Prime Bird in 2024

[DraymondGold]

...
"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!"

Great movie reference, but were you going for "apples?"

Crap! How am I gonna mess up a great quote like that?? Yeah apples haha

[tsherkin]

...
"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!"

Great movie reference, but were you going for "apples?"

Crap! How am I gonna mess up a great quote like that?? Yeah apples haha

Still, evoking one of Matt Damon's best lines? Bueno.

[SinceGatlingWasARookie]

He definitely wouldn’t be as good as Jokic. He did have better playoff numbers in ‘86 than SGA though, and I imagine that just being in a modern system he’d take a lot more threes so #2 sounds right.

Bird was a very good low post scorer and a very good broken play garbage man scorer.

Surround Bird with 3 point shooters and then set Bird free against a floor spaced out by 3 point shooters and Bird may go wild inside.

You had to stop Bird’s dribble which was not so hard because Bird did not have a good dribble. You put speed like Michael Cooper or Paul Pressey on Bird and they could stop Bird’s dribble one on one but if Bird posted them on one they were too small and Bird would score inside. Normally there was always a help defender to stop Bird from posting up the fast small defenders.

Bird would eat up a power forward inside one on one because power forwards were not quick enough to end Bird’s dribble like fast smaller Cooper and Pressey could.

1981 and 1984 Bird were not good 3 point shooters like 1986 Bird was.

I do not think modern Bird would shoot more 3s because modern NBA has other guys to shoot 3s but lacks guys that could do what Bird could do inside.

[theonlyclutch]

He definitely wouldn’t be as good as Jokic. He did have better playoff numbers in ‘86 than SGA though, and I imagine that just being in a modern system he’d take a lot more threes so #2 sounds right.

Bird was a very good low post scorer and a very good broken play garbage man scorer.

Surround Bird with 3 point shooters and then set Bird free against a floor spaced out by 3 point shooters and Bird may go wild inside.

You had to stop Bird’s dribble which was not so hard because Bird did not have a good dribble. You put speed like Michael Cooper or Paul Pressey on Bird and they could stop Bird’s dribble one on one but if Bird posted them on one they were too small and Bird would score inside. Normally there was always a help defender to stop Bird from posting up the fast small defenders.

Bird would eat up a power forward inside one on one because power forwards were not quick enough to end Bird’s dribble like fast smaller Cooper and Pressey could.

1981 and 1984 Bird were not good 3 point shooters like 1986 Bird was.

I do not think modern Bird would shoot more 3s because modern NBA has other guys to shoot 3s but lacks guys that could do what Bird could do inside.

Bird was able to slip free to score off so many broken plays largely because he played with a lot of offensive talent, Mchale/Parish being plenty dangerous with the ball in their hands prevented defenses from keeping their attentions focused on Bird at all times. Contrast now where teams are assigning people to faceguard Luka/SGA/Curry etc for entire plays because offensive talent is significantly more evenly distributed throughout the league, it's just a totally different opportunity set for stars nowadays.

[homecourtloss]

Spoiler:

Part 1: Argument Clarification

Heliocentric vs Motion Offenses
Hi all -- I want to push back slightly against two opinions raised here:
1. "the best offenses are helio offenses"
2. "the last 10 years have shown that teams understand exactly what they want to do on offense, and have the ball in the hands of a primary offense who is much better at it than others on his team, and instead of “ball movement,” they get to their spots through their favorite actions using fewer passes. Teams are making fewer and fewer passes over the last 10 years, but raw offenses are getting better and better."

Re: 2, saying that teams (in general, with no qualifications) all want to put the ball in the hands of their best offensive player more than before, just citing a reduction in passing numbers from 10 years ago, as the explanation for why raw offensive rating is higher now... seems to miss something so obvious it left my scratching my head.

Putting aside the misleading stuff (there are plenty of diverse play styles now, not all teams today are or want to be heliocentric, plenty of non-heliocentric teams today have also shown to have improved offensive results) ... this misses the 3 point revolution! It's called the 3 point revolution, not the heliocentric revolution, for a reason. Comparing 2014 vs 2024:
-the average regular season passes per game has gone from 295.1 to 281.3, a reduction of -4.6%.
-the average regular season 3PA per game has gone from 21.5 to 35.1, an increase of 33.4%.
Are we really attributing the improvement in modern raw offensive rating primarily to a 4.6% reduction in passing, as opposed to a 33.4% increase in three point shooting??
It also misses that motion offenses does not necessarily require more passing: there may just be more "motion". In a motion offense, it's perfectly valid to have a passer set up (e.g. at the top of the key, nail, or high post) while players move off-ball, cut to the basket, set off-ball screens, use the gravity of an off-ball offensive threat to create an opening, etc., then make one pass to the most open player who then shoots. You can have a heliocentric player who's a high-volume passer, and a motion player who prioritizes off-ball movement more than passing, and the team results could produce a similar volume of passing.

It may be true that motion offenses tend to have more passing, but a tiny bit less passing today =/= less motion today. And indeed, we have also seen a rise in 3 point shooters who move with or without the ball in the modern era. This is not inconsistent with motion principles.

In short: The rise of the 3 point shooting drove the boom in modern offenses, very obviously much more than a slight decrease in passing or the rise in some subset of teams playing more heliocentric.

I'll come back later and post more, but right off the bat I want to mention a few things.

Re: 2, saying that teams (in general, with no qualifications) all want to put the ball in the hands of their best offensive player more than before, just citing a reduction in passing numbers from 10 years ago, as the explanation for why raw offensive rating is higher now... seems to miss something so obvious it left my scratching my head.

This is a strawman. I didn't say that the reduced number of passes is what directly led to better offenses, but rather it is part of the general pattern of what teams want to do and therefore a byproduct, i.e., get to their spots with the players they want taking the shots that they want which by its very nature requires fewer passes.

Perhaps I'm misinterpreting your first post, but it sounds like you're arguing
1. that "putting ball in the hands of a primary offensive" player has been the primary driver of "raw offenses are getting better and better", "instead of ball movement"
(I.e., heliocentrism has been the driver of the recent offensive explosion, and that it produces better offensive results than motion offenses)
2. and that there's evidence for point 1 in the form of 2a) "inverse relationship between passes per 100 possessions and ORtg", perhaps along with 2b) "the examples [therealbig3] mention about the very best offenses of all time”.

I’m saying this all misses the point. I’m saying the increase in offensive rating was dominated by other factors that your first post doesn’t mention. I’m saying teams can get “the players they want taking the shots that they want”, but that does not “by… very nature require fewer passes”. Teams want to maximize offensive efficiency on offense (as well as possessions), which entails efficient shots (always preferably open; preferably at the rim or at the 3 point line, but midrange can be acceptable if it’s open or diversifies the offense to make it harder to defend or from a player who’s comfortable there). I’m saying teams can generate those efficient shots with a variety of on-ball, off-ball, passing, screening, and spacing forms of creation. Passing, ball movement, and off-ball movement are all inarguably methods of generating an open or efficient shot, which is what teams want.

Specifically, I’m arguing:
1. that the increase in volume and accuracy in 3 point shooting is ~~ by far ~~ the clear driver of the recent offensive explosion.
2. that any changes in passing (or heliocentrism more generally) over the decade has had a significantly weaker relation to why offenses are improving.
-> 1 and 2 are statistically justified, and shown below

3. that any relatively weak negative correlation between passing and offensive rating is not sufficient evidence to prove that heliocentrism is better than ball movement in general, and that these relatively weak correlations across all teams are not applicable to teams at the top level or to top-end players like the ones discussed here.
(i.e. that your 2a and 2b are wrong)
-> 3 is statistically justified, and shown below

4. that, in fact, if there is any trend in the noise among all-time team play styles, it suggests that movement offenses produce as good if not better results than heliocentrism. (but that analysis of hyper small-sample playoff-only all-time team peak play styles can be fraught with uncertainty/error/bias)
-> 4 is statistically supported across basically any survey of top teams that goes deeper than a sample of just one or two teams, as per my previous post

5. and that, regardless, passing is not an ideal differentiator for motion offenses vs heliocentric offenses, as players in a motion offense can move off ball without necessarily increasing passing, not all heliocentric offenses pass the least, and there's a variety of other play styles besides motion and heliocentrism and factors more generally that could muddy this trend.
(i.e. that 2a is more wrong).
-> 5 is readily obvious from any film study of teams with different play styles.

Putting aside the misleading stuff (there are plenty of diverse play styles now, not all teams today are or want to be heliocentric, plenty of non-heliocentric teams today have also shown to have improved offensive results) ... this misses the 3 point revolution! It's called the 3 point revolution, not the heliocentric revolution, for a reason. Comparing 2014 vs 2024:
-the average regular season passes per game has gone from 295.1 to 281.3, a reduction of -4.6%.
-the average regular season 3PA per game has gone from 21.5 to 35.1, an increase of 33.4%.
Are we really attributing the improvement in modern raw offensive rating primarily to a 4.6% reduction in passing, as opposed to a 33.4% increase in three point shooting??

1. Yes, non-helio teams and teams that make more passes have improved their offenses as well, but from 2014 to 2024, the years for which we have public data, more passes per 100 possessions tends to correlate with lower ORtgs. There are seasons from 2013-2014 to 2023-2024 in which there is a strongish inverse relationship between passes per 100 possessions and ORtg, i.e., more passes means lower ORtg, but NO SEASONS in which there is even more than a neutral relationship between passes per 100 possessions and ORtg. Not a single season. So, everyone is taking more threes, but the teams making fewer passes within this new environment of taking threes are also faring better.

2. The reduction in passing is more than what you're stating because you're not accounting for pace. Take a look at passes per 100 possessions. In 2013-2014, it was 314.5 passes per 100 and in 2023-2024, it was 285.6 If you look at 2014-2015 to 2022-2023, it went from 319.5 to 284.4. That's a change of 11%, not 4.6%. From 2014 to 2014 it would be a 9.2% change.

3. There's not any relationship between passes per 100 possessions (other than one, maybe another season) and three point attempt rate--the differences in ORtg are happening elsewhere.

Three point attempt rate and passes per 100 possessions

2014: slight inverse, i.e, more passes led to lower three point attempt rate
2015: zero
2016: zero
2017: zero
2018: strongest inverse relationship
2019: slight negative
2020: stronger inverse
2021: zero
2022: zero
2023: slight positive
2024: zero

So, for every season in which we have data, fewer passes tend to either lead to the same amount of three point attempts, or more three point attempts and generally higher ORtgs. Either way, it doesn't speak much to making more passes--either it leads to fewer three point attempts and a lower ORtg, same amount of three point attempts and a lower ORtg, greater three point attempt rate (one season) and a slightly great ORTg.

But more passes per 100 tend to lead to lower ORtgs.

ORtg and passes per 100 possessions

2014: slight inverse, i.e, more passes led to lower ORtg
2015: zero or slight inverse
2016: zero or slight inverse
2017: slight inverse
2018: strongest inverse relationship
2019: slight negative
2020: stronger inverse
2021: slight inverse, i.e, more passes led to lower ORtg
2022: slight inverse, i.e, more passes led to lower ORtg
2023: basically zero or slight positive, i.e., more passes led to a higher ORtg
2024: zero

Now... I haven't even gotten to average seconds per touch...

A brief quibble, before getting into the data: Why is it necessary to normalize for pace when looking at total passing?

When pace increases, teams play more possessions in a game, which require possessions are ending more quickly. If possessions are ending more quickly, that requires that some amount of points, assists (which sometimes come with points), rebounds, blocks, steals, and/or turnovers — all the things that might end a possession — increase. But passing is not a thing that ends a possession. If a team passes every 10 seconds they have the ball on average, and they have the ball for the same amount of time per game (i.e. they still have the ball ~50% of the time), then they will have the same number of total passes per game, regardless of their pace.

So it’s not clear, at least not a priori, that normalizing passes per 100 possessions is a better measure of ‘ball movement’ (and again ball movement is not the same as motion offense). If you plot passes vs pace, and draw a line of best fit, it looks like you’re holding a ruler up against a Jackson Pollack painting. Sure, you get out a negative slope, but it’s just great statistics.

Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses? That's not justified, as I'll show in a sec. But since it’s easy to look at both passes per game and passes per 100, I'll look at both.

Part 2: Statistical Evidence for the influence of 3 Point Shooting vs Passing on Offensive Ratings
1. 3 point shooting is what’s driving improved modern ORtg, and decreased passing is a poor model in comparison.
You claim that “putting ball in the hands of a primary offensive" player is the major driver of recent improvements offensive rating (i.e. that heliocentrism > movement offenses), and claim that a) the “strongish inverse relationship between passes per 100 possessions and ORtg” shows this, along with b) the best offenses ever are more heliocentric.

I’ve already disproved b in my last post. I’ve also claimed that a. is missing the point, that improved 3 point shooting is a significantly better driver, that the trend of passing vs ORtg is not very strong in comparison, and that it is insufficient to argue heliocentrism > movement offenses for good teams.

Let’s look at the R^2 value and p value for passes made per game vs ORtg, passes made per 100 vs ORtg, 3PA per game vs Ortg, and 3PA per 100 vs ORtg, across all available seasons (2014–2024 regular season)

What are these terms? They’re very standard measures in statistics.
Model: In this case, the model is the idea that ORtg would be proportional to either passes per game, passes per 100, 3PA per game, 3PA per 100.
R^2: Measures how much your model explains variation in the data, and thus can tell you how well your model fits the data. Range: 0 to 1. Best case: 1 is best, a minimum threshold of 0.5 is often set for an acceptable R^2 value, although in noisier cases (e.g. studies of human behavior) a minimum good threshold can be set to 0.4 or even lower.
p-value: Measures the probability that your model is better than the null hypothesis (i.e. that the trend doesn’t exist), and thus measures whether your model is statistically significant. Best case: 0 is best, and a maximum threshold of 0.05 is often set as an acceptable p-value.

How big is the best-fit slope?
-passes made per game vs ORtg: slope = -0.08
-passes made per 100 vs ORtg: -0.08
-3PA per game vs Ortg: 0.45
-3PA per 100 vs ORtg: 0.48

So the best-fit dependence suggests ORtg improves 600% more on average by taking one more 3PA per 100 than by taking one less pass per 100. The dependence on 3PA is much larger.


How do the models compare, from worst to best:
-passes made per game vs ORtg: R^2 = 0.13, p-value = 1.5 * 10^-11
-passes made per 100 vs ORtg: R^2 = 0.2, p-value = 8,4 * 10^-18
-3PA per game vs Ortg: R^2 = 0.40, p-value = 1.7 * 10^-38
-3PA per 100 vs ORtg: R^2 = 0.41, p-value = 7.0 * 10^-39

So increased 3 point shooting attempts fits the improved ORtg data over 300% better than the decrease in passing, and over 200% better than increased passing per 100. And 3 point shooting attempts have a better p-value by over 20 orders of magnitude!

Saying that decreases in passing improved ORtg is statistically significant (it’s better than doing nothing), but it is below any normal threshold for how well a good model should explain the variation in the data.

As I predicted, increased 3 point shooting volume is — by a country mile — a better model for explaining the improved ORtg in recent seasons than decreased passing. This doesn’t even include improved 3P%, which also has a significantly better R^2 (0.35) and p-value (1.2 * 10^-31), which would make overall 3 point shooting an even better model for explaining the improved ORtg in recent seasons.

In comparison, focusing on the decrease in passing to explain the ORtg improvements just misses the point. It is furthermore not a great proxy for movement offenses (as above, it misses off-ball movement). And therefore does not act as sufficient evidence to justify that heliocentrism > movement offenses, or that “putting ball in the hands of a primary offensive" player is the major driver of recent improvements offensive rating instead of just teams shooting 3s more (alongside all the supplemental spacing benefits of doing so).

2. There is no clear trend that decreasing passing improves offenses on good teams.
Let’s perform the same procedure as above, but filtering just for good teams.

If you look at passes per 100 vs ORtg for teams with ORtg > {113 (i.e. the top 95 teams of the past 10 years), 114, 115, and 117}, the best R^2 value is 0.04 (bad) and the best p-value is 0.14 (bad).
If you look at passes per 100 vs relative ORtg for teams with rORtg > {+0 (i.e. the top 166 teams of the past 10 years), +1, +2, +3, +4, and +5}, the best R^2 value is 0.002 (absolute trash) and the best p-value is 0.52 (absolute trash).

In other words, there is no statistically significant argument that more passing leads to worse offenses for teams with ORtg > 113 (top 95 teams of the last decade) or even for relative rORtg > 0 (top 166 teams of the decade). So unless you think Bird or whatever motion star you care to think of isn’t capable of raising a team to even a league average offense (which would be silly), the idea that the passing data somehow suggests motion offenses produce worse top-end results than heliocentric offenses is bogus.

You might push back against this by saying it’s a matter of sample size. Maybe 95 teams or 166 teams aren’t a large enough sample size to pick out the trend, so maybe passing is still worse for top-end offenses. But if that’s the case, if 95 teams or 166 teams aren’t a large enough sample to pick out the trend, how strong could this trend really be?? If it’s such a weak trend, then it’s pretty clearly unreasonable to use this to justify saying heliocentrism > motion for above-average offenses.

This actually raises something interesting: if you believe the trend that more passing produces worse offenses is real, it is only justifiable when applied to bad / below-average offenses. Why could this be? Here’s a few possibilities, although I”m open to other ideas:
a) Causal: bad teams with bad passing hurt their offenses with turnovers, and these teams have reduced turnovers from bad passes over the past decade.
There’s a statistically significant correlation between turnovers per 100 and offensive rating (it’s actually more statistically significant than your claimed passing vs ORtg), and there’s a clear causal connection (if you turn the ball over, you don’t get a good shot up that possession, which reduces your ORtg).
Bad offensive teams with bad passers used to turn the ball over more, but have since reduced passes that might lead to turnovers, thus improving their ORtg (alongside improved 3 point shooting being a bigger factor in their improved ORtg). Additionally, turnovers may also have been reduced by the wider passing lanes caused by 3 point shooting.

b) Symptomatic: bad teams aren’t able to break down the defenses as effectively, and so they end up passing the ball more to restart new actions as more of their initial actions fail.
In this model, bad teams from a decade ago were impotent passers, playing hot potato hoping one of their teammates could generate something better without much success from anyone.
Since then, it’s become easier for teams to break down defenses (aided primarily by increased 3 point shooting), and so there’s less of a need for impotent hot-potato passing.

In either case, good teams either a) gain the benefits of passing to balance or even outweigh the downsides of turnovers, or b), show less symptomatic impotent passing as they’re better able to break down defenses (including with passing, but also with a variety of other methods), and thus good teams show no correlation between passes per 100 and worse ORtg.

tldr:
-teams are better offenses now because of 3 point shooting much more than decreased passing.
-passing does not produce worse offenses on above-average teams.
-there's not enough evidence to claim heliocentrism > motion, based on what's described here, and based on the surveys of top level teams mentioned in my previous posts.

...
"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!" EDIT: APPLES!!

Once again, you have written a wall of text addressing a strawman, and you have been vigorously attacking that strawman. Here is the original conversation:

Helio offenses are easier to defend, the ball and the threat are in the same spot and move in unison. An off ball master like Bird is harder to defend because the ball and the threat are moving separately and are in different spots.

Do you have proof of this? Because the best offenses are helio offenses. By a clear margin. Helios like LeBron and Nash led better playoff offenses than Curry did on the Warriors...not to mention that Curry is plenty helio himself, he just does more off-ball than the rest.

And the Warriors offense stalled pretty much whenever Curry was NOT on ball as much as he should have been, but when they stubbornly stuck to moving him around off ball. Bird doesn't really have a choice, he can't be on ball, he doesn't have the skills for it.

In addition to the examples you mention about the very best offenses of all time, the last 10 years have shown that teams understand exactly what they want to do on offense, and have the ball in the hands of a primary offense who is much better at it than others on his team, and instead of “ball movement,” they get to their spots through their favorite actions using fewer passes.

Teams are making fewer and fewer passes over the last 10 years, but raw offenses are getting better and better. People will say, “But there’s no defense…” so if that were the case, how come teams that make more passes, have more ball movement, still, in general, have worse offenses even within this worse defense era?

Where did I say anything about threes in my initial discussion? Teams knowing exactly what they want to INCLUDES taking threes. The first nested comment was about Helios being easier to defend, but we’ve seen Helios along with teams that make fewer passes and holding the ball longer do as well or better than teams that move the ball more and don’t hold on to the ball as long. YOU then brought up threes, and begin attacking a strawman.

A brief quibble, before getting into the data: Why is it necessary to normalize for pace when looking at total passing?

When pace increases, teams play more possessions in a game, which require possessions are ending more quickly. If possessions are ending more quickly, that requires that some amount of points, assists (which sometimes come with points), rebounds, blocks, steals, and/or turnovers — all the things that might end a possession — increase. But passing is not a thing that ends a possession. If a team passes every 10 seconds they have the ball on average, and they have the ball for the same amount of time per game (i.e. they still have the ball ~50% of the time), then they will have the same number of total passes per game, regardless of their pace.

So it’s not clear, at least not a priori, that normalizing passes per 100 possessions is a better measure of ‘ball movement’ (and again ball movement is not the same as motion offense). If you plot passes vs pace, and draw a line of best fit, it looks like you’re holding a ruler up against a Jackson Pollack painting. Sure, you get out a negative slope, but it’s just great statistics.

Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses? That's not justified, as I'll show in a sec. But since it’s easy to look at both passes per game and passes per 100, I'll look at both.

For someone who likes using numbers, it’s really strange to see you ask why we need to adjust for pace. This seems mostly likely because you forgot about the need to adjust for pace and now came back with this to justify you overlooking/forgetting/not knowing to do this. If we want to know if teams are making more or fewer passes we have to adjust for how many opportunities there are to make passes. The league plays faster today than it did in 2014, YET there are fewer passes on average. You can’t compare average passes a team makes per game in a 94 possession pace environment vs. how many they make in a 98 possession environment. In a faster league, you would think there would be more passes yet there are fewer. You then top it off with yet another strawman: “Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses?” I didn’t make that argument. You’re arguing with yourself.

The central point is that it’s just accepted by most posters and the poster whose comment I was replying to that “moving the ball” and “not ball stopping” are better for offenses but we just don’t have proof of that. If it were true, we’d see SOME relationship between moving the ball+not holding the ball on offenses, but we don’t see any at all. Now, that doesn’t mean causation nor did I ever say that it did, but we also have no numbers telling us moving the ball and not holding on to the ball help offenses. This is over a course of 10 seasons with hundreds of thousands of touches and passes.

Obviously ORtgs have gone up so a straight comparison of passes per 100 and ORtgs is going to look like this:

But if we adjust for relative passes per 100 (i.e., relative to the NBA season average) and relative ORtg we still see this:

Btw, people might ask what happens if you take out transition offense and look only at half court offense—more passes in the half court should help, right? This is more time intensive to show, but I looked at two seasons and there was zero relationship and even slightly inverse.

Now what about holding the ball? Without relative comparison:

And with relative time holding the ball and relative ORtgs:

Conclusion: making more passes, moving the ball and not holding on to the ball don’t seem to have any effect on offenses and if there is any, it’s a slightly negative one. This doesn’t mean that it’s a bad thing to “move the ball,” or that a team can’t have a good offense making more passes or not holding on to the ball, but the data we have (at least from the 2013-2014do not support the long held beliefs about making more passes and not holding on to the ball being inherently better for offenses.

"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!"

Corny and embarrassing

[falcolombardi]

Spoiler:

Part 1: Argument Clarification

I'll come back later and post more, but right off the bat I want to mention a few things.



This is a strawman. I didn't say that the reduced number of passes is what directly led to better offenses, but rather it is part of the general pattern of what teams want to do and therefore a byproduct, i.e., get to their spots with the players they want taking the shots that they want which by its very nature requires fewer passes.

Perhaps I'm misinterpreting your first post, but it sounds like you're arguing
1. that "putting ball in the hands of a primary offensive" player has been the primary driver of "raw offenses are getting better and better", "instead of ball movement"
(I.e., heliocentrism has been the driver of the recent offensive explosion, and that it produces better offensive results than motion offenses)
2. and that there's evidence for point 1 in the form of 2a) "inverse relationship between passes per 100 possessions and ORtg", perhaps along with 2b) "the examples [therealbig3] mention about the very best offenses of all time”.

I’m saying this all misses the point. I’m saying the increase in offensive rating was dominated by other factors that your first post doesn’t mention. I’m saying teams can get “the players they want taking the shots that they want”, but that does not “by… very nature require fewer passes”. Teams want to maximize offensive efficiency on offense (as well as possessions), which entails efficient shots (always preferably open; preferably at the rim or at the 3 point line, but midrange can be acceptable if it’s open or diversifies the offense to make it harder to defend or from a player who’s comfortable there). I’m saying teams can generate those efficient shots with a variety of on-ball, off-ball, passing, screening, and spacing forms of creation. Passing, ball movement, and off-ball movement are all inarguably methods of generating an open or efficient shot, which is what teams want.

Specifically, I’m arguing:
1. that the increase in volume and accuracy in 3 point shooting is ~~ by far ~~ the clear driver of the recent offensive explosion.
2. that any changes in passing (or heliocentrism more generally) over the decade has had a significantly weaker relation to why offenses are improving.
-> 1 and 2 are statistically justified, and shown below

3. that any relatively weak negative correlation between passing and offensive rating is not sufficient evidence to prove that heliocentrism is better than ball movement in general, and that these relatively weak correlations across all teams are not applicable to teams at the top level or to top-end players like the ones discussed here.
(i.e. that your 2a and 2b are wrong)
-> 3 is statistically justified, and shown below

4. that, in fact, if there is any trend in the noise among all-time team play styles, it suggests that movement offenses produce as good if not better results than heliocentrism. (but that analysis of hyper small-sample playoff-only all-time team peak play styles can be fraught with uncertainty/error/bias)
-> 4 is statistically supported across basically any survey of top teams that goes deeper than a sample of just one or two teams, as per my previous post

5. and that, regardless, passing is not an ideal differentiator for motion offenses vs heliocentric offenses, as players in a motion offense can move off ball without necessarily increasing passing, not all heliocentric offenses pass the least, and there's a variety of other play styles besides motion and heliocentrism and factors more generally that could muddy this trend.
(i.e. that 2a is more wrong).
-> 5 is readily obvious from any film study of teams with different play styles.

1. Yes, non-helio teams and teams that make more passes have improved their offenses as well, but from 2014 to 2024, the years for which we have public data, more passes per 100 possessions tends to correlate with lower ORtgs. There are seasons from 2013-2014 to 2023-2024 in which there is a strongish inverse relationship between passes per 100 possessions and ORtg, i.e., more passes means lower ORtg, but NO SEASONS in which there is even more than a neutral relationship between passes per 100 possessions and ORtg. Not a single season. So, everyone is taking more threes, but the teams making fewer passes within this new environment of taking threes are also faring better.

2. The reduction in passing is more than what you're stating because you're not accounting for pace. Take a look at passes per 100 possessions. In 2013-2014, it was 314.5 passes per 100 and in 2023-2024, it was 285.6 If you look at 2014-2015 to 2022-2023, it went from 319.5 to 284.4. That's a change of 11%, not 4.6%. From 2014 to 2014 it would be a 9.2% change.

3. There's not any relationship between passes per 100 possessions (other than one, maybe another season) and three point attempt rate--the differences in ORtg are happening elsewhere.

Three point attempt rate and passes per 100 possessions

2014: slight inverse, i.e, more passes led to lower three point attempt rate
2015: zero
2016: zero
2017: zero
2018: strongest inverse relationship
2019: slight negative
2020: stronger inverse
2021: zero
2022: zero
2023: slight positive
2024: zero

So, for every season in which we have data, fewer passes tend to either lead to the same amount of three point attempts, or more three point attempts and generally higher ORtgs. Either way, it doesn't speak much to making more passes--either it leads to fewer three point attempts and a lower ORtg, same amount of three point attempts and a lower ORtg, greater three point attempt rate (one season) and a slightly great ORTg.

But more passes per 100 tend to lead to lower ORtgs.

ORtg and passes per 100 possessions

2014: slight inverse, i.e, more passes led to lower ORtg
2015: zero or slight inverse
2016: zero or slight inverse
2017: slight inverse
2018: strongest inverse relationship
2019: slight negative
2020: stronger inverse
2021: slight inverse, i.e, more passes led to lower ORtg
2022: slight inverse, i.e, more passes led to lower ORtg
2023: basically zero or slight positive, i.e., more passes led to a higher ORtg
2024: zero

Now... I haven't even gotten to average seconds per touch...

A brief quibble, before getting into the data: Why is it necessary to normalize for pace when looking at total passing?

When pace increases, teams play more possessions in a game, which require possessions are ending more quickly. If possessions are ending more quickly, that requires that some amount of points, assists (which sometimes come with points), rebounds, blocks, steals, and/or turnovers — all the things that might end a possession — increase. But passing is not a thing that ends a possession. If a team passes every 10 seconds they have the ball on average, and they have the ball for the same amount of time per game (i.e. they still have the ball ~50% of the time), then they will have the same number of total passes per game, regardless of their pace.

So it’s not clear, at least not a priori, that normalizing passes per 100 possessions is a better measure of ‘ball movement’ (and again ball movement is not the same as motion offense). If you plot passes vs pace, and draw a line of best fit, it looks like you’re holding a ruler up against a Jackson Pollack painting. Sure, you get out a negative slope, but it’s just great statistics.

Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses? That's not justified, as I'll show in a sec. But since it’s easy to look at both passes per game and passes per 100, I'll look at both.

Part 2: Statistical Evidence for the influence of 3 Point Shooting vs Passing on Offensive Ratings
1. 3 point shooting is what’s driving improved modern ORtg, and decreased passing is a poor model in comparison.
You claim that “putting ball in the hands of a primary offensive" player is the major driver of recent improvements offensive rating (i.e. that heliocentrism > movement offenses), and claim that a) the “strongish inverse relationship between passes per 100 possessions and ORtg” shows this, along with b) the best offenses ever are more heliocentric.

I’ve already disproved b in my last post. I’ve also claimed that a. is missing the point, that improved 3 point shooting is a significantly better driver, that the trend of passing vs ORtg is not very strong in comparison, and that it is insufficient to argue heliocentrism > movement offenses for good teams.

Let’s look at the R^2 value and p value for passes made per game vs ORtg, passes made per 100 vs ORtg, 3PA per game vs Ortg, and 3PA per 100 vs ORtg, across all available seasons (2014–2024 regular season)

What are these terms? They’re very standard measures in statistics.
Model: In this case, the model is the idea that ORtg would be proportional to either passes per game, passes per 100, 3PA per game, 3PA per 100.
R^2: Measures how much your model explains variation in the data, and thus can tell you how well your model fits the data. Range: 0 to 1. Best case: 1 is best, a minimum threshold of 0.5 is often set for an acceptable R^2 value, although in noisier cases (e.g. studies of human behavior) a minimum good threshold can be set to 0.4 or even lower.
p-value: Measures the probability that your model is better than the null hypothesis (i.e. that the trend doesn’t exist), and thus measures whether your model is statistically significant. Best case: 0 is best, and a maximum threshold of 0.05 is often set as an acceptable p-value.

How big is the best-fit slope?
-passes made per game vs ORtg: slope = -0.08
-passes made per 100 vs ORtg: -0.08
-3PA per game vs Ortg: 0.45
-3PA per 100 vs ORtg: 0.48

So the best-fit dependence suggests ORtg improves 600% more on average by taking one more 3PA per 100 than by taking one less pass per 100. The dependence on 3PA is much larger.


How do the models compare, from worst to best:
-passes made per game vs ORtg: R^2 = 0.13, p-value = 1.5 * 10^-11
-passes made per 100 vs ORtg: R^2 = 0.2, p-value = 8,4 * 10^-18
-3PA per game vs Ortg: R^2 = 0.40, p-value = 1.7 * 10^-38
-3PA per 100 vs ORtg: R^2 = 0.41, p-value = 7.0 * 10^-39

So increased 3 point shooting attempts fits the improved ORtg data over 300% better than the decrease in passing, and over 200% better than increased passing per 100. And 3 point shooting attempts have a better p-value by over 20 orders of magnitude!

Saying that decreases in passing improved ORtg is statistically significant (it’s better than doing nothing), but it is below any normal threshold for how well a good model should explain the variation in the data.

As I predicted, increased 3 point shooting volume is — by a country mile — a better model for explaining the improved ORtg in recent seasons than decreased passing. This doesn’t even include improved 3P%, which also has a significantly better R^2 (0.35) and p-value (1.2 * 10^-31), which would make overall 3 point shooting an even better model for explaining the improved ORtg in recent seasons.

In comparison, focusing on the decrease in passing to explain the ORtg improvements just misses the point. It is furthermore not a great proxy for movement offenses (as above, it misses off-ball movement). And therefore does not act as sufficient evidence to justify that heliocentrism > movement offenses, or that “putting ball in the hands of a primary offensive" player is the major driver of recent improvements offensive rating instead of just teams shooting 3s more (alongside all the supplemental spacing benefits of doing so).

2. There is no clear trend that decreasing passing improves offenses on good teams.
Let’s perform the same procedure as above, but filtering just for good teams.

If you look at passes per 100 vs ORtg for teams with ORtg > {113 (i.e. the top 95 teams of the past 10 years), 114, 115, and 117}, the best R^2 value is 0.04 (bad) and the best p-value is 0.14 (bad).
If you look at passes per 100 vs relative ORtg for teams with rORtg > {+0 (i.e. the top 166 teams of the past 10 years), +1, +2, +3, +4, and +5}, the best R^2 value is 0.002 (absolute trash) and the best p-value is 0.52 (absolute trash).

In other words, there is no statistically significant argument that more passing leads to worse offenses for teams with ORtg > 113 (top 95 teams of the last decade) or even for relative rORtg > 0 (top 166 teams of the decade). So unless you think Bird or whatever motion star you care to think of isn’t capable of raising a team to even a league average offense (which would be silly), the idea that the passing data somehow suggests motion offenses produce worse top-end results than heliocentric offenses is bogus.

You might push back against this by saying it’s a matter of sample size. Maybe 95 teams or 166 teams aren’t a large enough sample size to pick out the trend, so maybe passing is still worse for top-end offenses. But if that’s the case, if 95 teams or 166 teams aren’t a large enough sample to pick out the trend, how strong could this trend really be?? If it’s such a weak trend, then it’s pretty clearly unreasonable to use this to justify saying heliocentrism > motion for above-average offenses.

This actually raises something interesting: if you believe the trend that more passing produces worse offenses is real, it is only justifiable when applied to bad / below-average offenses. Why could this be? Here’s a few possibilities, although I”m open to other ideas:
a) Causal: bad teams with bad passing hurt their offenses with turnovers, and these teams have reduced turnovers from bad passes over the past decade.
There’s a statistically significant correlation between turnovers per 100 and offensive rating (it’s actually more statistically significant than your claimed passing vs ORtg), and there’s a clear causal connection (if you turn the ball over, you don’t get a good shot up that possession, which reduces your ORtg).
Bad offensive teams with bad passers used to turn the ball over more, but have since reduced passes that might lead to turnovers, thus improving their ORtg (alongside improved 3 point shooting being a bigger factor in their improved ORtg). Additionally, turnovers may also have been reduced by the wider passing lanes caused by 3 point shooting.

b) Symptomatic: bad teams aren’t able to break down the defenses as effectively, and so they end up passing the ball more to restart new actions as more of their initial actions fail.
In this model, bad teams from a decade ago were impotent passers, playing hot potato hoping one of their teammates could generate something better without much success from anyone.
Since then, it’s become easier for teams to break down defenses (aided primarily by increased 3 point shooting), and so there’s less of a need for impotent hot-potato passing.

In either case, good teams either a) gain the benefits of passing to balance or even outweigh the downsides of turnovers, or b), show less symptomatic impotent passing as they’re better able to break down defenses (including with passing, but also with a variety of other methods), and thus good teams show no correlation between passes per 100 and worse ORtg.

tldr:
-teams are better offenses now because of 3 point shooting much more than decreased passing.
-passing does not produce worse offenses on above-average teams.
-there's not enough evidence to claim heliocentrism > motion, based on what's described here, and based on the surveys of top level teams mentioned in my previous posts.

...
"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!" EDIT: APPLES!!

Once again, you have written a wall of text addressing a strawman, and you have been vigorously attacking that strawman. Here is the original conversation:

Do you have proof of this? Because the best offenses are helio offenses. By a clear margin. Helios like LeBron and Nash led better playoff offenses than Curry did on the Warriors...not to mention that Curry is plenty helio himself, he just does more off-ball than the rest.

And the Warriors offense stalled pretty much whenever Curry was NOT on ball as much as he should have been, but when they stubbornly stuck to moving him around off ball. Bird doesn't really have a choice, he can't be on ball, he doesn't have the skills for it.

In addition to the examples you mention about the very best offenses of all time, the last 10 years have shown that teams understand exactly what they want to do on offense, and have the ball in the hands of a primary offense who is much better at it than others on his team, and instead of “ball movement,” they get to their spots through their favorite actions using fewer passes.

Teams are making fewer and fewer passes over the last 10 years, but raw offenses are getting better and better. People will say, “But there’s no defense…” so if that were the case, how come teams that make more passes, have more ball movement, still, in general, have worse offenses even within this worse defense era?

Where did I say anything about threes in my initial discussion? Teams knowing exactly what they want to INCLUDES taking threes. The first nested comment was about Helios being easier to defend, but we’ve seen Helios along with teams that make fewer passes and holding the ball longer do as well or better than teams that move the ball more and don’t hold on to the ball as long. YOU then brought up threes, and begin attacking a strawman.

A brief quibble, before getting into the data: Why is it necessary to normalize for pace when looking at total passing?

When pace increases, teams play more possessions in a game, which require possessions are ending more quickly. If possessions are ending more quickly, that requires that some amount of points, assists (which sometimes come with points), rebounds, blocks, steals, and/or turnovers — all the things that might end a possession — increase. But passing is not a thing that ends a possession. If a team passes every 10 seconds they have the ball on average, and they have the ball for the same amount of time per game (i.e. they still have the ball ~50% of the time), then they will have the same number of total passes per game, regardless of their pace.

So it’s not clear, at least not a priori, that normalizing passes per 100 possessions is a better measure of ‘ball movement’ (and again ball movement is not the same as motion offense). If you plot passes vs pace, and draw a line of best fit, it looks like you’re holding a ruler up against a Jackson Pollack painting. Sure, you get out a negative slope, but it’s just great statistics.

Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses? That's not justified, as I'll show in a sec. But since it’s easy to look at both passes per game and passes per 100, I'll look at both.

For someone who likes using numbers, it’s really strange to see you ask why we need to adjust for pace. This seems mostly likely because you forgot about the need to adjust for pace and now came back with this to justify you overlooking/forgetting/not knowing to do this. If we want to know if teams are making more or fewer passes we have to adjust for how many opportunities there are to make passes. The league plays faster today than it did in 2014, YET there are fewer passes on average. You can’t compare average passes a team makes per game in a 94 possession pace environment vs. how many they make in a 98 possession environment. In a faster league, you would think there would be more passes yet there are fewer. You then top it off with yet another strawman: “Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses?” I didn’t make that argument. You’re arguing with yourself.

The central point is that it’s just accepted by most posters and the poster whose comment I was replying to that “moving the ball” and “not ball stopping” are better for offenses but we just don’t have proof of that. If it were true, we’d see SOME relationship between moving the ball+not holding the ball on offenses, but we don’t see any at all. Now, that doesn’t mean causation nor did I ever say that it did, but we also have no numbers telling us moving the ball and not holding on to the ball help offenses. This is over a course of 10 seasons with hundreds of thousands of touches and passes.

Obviously ORtgs have gone up so a straight comparison of passes per 100 and ORtgs is going to look like this:

But if we adjust for relative passes per 100 (i.e., relative to the NBA season average) and relative ORtg we still see this:

Btw, people might ask what happens if you take out transition offense and look only at half court offense—more passes in the half court should help, right? This is more intense to show, but I looked at two seasons and there was zero relationship and even slightly inverse.

Now what about holding the ball? Without relative comparison:

And with relative time holding the ball and relative ORtgs:

Conclusion: making more passes, moving the ball and not holding on to the ball don’t seem to have any effect on offenses and if there is any, it’s a slightly negative one. This doesn’t mean that it’s a bad thing to “move the ball,” or that a team can’t have a good offense making more passes or not holding on to the ball, but the data we have (at least from the 2013-2014do not support the long held beliefs about making more passes and not holding on to the ball being inherently better for offenses.
a7

"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!"

Corny and embarrassing

A lot of the port/off ball arguments used against on-ball players essentially have always came down to hipotheticals since by the numbers they have never held up to amy scrutiny

The best example i cam think of this is how on-ball players having the best offensive dominance stretches ever with their teams amd usually more impressive than their theorically more off ball (and supposedly more portable and higher ceiling) peers is systematically ignored

I have seen ben taylor hold ball dominance against magic in all time discussions as if he and nash didnt lead the most dominant team oheffenses run ever.

At some point port is just a justification to rank your favorite players higher based on a theory that superficially makes sense but doesnt even scratch basketball beyond the surface

[homecourtloss]

Spoiler:

Part 1: Argument Clarification

Perhaps I'm misinterpreting your first post, but it sounds like you're arguing
1. that "putting ball in the hands of a primary offensive" player has been the primary driver of "raw offenses are getting better and better", "instead of ball movement"
(I.e., heliocentrism has been the driver of the recent offensive explosion, and that it produces better offensive results than motion offenses)
2. and that there's evidence for point 1 in the form of 2a) "inverse relationship between passes per 100 possessions and ORtg", perhaps along with 2b) "the examples [therealbig3] mention about the very best offenses of all time”.

I’m saying this all misses the point. I’m saying the increase in offensive rating was dominated by other factors that your first post doesn’t mention. I’m saying teams can get “the players they want taking the shots that they want”, but that does not “by… very nature require fewer passes”. Teams want to maximize offensive efficiency on offense (as well as possessions), which entails efficient shots (always preferably open; preferably at the rim or at the 3 point line, but midrange can be acceptable if it’s open or diversifies the offense to make it harder to defend or from a player who’s comfortable there). I’m saying teams can generate those efficient shots with a variety of on-ball, off-ball, passing, screening, and spacing forms of creation. Passing, ball movement, and off-ball movement are all inarguably methods of generating an open or efficient shot, which is what teams want.

Specifically, I’m arguing:
1. that the increase in volume and accuracy in 3 point shooting is ~~ by far ~~ the clear driver of the recent offensive explosion.
2. that any changes in passing (or heliocentrism more generally) over the decade has had a significantly weaker relation to why offenses are improving.
-> 1 and 2 are statistically justified, and shown below

3. that any relatively weak negative correlation between passing and offensive rating is not sufficient evidence to prove that heliocentrism is better than ball movement in general, and that these relatively weak correlations across all teams are not applicable to teams at the top level or to top-end players like the ones discussed here.
(i.e. that your 2a and 2b are wrong)
-> 3 is statistically justified, and shown below

4. that, in fact, if there is any trend in the noise among all-time team play styles, it suggests that movement offenses produce as good if not better results than heliocentrism. (but that analysis of hyper small-sample playoff-only all-time team peak play styles can be fraught with uncertainty/error/bias)
-> 4 is statistically supported across basically any survey of top teams that goes deeper than a sample of just one or two teams, as per my previous post

5. and that, regardless, passing is not an ideal differentiator for motion offenses vs heliocentric offenses, as players in a motion offense can move off ball without necessarily increasing passing, not all heliocentric offenses pass the least, and there's a variety of other play styles besides motion and heliocentrism and factors more generally that could muddy this trend.
(i.e. that 2a is more wrong).
-> 5 is readily obvious from any film study of teams with different play styles.

A brief quibble, before getting into the data: Why is it necessary to normalize for pace when looking at total passing?

When pace increases, teams play more possessions in a game, which require possessions are ending more quickly. If possessions are ending more quickly, that requires that some amount of points, assists (which sometimes come with points), rebounds, blocks, steals, and/or turnovers — all the things that might end a possession — increase. But passing is not a thing that ends a possession. If a team passes every 10 seconds they have the ball on average, and they have the ball for the same amount of time per game (i.e. they still have the ball ~50% of the time), then they will have the same number of total passes per game, regardless of their pace.

So it’s not clear, at least not a priori, that normalizing passes per 100 possessions is a better measure of ‘ball movement’ (and again ball movement is not the same as motion offense). If you plot passes vs pace, and draw a line of best fit, it looks like you’re holding a ruler up against a Jackson Pollack painting. Sure, you get out a negative slope, but it’s just great statistics.

Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses? That's not justified, as I'll show in a sec. But since it’s easy to look at both passes per game and passes per 100, I'll look at both.

Part 2: Statistical Evidence for the influence of 3 Point Shooting vs Passing on Offensive Ratings
1. 3 point shooting is what’s driving improved modern ORtg, and decreased passing is a poor model in comparison.
You claim that “putting ball in the hands of a primary offensive" player is the major driver of recent improvements offensive rating (i.e. that heliocentrism > movement offenses), and claim that a) the “strongish inverse relationship between passes per 100 possessions and ORtg” shows this, along with b) the best offenses ever are more heliocentric.

I’ve already disproved b in my last post. I’ve also claimed that a. is missing the point, that improved 3 point shooting is a significantly better driver, that the trend of passing vs ORtg is not very strong in comparison, and that it is insufficient to argue heliocentrism > movement offenses for good teams.

Let’s look at the R^2 value and p value for passes made per game vs ORtg, passes made per 100 vs ORtg, 3PA per game vs Ortg, and 3PA per 100 vs ORtg, across all available seasons (2014–2024 regular season)

What are these terms? They’re very standard measures in statistics.
Model: In this case, the model is the idea that ORtg would be proportional to either passes per game, passes per 100, 3PA per game, 3PA per 100.
R^2: Measures how much your model explains variation in the data, and thus can tell you how well your model fits the data. Range: 0 to 1. Best case: 1 is best, a minimum threshold of 0.5 is often set for an acceptable R^2 value, although in noisier cases (e.g. studies of human behavior) a minimum good threshold can be set to 0.4 or even lower.
p-value: Measures the probability that your model is better than the null hypothesis (i.e. that the trend doesn’t exist), and thus measures whether your model is statistically significant. Best case: 0 is best, and a maximum threshold of 0.05 is often set as an acceptable p-value.

How big is the best-fit slope?
-passes made per game vs ORtg: slope = -0.08
-passes made per 100 vs ORtg: -0.08
-3PA per game vs Ortg: 0.45
-3PA per 100 vs ORtg: 0.48

So the best-fit dependence suggests ORtg improves 600% more on average by taking one more 3PA per 100 than by taking one less pass per 100. The dependence on 3PA is much larger.


How do the models compare, from worst to best:
-passes made per game vs ORtg: R^2 = 0.13, p-value = 1.5 * 10^-11
-passes made per 100 vs ORtg: R^2 = 0.2, p-value = 8,4 * 10^-18
-3PA per game vs Ortg: R^2 = 0.40, p-value = 1.7 * 10^-38
-3PA per 100 vs ORtg: R^2 = 0.41, p-value = 7.0 * 10^-39

So increased 3 point shooting attempts fits the improved ORtg data over 300% better than the decrease in passing, and over 200% better than increased passing per 100. And 3 point shooting attempts have a better p-value by over 20 orders of magnitude!

Saying that decreases in passing improved ORtg is statistically significant (it’s better than doing nothing), but it is below any normal threshold for how well a good model should explain the variation in the data.

As I predicted, increased 3 point shooting volume is — by a country mile — a better model for explaining the improved ORtg in recent seasons than decreased passing. This doesn’t even include improved 3P%, which also has a significantly better R^2 (0.35) and p-value (1.2 * 10^-31), which would make overall 3 point shooting an even better model for explaining the improved ORtg in recent seasons.

In comparison, focusing on the decrease in passing to explain the ORtg improvements just misses the point. It is furthermore not a great proxy for movement offenses (as above, it misses off-ball movement). And therefore does not act as sufficient evidence to justify that heliocentrism > movement offenses, or that “putting ball in the hands of a primary offensive" player is the major driver of recent improvements offensive rating instead of just teams shooting 3s more (alongside all the supplemental spacing benefits of doing so).

2. There is no clear trend that decreasing passing improves offenses on good teams.
Let’s perform the same procedure as above, but filtering just for good teams.

If you look at passes per 100 vs ORtg for teams with ORtg > {113 (i.e. the top 95 teams of the past 10 years), 114, 115, and 117}, the best R^2 value is 0.04 (bad) and the best p-value is 0.14 (bad).
If you look at passes per 100 vs relative ORtg for teams with rORtg > {+0 (i.e. the top 166 teams of the past 10 years), +1, +2, +3, +4, and +5}, the best R^2 value is 0.002 (absolute trash) and the best p-value is 0.52 (absolute trash).

In other words, there is no statistically significant argument that more passing leads to worse offenses for teams with ORtg > 113 (top 95 teams of the last decade) or even for relative rORtg > 0 (top 166 teams of the decade). So unless you think Bird or whatever motion star you care to think of isn’t capable of raising a team to even a league average offense (which would be silly), the idea that the passing data somehow suggests motion offenses produce worse top-end results than heliocentric offenses is bogus.

You might push back against this by saying it’s a matter of sample size. Maybe 95 teams or 166 teams aren’t a large enough sample size to pick out the trend, so maybe passing is still worse for top-end offenses. But if that’s the case, if 95 teams or 166 teams aren’t a large enough sample to pick out the trend, how strong could this trend really be?? If it’s such a weak trend, then it’s pretty clearly unreasonable to use this to justify saying heliocentrism > motion for above-average offenses.

This actually raises something interesting: if you believe the trend that more passing produces worse offenses is real, it is only justifiable when applied to bad / below-average offenses. Why could this be? Here’s a few possibilities, although I”m open to other ideas:
a) Causal: bad teams with bad passing hurt their offenses with turnovers, and these teams have reduced turnovers from bad passes over the past decade.
There’s a statistically significant correlation between turnovers per 100 and offensive rating (it’s actually more statistically significant than your claimed passing vs ORtg), and there’s a clear causal connection (if you turn the ball over, you don’t get a good shot up that possession, which reduces your ORtg).
Bad offensive teams with bad passers used to turn the ball over more, but have since reduced passes that might lead to turnovers, thus improving their ORtg (alongside improved 3 point shooting being a bigger factor in their improved ORtg). Additionally, turnovers may also have been reduced by the wider passing lanes caused by 3 point shooting.

b) Symptomatic: bad teams aren’t able to break down the defenses as effectively, and so they end up passing the ball more to restart new actions as more of their initial actions fail.
In this model, bad teams from a decade ago were impotent passers, playing hot potato hoping one of their teammates could generate something better without much success from anyone.
Since then, it’s become easier for teams to break down defenses (aided primarily by increased 3 point shooting), and so there’s less of a need for impotent hot-potato passing.

In either case, good teams either a) gain the benefits of passing to balance or even outweigh the downsides of turnovers, or b), show less symptomatic impotent passing as they’re better able to break down defenses (including with passing, but also with a variety of other methods), and thus good teams show no correlation between passes per 100 and worse ORtg.

tldr:
-teams are better offenses now because of 3 point shooting much more than decreased passing.
-passing does not produce worse offenses on above-average teams.
-there's not enough evidence to claim heliocentrism > motion, based on what's described here, and based on the surveys of top level teams mentioned in my previous posts.

...
"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!" EDIT: APPLES!!

Once again, you have written a wall of text addressing a strawman, and you have been vigorously attacking that strawman. Here is the original conversation:

In addition to the examples you mention about the very best offenses of all time, the last 10 years have shown that teams understand exactly what they want to do on offense, and have the ball in the hands of a primary offense who is much better at it than others on his team, and instead of “ball movement,” they get to their spots through their favorite actions using fewer passes.

Teams are making fewer and fewer passes over the last 10 years, but raw offenses are getting better and better. People will say, “But there’s no defense…” so if that were the case, how come teams that make more passes, have more ball movement, still, in general, have worse offenses even within this worse defense era?

Where did I say anything about threes in my initial discussion? Teams knowing exactly what they want to INCLUDES taking threes. The first nested comment was about Helios being easier to defend, but we’ve seen Helios along with teams that make fewer passes and holding the ball longer do as well or better than teams that move the ball more and don’t hold on to the ball as long. YOU then brought up threes, and begin attacking a strawman.

A brief quibble, before getting into the data: Why is it necessary to normalize for pace when looking at total passing?

When pace increases, teams play more possessions in a game, which require possessions are ending more quickly. If possessions are ending more quickly, that requires that some amount of points, assists (which sometimes come with points), rebounds, blocks, steals, and/or turnovers — all the things that might end a possession — increase. But passing is not a thing that ends a possession. If a team passes every 10 seconds they have the ball on average, and they have the ball for the same amount of time per game (i.e. they still have the ball ~50% of the time), then they will have the same number of total passes per game, regardless of their pace.

So it’s not clear, at least not a priori, that normalizing passes per 100 possessions is a better measure of ‘ball movement’ (and again ball movement is not the same as motion offense). If you plot passes vs pace, and draw a line of best fit, it looks like you’re holding a ruler up against a Jackson Pollack painting. Sure, you get out a negative slope, but it’s just great statistics.

Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses? That's not justified, as I'll show in a sec. But since it’s easy to look at both passes per game and passes per 100, I'll look at both.

For someone who likes using numbers, it’s really strange to see you ask why we need to adjust for pace. This seems mostly likely because you forgot about the need to adjust for pace and now came back with this to justify you overlooking/forgetting/not knowing to do this. If we want to know if teams are making more or fewer passes we have to adjust for how many opportunities there are to make passes. The league plays faster today than it did in 2014, YET there are fewer passes on average. You can’t compare average passes a team makes per game in a 94 possession pace environment vs. how many they make in a 98 possession environment. In a faster league, you would think there would be more passes yet there are fewer. You then top it off with yet another strawman: “Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses?” I didn’t make that argument. You’re arguing with yourself.

The central point is that it’s just accepted by most posters and the poster whose comment I was replying to that “moving the ball” and “not ball stopping” are better for offenses but we just don’t have proof of that. If it were true, we’d see SOME relationship between moving the ball+not holding the ball on offenses, but we don’t see any at all. Now, that doesn’t mean causation nor did I ever say that it did, but we also have no numbers telling us moving the ball and not holding on to the ball help offenses. This is over a course of 10 seasons with hundreds of thousands of touches and passes.

Obviously ORtgs have gone up so a straight comparison of passes per 100 and ORtgs is going to look like this:

But if we adjust for relative passes per 100 (i.e., relative to the NBA season average) and relative ORtg we still see this:

Btw, people might ask what happens if you take out transition offense and look only at half court offense—more passes in the half court should help, right? This is more intense to show, but I looked at two seasons and there was zero relationship and even slightly inverse.

Now what about holding the ball? Without relative comparison:

And with relative time holding the ball and relative ORtgs:

Conclusion: making more passes, moving the ball and not holding on to the ball don’t seem to have any effect on offenses and if there is any, it’s a slightly negative one. This doesn’t mean that it’s a bad thing to “move the ball,” or that a team can’t have a good offense making more passes or not holding on to the ball, but the data we have (at least from the 2013-2014do not support the long held beliefs about making more passes and not holding on to the ball being inherently better for offenses.
a7

"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!"

Corny and embarrassing

A lot of the port/off ball arguments used against on-ball players essentially have always came down to hipotheticals since by the numbers they have never held up to amy scrutiny

The best example i cam think of this is how on-ball players having the best offensive dominance stretches ever with their teams amd usually more impressive than their theorically more off ball (and supposedly more portable and higher ceiling) peers is systematically ignored

I have seen ben taylor hold ball dominance against magic in all time discussions as if he and nash didnt lead the most dominant team oheffenses run ever.

At some point port is just a justification to rank your favorite players higher based on a theory that superficially makes sense but doesnt even scratch basketball beyond the surface

Indeed. There has to be some data that show more passes, higher frequency of passes, not holding on to the ball, etc., lead to better offense but they don’t in and of themselves. There’s nothing that shows that in the timeframe we have data. If more passes, not holding onto the ball in and of itself was beneficial, we’d see at least SOMETHING, but in this case, we actually see the opposite (slightly, but there).

[Djoker]

Do y'all realize that those graphs above show really weak correlation? Not saying anything regarding time of possession or passing and how it impacts ORtg but R squared of 0.2 is meaningless when trying to prove a point!

[Bidofo]

Indeed. There has to be some data that show more passes, higher frequency of passes, not holding on to the ball, etc., lead to better offense but they don’t in and of themselves. There’s nothing that shows that in the timeframe we have data. If more passes, not holding onto the ball in and of itself was beneficial, we’d see at least SOMETHING, but in this case, we actually see the opposite (slightly, but there).

Could you plot passes and seconds per touch with TOV as well? Intuitively there would be a connection there. The golden standard for motion offense Warriors were always fairly bad in that regard (even 28th in 2018 and 29th in 2022!)

[SNPA]

Spoiler:

Part 1: Argument Clarification

I'll come back later and post more, but right off the bat I want to mention a few things.



This is a strawman. I didn't say that the reduced number of passes is what directly led to better offenses, but rather it is part of the general pattern of what teams want to do and therefore a byproduct, i.e., get to their spots with the players they want taking the shots that they want which by its very nature requires fewer passes.

Perhaps I'm misinterpreting your first post, but it sounds like you're arguing
1. that "putting ball in the hands of a primary offensive" player has been the primary driver of "raw offenses are getting better and better", "instead of ball movement"
(I.e., heliocentrism has been the driver of the recent offensive explosion, and that it produces better offensive results than motion offenses)
2. and that there's evidence for point 1 in the form of 2a) "inverse relationship between passes per 100 possessions and ORtg", perhaps along with 2b) "the examples [therealbig3] mention about the very best offenses of all time”.

I’m saying this all misses the point. I’m saying the increase in offensive rating was dominated by other factors that your first post doesn’t mention. I’m saying teams can get “the players they want taking the shots that they want”, but that does not “by… very nature require fewer passes”. Teams want to maximize offensive efficiency on offense (as well as possessions), which entails efficient shots (always preferably open; preferably at the rim or at the 3 point line, but midrange can be acceptable if it’s open or diversifies the offense to make it harder to defend or from a player who’s comfortable there). I’m saying teams can generate those efficient shots with a variety of on-ball, off-ball, passing, screening, and spacing forms of creation. Passing, ball movement, and off-ball movement are all inarguably methods of generating an open or efficient shot, which is what teams want.

Specifically, I’m arguing:
1. that the increase in volume and accuracy in 3 point shooting is ~~ by far ~~ the clear driver of the recent offensive explosion.
2. that any changes in passing (or heliocentrism more generally) over the decade has had a significantly weaker relation to why offenses are improving.
-> 1 and 2 are statistically justified, and shown below

3. that any relatively weak negative correlation between passing and offensive rating is not sufficient evidence to prove that heliocentrism is better than ball movement in general, and that these relatively weak correlations across all teams are not applicable to teams at the top level or to top-end players like the ones discussed here.
(i.e. that your 2a and 2b are wrong)
-> 3 is statistically justified, and shown below

4. that, in fact, if there is any trend in the noise among all-time team play styles, it suggests that movement offenses produce as good if not better results than heliocentrism. (but that analysis of hyper small-sample playoff-only all-time team peak play styles can be fraught with uncertainty/error/bias)
-> 4 is statistically supported across basically any survey of top teams that goes deeper than a sample of just one or two teams, as per my previous post

5. and that, regardless, passing is not an ideal differentiator for motion offenses vs heliocentric offenses, as players in a motion offense can move off ball without necessarily increasing passing, not all heliocentric offenses pass the least, and there's a variety of other play styles besides motion and heliocentrism and factors more generally that could muddy this trend.
(i.e. that 2a is more wrong).
-> 5 is readily obvious from any film study of teams with different play styles.

1. Yes, non-helio teams and teams that make more passes have improved their offenses as well, but from 2014 to 2024, the years for which we have public data, more passes per 100 possessions tends to correlate with lower ORtgs. There are seasons from 2013-2014 to 2023-2024 in which there is a strongish inverse relationship between passes per 100 possessions and ORtg, i.e., more passes means lower ORtg, but NO SEASONS in which there is even more than a neutral relationship between passes per 100 possessions and ORtg. Not a single season. So, everyone is taking more threes, but the teams making fewer passes within this new environment of taking threes are also faring better.

2. The reduction in passing is more than what you're stating because you're not accounting for pace. Take a look at passes per 100 possessions. In 2013-2014, it was 314.5 passes per 100 and in 2023-2024, it was 285.6 If you look at 2014-2015 to 2022-2023, it went from 319.5 to 284.4. That's a change of 11%, not 4.6%. From 2014 to 2014 it would be a 9.2% change.

3. There's not any relationship between passes per 100 possessions (other than one, maybe another season) and three point attempt rate--the differences in ORtg are happening elsewhere.

Three point attempt rate and passes per 100 possessions

2014: slight inverse, i.e, more passes led to lower three point attempt rate
2015: zero
2016: zero
2017: zero
2018: strongest inverse relationship
2019: slight negative
2020: stronger inverse
2021: zero
2022: zero
2023: slight positive
2024: zero

So, for every season in which we have data, fewer passes tend to either lead to the same amount of three point attempts, or more three point attempts and generally higher ORtgs. Either way, it doesn't speak much to making more passes--either it leads to fewer three point attempts and a lower ORtg, same amount of three point attempts and a lower ORtg, greater three point attempt rate (one season) and a slightly great ORTg.

But more passes per 100 tend to lead to lower ORtgs.

ORtg and passes per 100 possessions

2014: slight inverse, i.e, more passes led to lower ORtg
2015: zero or slight inverse
2016: zero or slight inverse
2017: slight inverse
2018: strongest inverse relationship
2019: slight negative
2020: stronger inverse
2021: slight inverse, i.e, more passes led to lower ORtg
2022: slight inverse, i.e, more passes led to lower ORtg
2023: basically zero or slight positive, i.e., more passes led to a higher ORtg
2024: zero

Now... I haven't even gotten to average seconds per touch...

A brief quibble, before getting into the data: Why is it necessary to normalize for pace when looking at total passing?

When pace increases, teams play more possessions in a game, which require possessions are ending more quickly. If possessions are ending more quickly, that requires that some amount of points, assists (which sometimes come with points), rebounds, blocks, steals, and/or turnovers — all the things that might end a possession — increase. But passing is not a thing that ends a possession. If a team passes every 10 seconds they have the ball on average, and they have the ball for the same amount of time per game (i.e. they still have the ball ~50% of the time), then they will have the same number of total passes per game, regardless of their pace.

So it’s not clear, at least not a priori, that normalizing passes per 100 possessions is a better measure of ‘ball movement’ (and again ball movement is not the same as motion offense). If you plot passes vs pace, and draw a line of best fit, it looks like you’re holding a ruler up against a Jackson Pollack painting. Sure, you get out a negative slope, but it’s just great statistics.

Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses? That's not justified, as I'll show in a sec. But since it’s easy to look at both passes per game and passes per 100, I'll look at both.

Part 2: Statistical Evidence for the influence of 3 Point Shooting vs Passing on Offensive Ratings
1. 3 point shooting is what’s driving improved modern ORtg, and decreased passing is a poor model in comparison.
You claim that “putting ball in the hands of a primary offensive" player is the major driver of recent improvements offensive rating (i.e. that heliocentrism > movement offenses), and claim that a) the “strongish inverse relationship between passes per 100 possessions and ORtg” shows this, along with b) the best offenses ever are more heliocentric.

I’ve already disproved b in my last post. I’ve also claimed that a. is missing the point, that improved 3 point shooting is a significantly better driver, that the trend of passing vs ORtg is not very strong in comparison, and that it is insufficient to argue heliocentrism > movement offenses for good teams.

Let’s look at the R^2 value and p value for passes made per game vs ORtg, passes made per 100 vs ORtg, 3PA per game vs Ortg, and 3PA per 100 vs ORtg, across all available seasons (2014–2024 regular season)

What are these terms? They’re very standard measures in statistics.
Model: In this case, the model is the idea that ORtg would be proportional to either passes per game, passes per 100, 3PA per game, 3PA per 100.
R^2: Measures how much your model explains variation in the data, and thus can tell you how well your model fits the data. Range: 0 to 1. Best case: 1 is best, a minimum threshold of 0.5 is often set for an acceptable R^2 value, although in noisier cases (e.g. studies of human behavior) a minimum good threshold can be set to 0.4 or even lower.
p-value: Measures the probability that your model is better than the null hypothesis (i.e. that the trend doesn’t exist), and thus measures whether your model is statistically significant. Best case: 0 is best, and a maximum threshold of 0.05 is often set as an acceptable p-value.

How big is the best-fit slope?
-passes made per game vs ORtg: slope = -0.08
-passes made per 100 vs ORtg: -0.08
-3PA per game vs Ortg: 0.45
-3PA per 100 vs ORtg: 0.48

So the best-fit dependence suggests ORtg improves 600% more on average by taking one more 3PA per 100 than by taking one less pass per 100. The dependence on 3PA is much larger.


How do the models compare, from worst to best:
-passes made per game vs ORtg: R^2 = 0.13, p-value = 1.5 * 10^-11
-passes made per 100 vs ORtg: R^2 = 0.2, p-value = 8,4 * 10^-18
-3PA per game vs Ortg: R^2 = 0.40, p-value = 1.7 * 10^-38
-3PA per 100 vs ORtg: R^2 = 0.41, p-value = 7.0 * 10^-39

So increased 3 point shooting attempts fits the improved ORtg data over 300% better than the decrease in passing, and over 200% better than increased passing per 100. And 3 point shooting attempts have a better p-value by over 20 orders of magnitude!

Saying that decreases in passing improved ORtg is statistically significant (it’s better than doing nothing), but it is below any normal threshold for how well a good model should explain the variation in the data.

As I predicted, increased 3 point shooting volume is — by a country mile — a better model for explaining the improved ORtg in recent seasons than decreased passing. This doesn’t even include improved 3P%, which also has a significantly better R^2 (0.35) and p-value (1.2 * 10^-31), which would make overall 3 point shooting an even better model for explaining the improved ORtg in recent seasons.

In comparison, focusing on the decrease in passing to explain the ORtg improvements just misses the point. It is furthermore not a great proxy for movement offenses (as above, it misses off-ball movement). And therefore does not act as sufficient evidence to justify that heliocentrism > movement offenses, or that “putting ball in the hands of a primary offensive" player is the major driver of recent improvements offensive rating instead of just teams shooting 3s more (alongside all the supplemental spacing benefits of doing so).

2. There is no clear trend that decreasing passing improves offenses on good teams.
Let’s perform the same procedure as above, but filtering just for good teams.

If you look at passes per 100 vs ORtg for teams with ORtg > {113 (i.e. the top 95 teams of the past 10 years), 114, 115, and 117}, the best R^2 value is 0.04 (bad) and the best p-value is 0.14 (bad).
If you look at passes per 100 vs relative ORtg for teams with rORtg > {+0 (i.e. the top 166 teams of the past 10 years), +1, +2, +3, +4, and +5}, the best R^2 value is 0.002 (absolute trash) and the best p-value is 0.52 (absolute trash).

In other words, there is no statistically significant argument that more passing leads to worse offenses for teams with ORtg > 113 (top 95 teams of the last decade) or even for relative rORtg > 0 (top 166 teams of the decade). So unless you think Bird or whatever motion star you care to think of isn’t capable of raising a team to even a league average offense (which would be silly), the idea that the passing data somehow suggests motion offenses produce worse top-end results than heliocentric offenses is bogus.

You might push back against this by saying it’s a matter of sample size. Maybe 95 teams or 166 teams aren’t a large enough sample size to pick out the trend, so maybe passing is still worse for top-end offenses. But if that’s the case, if 95 teams or 166 teams aren’t a large enough sample to pick out the trend, how strong could this trend really be?? If it’s such a weak trend, then it’s pretty clearly unreasonable to use this to justify saying heliocentrism > motion for above-average offenses.

This actually raises something interesting: if you believe the trend that more passing produces worse offenses is real, it is only justifiable when applied to bad / below-average offenses. Why could this be? Here’s a few possibilities, although I”m open to other ideas:
a) Causal: bad teams with bad passing hurt their offenses with turnovers, and these teams have reduced turnovers from bad passes over the past decade.
There’s a statistically significant correlation between turnovers per 100 and offensive rating (it’s actually more statistically significant than your claimed passing vs ORtg), and there’s a clear causal connection (if you turn the ball over, you don’t get a good shot up that possession, which reduces your ORtg).
Bad offensive teams with bad passers used to turn the ball over more, but have since reduced passes that might lead to turnovers, thus improving their ORtg (alongside improved 3 point shooting being a bigger factor in their improved ORtg). Additionally, turnovers may also have been reduced by the wider passing lanes caused by 3 point shooting.

b) Symptomatic: bad teams aren’t able to break down the defenses as effectively, and so they end up passing the ball more to restart new actions as more of their initial actions fail.
In this model, bad teams from a decade ago were impotent passers, playing hot potato hoping one of their teammates could generate something better without much success from anyone.
Since then, it’s become easier for teams to break down defenses (aided primarily by increased 3 point shooting), and so there’s less of a need for impotent hot-potato passing.

In either case, good teams either a) gain the benefits of passing to balance or even outweigh the downsides of turnovers, or b), show less symptomatic impotent passing as they’re better able to break down defenses (including with passing, but also with a variety of other methods), and thus good teams show no correlation between passes per 100 and worse ORtg.

tldr:
-teams are better offenses now because of 3 point shooting much more than decreased passing.
-passing does not produce worse offenses on above-average teams.
-there's not enough evidence to claim heliocentrism > motion, based on what's described here, and based on the surveys of top level teams mentioned in my previous posts.

...
"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!" EDIT: APPLES!!

Once again, you have written a wall of text addressing a strawman, and you have been vigorously attacking that strawman. Here is the original conversation:

Do you have proof of this? Because the best offenses are helio offenses. By a clear margin. Helios like LeBron and Nash led better playoff offenses than Curry did on the Warriors...not to mention that Curry is plenty helio himself, he just does more off-ball than the rest.

And the Warriors offense stalled pretty much whenever Curry was NOT on ball as much as he should have been, but when they stubbornly stuck to moving him around off ball. Bird doesn't really have a choice, he can't be on ball, he doesn't have the skills for it.

In addition to the examples you mention about the very best offenses of all time, the last 10 years have shown that teams understand exactly what they want to do on offense, and have the ball in the hands of a primary offense who is much better at it than others on his team, and instead of “ball movement,” they get to their spots through their favorite actions using fewer passes.

Teams are making fewer and fewer passes over the last 10 years, but raw offenses are getting better and better. People will say, “But there’s no defense…” so if that were the case, how come teams that make more passes, have more ball movement, still, in general, have worse offenses even within this worse defense era?

Where did I say anything about threes in my initial discussion? Teams knowing exactly what they want to INCLUDES taking threes. The first nested comment was about Helios being easier to defend, but we’ve seen Helios along with teams that make fewer passes and holding the ball longer do as well or better than teams that move the ball more and don’t hold on to the ball as long. YOU then brought up threes, and begin attacking a strawman.

A brief quibble, before getting into the data: Why is it necessary to normalize for pace when looking at total passing?

When pace increases, teams play more possessions in a game, which require possessions are ending more quickly. If possessions are ending more quickly, that requires that some amount of points, assists (which sometimes come with points), rebounds, blocks, steals, and/or turnovers — all the things that might end a possession — increase. But passing is not a thing that ends a possession. If a team passes every 10 seconds they have the ball on average, and they have the ball for the same amount of time per game (i.e. they still have the ball ~50% of the time), then they will have the same number of total passes per game, regardless of their pace.

So it’s not clear, at least not a priori, that normalizing passes per 100 possessions is a better measure of ‘ball movement’ (and again ball movement is not the same as motion offense). If you plot passes vs pace, and draw a line of best fit, it looks like you’re holding a ruler up against a Jackson Pollack painting. Sure, you get out a negative slope, but it’s just great statistics.

Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses? That's not justified, as I'll show in a sec. But since it’s easy to look at both passes per game and passes per 100, I'll look at both.

For someone who likes using numbers, it’s really strange to see you ask why we need to adjust for pace. This seems mostly likely because you forgot about the need to adjust for pace and now came back with this to justify you overlooking/forgetting/not knowing to do this. If we want to know if teams are making more or fewer passes we have to adjust for how many opportunities there are to make passes. The league plays faster today than it did in 2014, YET there are fewer passes on average. You can’t compare average passes a team makes per game in a 94 possession pace environment vs. how many they make in a 98 possession environment. In a faster league, you would think there would be more passes yet there are fewer. You then top it off with yet another strawman: “Regardless, are you really saying that a 9.2% reduction in passing has even a comparable effect to the 33.4% increase in 3 point shooting volume on offenses?” I didn’t make that argument. You’re arguing with yourself.

The central point is that it’s just accepted by most posters and the poster whose comment I was replying to that “moving the ball” and “not ball stopping” are better for offenses but we just don’t have proof of that. If it were true, we’d see SOME relationship between moving the ball+not holding the ball on offenses, but we don’t see any at all. Now, that doesn’t mean causation nor did I ever say that it did, but we also have no numbers telling us moving the ball and not holding on to the ball help offenses. This is over a course of 10 seasons with hundreds of thousands of touches and passes.

Obviously ORtgs have gone up so a straight comparison of passes per 100 and ORtgs is going to look like this:

But if we adjust for relative passes per 100 (i.e., relative to the NBA season average) and relative ORtg we still see this:

Btw, people might ask what happens if you take out transition offense and look only at half court offense—more passes in the half court should help, right? This is more time intensive to show, but I looked at two seasons and there was zero relationship and even slightly inverse.

Now what about holding the ball? Without relative comparison:

And with relative time holding the ball and relative ORtgs:

Conclusion: making more passes, moving the ball and not holding on to the ball don’t seem to have any effect on offenses and if there is any, it’s a slightly negative one. This doesn’t mean that it’s a bad thing to “move the ball,” or that a team can’t have a good offense making more passes or not holding on to the ball, but the data we have (at least from the 2013-2014do not support the long held beliefs about making more passes and not holding on to the ball being inherently better for offenses.

"Do you like applies?"
"... yeah."
"Well I got [numbers]! How do you like them applies!"

Corny and embarrassing

My original post you quoted isn’t about passing necessarily, it was about defensive schemes engaging helio’s vs off ball threats.

There is also a giant assumption built into this post, which is helio offenses pass less which seems intuitive but isn't universally true by default (maybe in some/many cases). The helio is ball dominant (reducing passing) but the players teams sometimes put around helio’s aren’t ball dominant….there’s only one ball. Those players tend to move the ball quickly if they don’t have a shot or driving lane. The whole 0.5 thing. Helio pounds the air out of it for 15 seconds, drives and kicks, it gets swung with two quick passes for a wide open three. There is not inherently a lack of passing there.

Also, the helio phenomenon is most a modern one, so it’s pace and space era. That is impacting a lot of the analysis. How would a helio offense do with ten guys playing inside the arch?

[AEnigma]

Do y'all realize that those graphs above show really weak correlation? Not saying anything regarding time of possession or passing and how it impacts ORtg but R squared of 0.2 is meaningless when trying to prove a point!

Hm, let me see, do we realise that.

The central point is that it’s just accepted by most posters and the poster whose comment I was replying to that “moving the ball” and “not ball stopping” are better for offenses but we just don’t have proof of that. If it were true, we’d see SOME relationship between moving the ball+not holding the ball on offenses, but we don’t see any at all.

Conclusion: making more passes, moving the ball and not holding on to the ball don’t seem to have any effect on offenses and if there is any, it’s a slightly negative one. This doesn’t mean that it’s a bad thing to “move the ball,” or that a team can’t have a good offense making more passes or not holding on to the ball, but the data we have (at least from the 2013-2014 do not support the long held beliefs about making more passes and not holding on to the ball being inherently better for offenses.

There has to be some data that show more passes, higher frequency of passes, not holding on to the ball, etc., lead to better offense but they don’t in and of themselves. There’s nothing that shows that in the timeframe we have data.

Unfortunately, to figure that out, you would need to read what was written rather than posting reactively as soon as you registered the graphs had some abstract connection to Lebron.

[homecourtloss]

Do y'all realize that those graphs above show really weak correlation? Not saying anything regarding time of possession or passing and how it impacts ORtg but R squared of 0.2 is meaningless when trying to prove a point!

Hm, let me see, do we realise that.

The central point is that it’s just accepted by most posters and the poster whose comment I was replying to that “moving the ball” and “not ball stopping” are better for offenses but we just don’t have proof of that. If it were true, we’d see SOME relationship between moving the ball+not holding the ball on offenses, but we don’t see any at all.

Conclusion: making more passes, moving the ball and not holding on to the ball don’t seem to have any effect on offenses and if there is any, it’s a slightly negative one. This doesn’t mean that it’s a bad thing to “move the ball,” or that a team can’t have a good offense making more passes or not holding on to the ball, but the data we have (at least from the 2013-2014 do not support the long held beliefs about making more passes and not holding on to the ball being inherently better for offenses.

There has to be some data that show more passes, higher frequency of passes, not holding on to the ball, etc., lead to better offense but they don’t in and of themselves. There’s nothing that shows that in the timeframe we have data.

Unfortunately, to figure that out, you would need to read what was written rather than posting reactively as soon as you registered the graphs had some abstract connection to Lebron.

Saved me some time – thank you!

[DraymondGold]

Do y'all realize that those graphs above show really weak correlation? Not saying anything regarding time of possession or passing and how it impacts ORtg but R squared of 0.2 is meaningless when trying to prove a point!

Hey Djoker! Yeah it's not really that compelling of a fit. And just to re-emphasize -- the trend completely disappears statistically when we look at above average teams. So the trend only exists for below-average teams.

If it's real -- and I've tried to emphasize reasons why it's not that compelling in my OP -- that could suggest something like bad teams are worse because they can't take care of the ball (causation: too much passing makes bad teams worse), or they're doing passing when they can't break down the defense in another way (correlation: the excess passing is a downstream symptom of an upstream offensive impotency).
But for good teams, the trend doesn't exist.

The irony is that, despite being used to defend the anti-port crowd's argument ("the numbers they have never held up to amy scrutiny")... this change across team value is actually a point in favor of port! The idea of portabilit is not that one playstyle is inherently more valuable than another a priori. It's about how value changes as your teammates/team gets better.
So if this trend is real (and I've tried to emphasize reasons why it's not that compelling) and can be applied in the way people want (if high passing teams can be a proxy of motion offense and/or of having a high-port best player, which it isn't the best for)... then it seems like helio styles are only better for worse teams, and that helio = motion for good teams. I.e., either helio faces more diminishing returns on good teams, or motion faces less diminishing returns on good teams. Which is exactly what port has been saying the whole time! The only disagreement is that the difference in diminishing returns is not quite as large as port-people have said, but given how utterly massive the error bars should be in this calculation... this is pretty clearly not damning.

There's a second irony, which is that if fit the Thinking Basketball port scores (which are based film/statistical studies of playstyle) to actual Top 100 team rankings (say Sansterre's OSRS or Fivethirtyeight)... you actually get something more statistically significant than just fitting the passing data homecourtloss proposed to the Top 100 teams of the last decade! It's a different sample of teams, but we don't have the port scores for all the best players on the past 100 teams of the decade. Maybe one day I'll post the studies I've done into that, but given how acrimonious, immature, and stubborn people get in conversations of port, it makes one hesitant to post the work they've done for their own curiosity. If we don't have a better community, we end up pushing away the best of us like Squared2020. Maybe people would just prefer more of an echo chamber. Or maybe intellectual, mature debates with people we disagree with is too much to ask for with sports fans of all ages. IDK.

All that to say, I'm probably done discussing with the other side in this thread, given how quickly it dissolved. In a final irony, despite continuously accusing me of strawmannirg (even where I just directly quote what he said), homecourtloss... just about ignored all the actual points I made about the legitimate statistical concerns with the data he was presenting (bad R^2 and completely failing p-values for the team levels that are relevant in a discussion of Prime Bird in this thread), and the qualitative takeaways I made (why too much passing might only be a limitation for bad teams not good ones)... just to post another set of plots that show the exact same limitations. And ended it with another immature personal attack on me (thread #2 in a row). I guess I or someone with a similar opinion to me offended him at some point, cause he sure used to make more productive posts than this. But it seems our conversation is past saving.

Djoker, if you or others have other ideas on how we might test port in the data, I've been brainstorming methods. Obviously many are subject to limitations:
-passing vs team result (raw and relative ORtg, Net Rtg / SRS / Overall SRS / ELO; overall and looking at changes across team quality), a la homecourt's plots above (obvious limitations already discussed)
-touches vs team result (raw and relative ORtg, Net Rtg / SRS / Overall SRS / ELO; overall and looking at changes across team quality), a la homecourt's plots above(obvious limitations already discussed)
-A survey of the offensive play styles of the top team results, with a sample size more than just the top ~2 like it's normally done (I've done a first-pass looking at Top 15 teams across various categories a page back, but the play styles are done by hand with room for disagreement, and the sample size could still be larger)
-TB's port ratings vs team results. (I've done a first pass at this wrt OSRS and ELO and the results are statistically significant, but not ready to post yet, given the likely torrent of nay-sayers it will get)
-Comparing the fit of TB's plus minus evaluations (his evaluations without port) to his healthy CORP evaluations (his evaluations with port) to team results
-Some sort of broad study of some impact metric vs player archetype, again across team quality. We would need quantifications for each of these. There's plenty of websites that attempt quantitative categorizations of player archetype, but many have flaws.
-Looking more into some method like Craftednba's attempt to quantify port, and compare it to team results / etc. I'm not crazy with the execution, but it's an interesting idea.
All of these have clear limitations, but they're possible ways forward. Open to suggestions as to whether any of these seem promising, or if you have any other ideas!

[tsherkin]

I figure I'd add, while the particulars of how he did it may require tweaking, AEnigma was doing some "quality of pass" type stuff in his tracking of Larry Bird. Like with most other things, I think the volume of passes is secondary to what they achieve.

Let's say your offense typically centers around a Lebron PnR. He brings the ball up the court, and in the half-court, you really only get the kick-out pass to the shooter. 1 pass. But it's a very effective one setting up the corner shot. That's not bad offense because of low pass volume. That's a very simple example which doesn't account for defensive adaptation, but just as an opener to the concept.

I think average defender distance and shot location are a lot more important for the final shot in the offense than how many passes went into it, you know what I mean? You don't NEED to have a pass-heavy offense, you need to have an offense that effectively engages the defense and produces high-quality looks around the basket, and open looks away therefrom. A team could play Ring Around the Rosie with the ball all day, and all that reversal will accomplish very little, after all. That's, again, an oversimplification, but there's no inherent value to passing just for the sake of moving the ball.

Perhaps we need categories of pass? Release pass on a dribble drive which didn't work. Kick out from the post. Pocket pass out of the PnR, that sort of thing?

[Djoker]

Do y'all realize that those graphs above show really weak correlation? Not saying anything regarding time of possession or passing and how it impacts ORtg but R squared of 0.2 is meaningless when trying to prove a point!

Hey Djoker! Yeah it's not really that compelling of a fit. And just to re-emphasize -- the trend completely disappears statistically when we look at above average teams. So the trend only exists for below-average teams.

If it's real -- and I've tried to emphasize reasons why it's not that compelling in my OP -- that could suggest something like bad teams are worse because they can't take care of the ball (causation: too much passing makes bad teams worse), or they're doing passing when they can't break down the defense in another way (correlation: the excess passing is a downstream symptom of an upstream offensive impotency).
But for good teams, the trend doesn't exist.

The irony is that, despite being used to defend the anti-port crowd's argument ("the numbers they have never held up to amy scrutiny")... this change across team value is actually a point in favor of port! The idea of portabilit is not that one playstyle is inherently more valuable than another a priori. It's about how value changes as your teammates/team gets better.
So if this trend is real (and I've tried to emphasize reasons why it's not that compelling) and can be applied in the way people want (if high passing teams can be a proxy of motion offense and/or of having a high-port best player, which it isn't the best for)... then it seems like helio styles are only better for worse teams, and that helio = motion for good teams. I.e., either helio faces more diminishing returns on good teams, or motion faces less diminishing returns on good teams. Which is exactly what port has been saying the whole time! The only disagreement is that the difference in diminishing returns is not quite as large as port-people have said, but given how utterly massive the error bars should be in this calculation... this is pretty clearly not damning.

There's a second irony, which is that if fit the Thinking Basketball port scores (which are based film/statistical studies of playstyle) to actual Top 100 team rankings (say Sansterre's OSRS or Fivethirtyeight)... you actually get something more statistically significant than just fitting the passing data homecourtloss proposed to the Top 100 teams of the last decade! It's a different sample of teams, but we don't have the port scores for all the best players on the past 100 teams of the decade. Maybe one day I'll post the studies I've done into that, but given how acrimonious, immature, and stubborn people get in conversations of port, it makes one hesitant to post the work they've done for their own curiosity. If we don't have a better community, we end up pushing away the best of us like Squared2020. Maybe people would just prefer more of an echo chamber. Or maybe intellectual, mature debates with people we disagree with is too much to ask for with sports fans of all ages. IDK.

All that to say, I'm probably done discussing with the other side in this thread, given how quickly it dissolved. In a final irony, despite continuously accusing me of strawmannirg (even where I just directly quote what he said), homecourtloss... just about ignored all the actual points I made about the legitimate statistical concerns with the data he was presenting (bad R^2 and completely failing p-values for the team levels that are relevant in a discussion of Prime Bird in this thread), and the qualitative takeaways I made (why too much passing might only be a limitation for bad teams not good ones)... just to post another set of plots that show the exact same limitations. And ended it with another immature personal attack on me (thread #2 in a row). I guess I or someone with a similar opinion to me offended him at some point, cause he sure used to make more productive posts than this. But it seems our conversation is past saving.

Djoker, if you or others have other ideas on how we might test port in the data, I've been brainstorming methods. Obviously many are subject to limitations:
-passing vs team result (raw and relative ORtg, Net Rtg / SRS / Overall SRS / ELO; overall and looking at changes across team quality), a la homecourt's plots above (obvious limitations already discussed)
-touches vs team result (raw and relative ORtg, Net Rtg / SRS / Overall SRS / ELO; overall and looking at changes across team quality), a la homecourt's plots above(obvious limitations already discussed)
-A survey of the offensive play styles of the top team results, with a sample size more than just the top ~2 like it's normally done (I've done a first-pass looking at Top 15 teams across various categories a page back, but the play styles are done by hand with room for disagreement, and the sample size could still be larger)
-TB's port ratings vs team results. (I've done a first pass at this wrt OSRS and ELO and the results are statistically significant, but not ready to post yet, given the likely torrent of nay-sayers it will get)
-Comparing the fit of TB's plus minus evaluations (his evaluations without port) to his healthy CORP evaluations (his evaluations with port) to team results
-Some sort of broad study of some impact metric vs player archetype, again across team quality. We would need quantifications for each of these. There's plenty of websites that attempt quantitative categorizations of player archetype, but many have flaws.
-Looking more into some method like Craftednba's attempt to quantify port, and compare it to team results / etc. I'm not crazy with the execution, but it's an interesting idea.
All of these have clear limitations, but they're possible ways forward. Open to suggestions as to whether any of these seem promising, or if you have any other ideas!

Great post.

I think the issue with any potential tests is the interplay of so many factors in basketball. As you said, bad teams passing more can lead to more turnovers. Even good teams passing more can lead to more turnovers if those particular players' weaknesses include taking care of the ball. Or more passing can be the result of other underlying issue as you said.

Another thing is that the optimal style of play may well be different in different eras. For instance we definitely know that an elite defensive big (hello Bill Russell) was the best type of superstar to have in the 60's (= most impact) but that clearly isn't so today. And prior to 2013-14, we don't have tracking for passes so any kind of study we do will only answer questions (or fail to answer questions ) for the last 11 seasons of the NBA. And even then I'd argue that the NBA in 2013-14 is substantially different than the NBA today so we have a basket full of apples and oranges that we're analyzing. But if we only study a few seasons, then we run into issues of small sample size.

Also it isn't obvious whether looking at passing/touches in relation to ORtg or Net Rtg is the right move. ORtg is obviously directly impacted by passing and DRtg obviously is not but again, interplay between a gazillion variables is there. For example, it could be that heliocentrism is subpar for offense but it actually helps the team overall because the other four players watching James Harden don't have to do much work on offense except stand at the 3pt line so they have a lot of energy on the defensive end! Or it could be that Harden himself gets so tired on offense that he blows up the whole team on the defensive end. Or that most shooters are poor rebounders and are getting smoked on the glass or something. Overall I'd argue that studying the effect of any variable on Net Rtg (as opposed to ORtg) is better for that reason.

Please post your study on portability vs. team results. Forget the torrent of naysayers. I wanna see it!

[SNPA]

I figure I'd add, while the particulars of how he did it may require tweaking, AEnigma was doing some "quality of pass" type stuff in his tracking of Larry Bird. Like with most other things, I think the volume of passes is secondary to what they achieve.

Let's say your offense typically centers around a Lebron PnR. He brings the ball up the court, and in the half-court, you really only get the kick-out pass to the shooter. 1 pass. But it's a very effective one setting up the corner shot. That's not bad offense because of low pass volume. That's a very simple example which doesn't account for defensive adaptation, but just as an opener to the concept.

I think average defender distance and shot location are a lot more important for the final shot in the offense than how many passes went into it, you know what I mean? You don't NEED to have a pass-heavy offense, you need to have an offense that effectively engages the defense and produces high-quality looks around the basket, and open looks away therefrom. A team could play Ring Around the Rosie with the ball all day, and all that reversal will accomplish very little, after all. That's, again, an oversimplification, but there's no inherent value to passing just for the sake of moving the ball.

Perhaps we need categories of pass? Release pass on a dribble drive which didn't work. Kick out from the post. Pocket pass out of the PnR, that sort of thing?

This is all true but it hurts James more than Bird. James’ game has to change more going back into eras where ten guys played inside the arch. There simply wasn’t enough space for his drive and kick game to work as well. Whereas Bird’s game gets better in the pace and space era. Give Bird more room and freedom of movement…yeah…you’re going to lose.

James is certainly the drive and kick master passer. Clearly his strength. Bird stands no chance in that category. Leading the break, again, James is a monster. Outside that…lobs and?

[theonlyclutch]

I figure I'd add, while the particulars of how he did it may require tweaking, AEnigma was doing some "quality of pass" type stuff in his tracking of Larry Bird. Like with most other things, I think the volume of passes is secondary to what they achieve.

Let's say your offense typically centers around a Lebron PnR. He brings the ball up the court, and in the half-court, you really only get the kick-out pass to the shooter. 1 pass. But it's a very effective one setting up the corner shot. That's not bad offense because of low pass volume. That's a very simple example which doesn't account for defensive adaptation, but just as an opener to the concept.

I think average defender distance and shot location are a lot more important for the final shot in the offense than how many passes went into it, you know what I mean? You don't NEED to have a pass-heavy offense, you need to have an offense that effectively engages the defense and produces high-quality looks around the basket, and open looks away therefrom. A team could play Ring Around the Rosie with the ball all day, and all that reversal will accomplish very little, after all. That's, again, an oversimplification, but there's no inherent value to passing just for the sake of moving the ball.

Perhaps we need categories of pass? Release pass on a dribble drive which didn't work. Kick out from the post. Pocket pass out of the PnR, that sort of thing?

This is all true but it hurts James more than Bird. James’ game has to change more going back into eras where ten guys played inside the arch. There simply wasn’t enough space for his drive and kick game to work as well. Whereas Bird’s game gets better in the pace and space era. Give Bird more room and freedom of movement…yeah…you’re going to lose.

James is certainly the drive and kick master passer. Clearly his strength. Bird stands no chance in that category. Leading the break, again, James is a monster. Outside that…lobs and?

Of course, what is conveniently absent in this discussion is that what Bird brings to an offense is inherently reliant on competent ball-handlers making sound decisions in a way that players like Luka/SGA/Lebron/Curry aren't. The same competent ball handlers which are made much harder to keep in the medium/long term with cap rules in 2024 being so punitive against stacking contracts.

[Woodsanity]

I rank him number 2 behind Jokic.