luka the average of opponents winning percentages (discounting games against the Mavs) is 57.3%.
(01-01-2022, 12:23 PM)BackToSquareOne Wrote: [ -> ]With Luka the average of opponents winning percentages (discounting games against the Mavs) is 46.9%.
Without Luka the average of opponents winning percentages (discounting games against the Mavs) is 57.3%.
Well, if you want to go in that direction --
With Luka, the Mavs are winning 57% of their games (more than the 53% that opponent winning percentage would predict).
Without Luka, the Mavs are winning 36% of their games (less than the 43% that the opponent winning percentage would predict).
My point isn't that people can't cobble together some stats to try to make it look like Luka isn't contributing much, or might even be dragging this band of merry men down. It is that even the stats being offered don't actually show that, the whole premise doesn't pass the smell test, and the most important thing is that the Mavs are a winning team with Luka and a losing team without him.
Haters gonna hate. (Not accusing you in particular, BTSO).
(01-01-2022, 12:23 PM)BackToSquareOne Wrote: [ -> ]With Luka the average of opponents winning percentages (discounting games against the Mavs) is 46.9%.
Without Luka the average of opponents winning percentages (discounting games against the Mavs) is 57.3%.
And how many of those high winning percentages teams were ravaged by Covid when the Mavs played them? More missing context. It just seems pointless to spend much time looking at extremely noisy stats like raw individual +/- when you could be looking at a quality advanced stat that removes so much of the context issues. They are not perfect either, and require a good sized sample and looking at context not captured (like coaching, player fit, playing with injury, etc).
(01-01-2022, 12:59 PM)mavsluvr Wrote: [ -> ]Well, if you want to go in that direction --
With Luka, the Mavs are winning 57% of their games (more than the 53% that opponent winning percentage would predict).
Without Luka, the Mavs are winning 36% of their games (less than the 43% that the opponent winning percentage would predict).
My point isn't that people can't cobble together some stats to try to make it look like Luka isn't contributing much, or might even be dragging this band of merry men down. It is that even the stats being offered don't actually show that, the whole premise doesn't pass the smell test, and the most important thing is that the Mavs are a winning team with Luka and a losing team without him.
Haters gonna hate. (Not accusing you in particular, BTSO).
I admire you even bother. Brunson net on/off in last 10 games Luka didn't play is minus 5.
(01-01-2022, 12:59 PM)mavsluvr Wrote: [ -> ]Well, if you want to go in that direction --
With Luka, the Mavs are winning 57% of their games (more than the 53% that opponent winning percentage would predict).
Without Luka, the Mavs are winning 36% of their games (less than the 43% that the opponent winning percentage would predict).
My point isn't that people can't cobble together some stats to try to make it look like Luka isn't contributing much, or might even be dragging this band of merry men down. It is that even the stats being offered don't actually show that, the whole premise doesn't pass the smell test, and the most important thing is that the Mavs are a winning team with Luka and a losing team without him.
Haters gonna hate. (Not accusing you in particular, BTSO).
I'll defend the driver of this particular conversation, even though I don't see +/- the same way he does. Kam isn't a "hater", he's been a lover of +/- for the decade I've known him. He's very consistant. He likes to use +/- and combine it with his eye test to make observations. I can't see any "hating" going on.
The NBA PIE stat (which is one of the best predictive stats in regard to winning) shows very similar things as the +/- numbers I have been sharing.
In the last 10 games (without Luka) the Mavs are 4th in the NBA with a PIE of 54.4 (over 50 means you are playing like a winning team). For reference GSW are 56.4 for the year, which is best in the NBA.
In the 25 games before that (in which Luka played in 21 of them) the Mavs were 24th in the NBA with a PIE of 47.7 (for reference DET is worst in the NBA with 43.4).
Point being, the Mavs are playing basketball much better in the last 10 games than they did in the first 25 and are currently performing like one of the better teams in the league. This is even despite being 5-5 in the last 10 games. You are indeed what your record says, but oftentimes teams playing like a losing team will win some games, and other times teams playing like a winning team will lose some games. The Mavs are playing like a winning team over the last ten games, 4th best in the NBA according to PIE.
(01-01-2022, 01:26 PM)omahen Wrote: [ -> ]I admire you even bother.
Bother is what we do around here. Why "bother" about any of it? I'm so glad we're not usually too dismissive of each other.
We have folks not liking what they've seen from Luka so far, mainly because Luka has spoiled them with his otherworldly greatness for three years.
We have folks that don't want to hear anything negative about Luka rejecting any talk of Luka having a down year.
Interestingly, these polar stances have resutled in some really good conversation, imo. And all the folks in the mushy middle have benefited.
(01-01-2022, 01:31 PM)Kammrath Wrote: [ -> ]The NBA PIE stat (which is one of the best predictive stats in regard to winning) shows very similar things as the +/- numbers I have been sharing.
In the last 10 games (without Luka) the Mavs are 4th in the NBA with a PIE of 54.4 (over 50 means you are playing like a winning team). For reference GSW are 56.4 for the year, which is best in the NBA.
In the 25 games before that (in which Luka played in 21 of them) the Mavs were 24th in the NBA with a PIE of 47.7 (for reference DET is worst in the NBA with 43.4).
Point being, the Mavs are playing basketball much better in the last 10 games than they did in the first 25 and are currently performing like one of the better teams in the league. This is even despite being 5-5 in the last 10 games. You are indeed what your record says, but oftentimes teams playing like a losing team will win some games, and other times teams playing like a winning team will lose some games. The Mavs are playing like a winning team over the last ten games, 4th best in the NBA according to PIE.
They have played a bunch of Covid ravished teams and the kings twice in the last 10 games. Do you really think that is a good sample to make any kind of conclusions from?
(01-01-2022, 01:39 PM)mvossman Wrote: [ -> ]They have played a bunch of Covid ravished teams and the kings twice in the last 10 games. Do you really think that is a good sample to make any kind of conclusions from?
Yes I do. ALL sample sizes are "small" in a very real sense (shoot, even 82 games isn't enough!) and so we simply take the sample sizes we have available to us and look at them carefully with wisdom and discernment.
During this stretch the Mavs are EQUALLY (if not moreso) ravaged by COVID and the Mavs are also missing their far and away single most talented basketball player. And yet despite that both my eyes and the stats say this is the best stretch of basketball they have played this season, not in the least being much more watchable and entertaining.
(01-01-2022, 12:17 PM)fifteenth Wrote: [ -> ]I think line-up data, 5-man, or even 3-man or pairing data is useful. Seems like the closer you get to 5-man, the more noise you remove. But even with 5-man, there is noise, because you don't know, just by the numbers, who that lineup has been playing against.
But individual +/- is impacted by so many things, and the number doesn't give you the context.
Regarding Luka, he hasn't been LUKA, but he's still been good compared to an average NBA player. You can point to things that Luka has done that could negatively impact a lineup (not getting back on defense because he'd rather chat with the ref), AND you can point to lineup issues that would impact Luka's +/- number. He has spent lots of minutes in the dreaded 2-big starting lineups that have tanked. And Luka has spent a lot of time in 4th quarter lineups that have tanked. I'd say Luka not being LUKA has contributed a little to the tanking of the 1st and 4th quarter lineups, but the lineups themselves have a been a huge problem because of non-Luka factors.
Partnow discusses the use of +/- to evaluate
lineups (as opposed to an individual player) at some length. This
is an area where the stat (as adjusted according to the team's algorithm) has proved useful. However, it still much be approached with knowledge of its limitations. As you point out,
you still need contextual information to evaluate it.
And an even bigger issue with it is sample size. These stats tend to start providing a meaningful basis for analysis, he says, at around 500 possessions (250 minutes or so), and very few lineups reach anything close to that much playing time together. (In the season he references, about .16% of all 5-man lineups played). When you go down to smaller combos, more of them reach the minutes threshold, but there is more external context to be accounted for.
Coaches and organizations can appear to have widely different views on +/-, but that may be more superficial than anything. He cites a couple of examples. One is the Mavericks, who are openly believers (while acknowledging the limitations). Carlisle relied on this type of analysis in choosing to start JJ Barea in the back half of the Miami series; they had lineup data indicating that the team performed particularly well when JJ lineups countered Miami lineups containing Mike Bibby. Pop, on the other hand, is dismissive of the nightly +/- stat, saying he normally doesn't even look at it, and if he does, he gives it little weight. That doesn't mean that the Spurs don't make excellent use of this type of analytics -- they do, and Pop is just correctly noting that the single-game individual plus-minus stats reported in the box score tell you the aggregate scoring margin in that game when a particular player was on the floor, but are 100% useless as far as predicting what the scoring margin will be when that player takes the floor again.
He thinks that these stats, as adjusted, can in some situations do a pretty fair job of categorizing lineup/players into good, medium, and bad tiers. However, they have next to no value in predicting outcomes with specificity. For example, if a lineup has a +5 in 250 minutes played, that would tend to indicate that a deeper look might suggest that it is generally a good lineup. However, it provides no reason to think that the next time the lineup takes the floor, it will have a +5, or even that its average play in the future will be +5.
(01-01-2022, 01:20 PM)mvossman Wrote: [ -> ]And how many of those high winning percentages teams were ravaged by Covid when the Mavs played them? More missing context. It just seems pointless to spend much time looking at extremely noisy stats like raw individual +/- when you could be looking at a quality advanced stat that removes so much of the context issues. They are not perfect either, and require a good sized sample and looking at context not captured (like coaching, player fit, playing with injury, etc).
I agree.
Not that you asked, but I think the stuff I like to look at most is box score stats, in the context of the eye test and lineup data. I also like some of the cool little situational defensive measures floating around.
What do you like other than the more recent adjusted +/- descendants?
I've read a few of the papers on or definitions of some of the metrics that have followed APM over the years, and I never really come away satisfied. I know the goal is to smooth the rough edges off of, and add context to raw +/-, but it still applies regression in a way that seems to benefit some and hurt other players in a way that's not just based on actual context, but just based on unintended bias within the adjustments themselves. But I'm not a probability and statistics guy, so I may be off on my assessment.
(01-01-2022, 01:26 PM)omahen Wrote: [ -> ]I admire you even bother. Brunson net on/off in last 10 games Luka didn't play is minus 5.
Believe me, every time I venture into this breach, I ask myself why, lol. This time, I was fresh off reading Seth's book, and thought maybe sharing a few of his insights would elicit some interesting discussion.
(01-01-2022, 01:46 PM)Kammrath Wrote: [ -> ]ALL sample sizes are "small" in a very real sense (shoot, even 82 games isn't enough!)
Yes, but some are smaller than others and some are larger than others. Sample size is actually a think studying statistics.
(01-01-2022, 01:46 PM)Kammrath Wrote: [ -> ]Yes I do. ALL sample sizes are "small" in a very real sense (shoot, even 82 games isn't enough!) and so we simply take the sample sizes we have available to us and look at them carefully with wisdom and discernment.
During this stretch the Mavs are EQUALLY (if not moreso) ravaged by COVID and the Mavs are also missing their far and away single most talented basketball player. And yet despite that both my eyes and the stats say this is the best stretch of basketball they have played this season, not in the least being much more watchable and entertaining.
Yes, this offense would be more fun to watch if they were always going against the Kings, or the Nurkless Blazers or a bunch of Covid replacement players. Do you really think that's because there is no Luka?
What is your point with all of this anyways? Do you really think this team is better without Luka? Think its time to trade him for a bushel of picks? Where are you going with this?