Sloan'25
Re: Sloan'25 panel topic
No, they haven't responded yet. I just look at team level data and mentioned the source, intending that to offer clarity (using same source). If they respond I'll specifically say so.
Re: Sloan'25 panel topic
Dean Oliver says:
"As a percentage of shots, layups have gone up in the last few years. But that does rely on play-by-play data labeling. Tracking data is a little less clear."
I had asked:
"From the data you see / use, do you agree or disagree with the statement that 'shots [at] the rim have remained the same [or about the same] during the increase in 3ptas'?
No response from BRef yet.
Mike G's modified data for dunks & layups suggest a mild decline. Probably go with that unless I see a different / better dataset.
0-3 feet appears less than ideal. 0-5 feet could be better.
"As a percentage of shots, layups have gone up in the last few years. But that does rely on play-by-play data labeling. Tracking data is a little less clear."
I had asked:
"From the data you see / use, do you agree or disagree with the statement that 'shots [at] the rim have remained the same [or about the same] during the increase in 3ptas'?
No response from BRef yet.
Mike G's modified data for dunks & layups suggest a mild decline. Probably go with that unless I see a different / better dataset.
0-3 feet appears less than ideal. 0-5 feet could be better.
Re: Sloan'25 panel topic
Layup % is at near high
according to BRef but at its lowest level now by Mike's revised numbers.
Seems like a general topic worthy of more study & comment.
according to BRef but at its lowest level now by Mike's revised numbers.
Seems like a general topic worthy of more study & comment.
Re: Sloan'25 panel topic
League average attempts per game by nba.com within 5 feet:
2019-20 31
20024-25 28.2
(my calculation, since league totals are absent)
Almost 10% decline here.
Not a lot but not nothing.
Would be more noticeable if trend continued another 5 years.
Should the talk shift from 'no decline' for at rim rate to modest decline? I think so. Unless someone has a dataset clearly with no decline and a case that it is the best dataset on the topic.
2019-20 31
20024-25 28.2
(my calculation, since league totals are absent)
Almost 10% decline here.
Not a lot but not nothing.
Would be more noticeable if trend continued another 5 years.
Should the talk shift from 'no decline' for at rim rate to modest decline? I think so. Unless someone has a dataset clearly with no decline and a case that it is the best dataset on the topic.
Re: Sloan'25 panel topic
BRef still hasn't responded.
Re: Sloan'25 panel topic
I think the idea that there is a problem and disconnect between the dunk and layup and shot distance data on basketball reference is incorrect.
You can calculate what the implied FG%s are for dunks and layups from the data there, and they only makes sense (i.e. dunks coming in around 90% and layups mid-50%s) if the denominator is as stated (total FGA). Conversely if you assume the denominate for layups is 2PAs, the answers over time swing wildly and would be way off (e.g. 97.1% in 2002 and 67% this year to recent date).
I think the confusion lies in the interpretation of dunks and layups. These categories describe finishes, whereas the shooting distances (must?) describe the take-off location of the shot. And, of course, many dunks and layups get taken from beyond 3 feet.
Supposing this, to interpret the data, it is helpful to lump together dunks and layups as a fraction of the shots taken from 0-3 and 3-10 feet (as no dunks and layups occur with the shooter leaving the floor from beyond that distance).
As such, you can see the change from 2002 (when the layup data first is posted in B-R.com) to this season is that "shots around the rim" (defined as dunks and layups) increased in share from 26.2% to 32.5% (with the aggregated FG% increasing marginally from 61.5% to 62.2%).
You can also calculate the numbers for the "non-dunks and non-layups" (NDNL) within 10 feet of the basket. The fraction of such attempts drops, from 2002 to present, from 17.3% to 11.5%. However, the FG% rises from 39.9% to 46.6%
I don't think any of this is surprising. The increase in dunks and layups over time (and the growing disconnect with the distance data) makes sense if you think of more of these shots coming from slashing to the basket, rather than from post-up play.
The (excruciatingly slow adaptation of the) three point shot had the predicted effect of getting rid of bad shots, and not all of those were long mid-range jumpers; some were within ten feet of the basket. Of course, the value of the NDNL FG% is underrepresented in that it doesn't include the premium of shots that were more likely to have resulted in free throws, but the point remains the same.
P.S. I should belabor the point about how truly awful the shots were within 10 feet of the basket that got replaced over time (with three pointers). The share of these NDNL shots decreased by 5.8 percentage points (17.3% to 11.5%, again, of all FGA). And the implied FG% of these no-longer-taken shots (so as to get the realized NDNL FG% to rise 6.7 percentage points - 39.9% to 46.6%) is a horrific 26.6%. For those who bemoan the aesthetics of today's NBA, this was the beautiful game of their gauzy memories.
You can calculate what the implied FG%s are for dunks and layups from the data there, and they only makes sense (i.e. dunks coming in around 90% and layups mid-50%s) if the denominator is as stated (total FGA). Conversely if you assume the denominate for layups is 2PAs, the answers over time swing wildly and would be way off (e.g. 97.1% in 2002 and 67% this year to recent date).
I think the confusion lies in the interpretation of dunks and layups. These categories describe finishes, whereas the shooting distances (must?) describe the take-off location of the shot. And, of course, many dunks and layups get taken from beyond 3 feet.
Supposing this, to interpret the data, it is helpful to lump together dunks and layups as a fraction of the shots taken from 0-3 and 3-10 feet (as no dunks and layups occur with the shooter leaving the floor from beyond that distance).
As such, you can see the change from 2002 (when the layup data first is posted in B-R.com) to this season is that "shots around the rim" (defined as dunks and layups) increased in share from 26.2% to 32.5% (with the aggregated FG% increasing marginally from 61.5% to 62.2%).
You can also calculate the numbers for the "non-dunks and non-layups" (NDNL) within 10 feet of the basket. The fraction of such attempts drops, from 2002 to present, from 17.3% to 11.5%. However, the FG% rises from 39.9% to 46.6%
I don't think any of this is surprising. The increase in dunks and layups over time (and the growing disconnect with the distance data) makes sense if you think of more of these shots coming from slashing to the basket, rather than from post-up play.
The (excruciatingly slow adaptation of the) three point shot had the predicted effect of getting rid of bad shots, and not all of those were long mid-range jumpers; some were within ten feet of the basket. Of course, the value of the NDNL FG% is underrepresented in that it doesn't include the premium of shots that were more likely to have resulted in free throws, but the point remains the same.
P.S. I should belabor the point about how truly awful the shots were within 10 feet of the basket that got replaced over time (with three pointers). The share of these NDNL shots decreased by 5.8 percentage points (17.3% to 11.5%, again, of all FGA). And the implied FG% of these no-longer-taken shots (so as to get the realized NDNL FG% to rise 6.7 percentage points - 39.9% to 46.6%) is a horrific 26.6%. For those who bemoan the aesthetics of today's NBA, this was the beautiful game of their gauzy memories.
Re: Sloan'25 panel topic
Yeah NDNLs are bad shots.
I don't miss them, have pointed them out as low fg% several other places recently, but did want to recognize the change in frequency by BRef data that some on panel were saying didn't exist.
I don't miss them, have pointed them out as low fg% several other places recently, but did want to recognize the change in frequency by BRef data that some on panel were saying didn't exist.
Re: Sloan'25
Just to complete the thought, we can add in the observed changes between 2002 and today from the other B-R.com zone distances (10 to 16 feet, and 16 feet to the 3 point line).
The inferred quality of mid-range attempts that were eliminated by transitioning to the "modern" 3P NBA from mid-range was 30.6% (with FG% from that distance rising from 38.8% to 44.4%). And FGAs from this distance decreased by 5.9%age points
For long-twos, by contrast, there was essentially no change in observed completion rates, actual FG% rising from 39.7% to just 40% (so a marginal 39.6%). But here was the biggest change in shot selection, with a decrease of 9.1%age points in FGAs.
So, the biggest per shot gains (with the possible caveat about anticipated foul rates) came from reducing non-dunk, non-layup attempts, but these were reduced less than 2PAs from the other two distances.
The inferred quality of mid-range attempts that were eliminated by transitioning to the "modern" 3P NBA from mid-range was 30.6% (with FG% from that distance rising from 38.8% to 44.4%). And FGAs from this distance decreased by 5.9%age points
For long-twos, by contrast, there was essentially no change in observed completion rates, actual FG% rising from 39.7% to just 40% (so a marginal 39.6%). But here was the biggest change in shot selection, with a decrease of 9.1%age points in FGAs.
So, the biggest per shot gains (with the possible caveat about anticipated foul rates) came from reducing non-dunk, non-layup attempts, but these were reduced less than 2PAs from the other two distances.
Re: Sloan'25
True dat. I look at the '90s Bulls highlights, and prime Jordan plays almost 'ugly' compared to SGA now.For those who bemoan the aesthetics of today's NBA, this was the beautiful game of their gauzy memories.
It should definitely be part of the relative 'efficiency' involved. And even then, there's the worth of creating foul trouble for opponents; and the effectiveness decline of a high-FG% & low-FT% player in the 'FT bonus situation'.... the biggest per shot gains (with the possible caveat about anticipated foul rates)...
Is it possible to find the FTA from any of these FGA zones? For player, team, or even just league-wide?
And how close to the basket are Turnovers occurring?
Re: Sloan'25
82games had some long ago articles with some tracking data on those details.
Should be available somewhere, some way. Maybe Synergy.
Should be available somewhere, some way. Maybe Synergy.
Re: Sloan'25
I've been mostly curious about % of PF drawn and TO within 2-pt and 3-pt territory.
Guessing that relatively few of each are outside the arc, but by assigning some % of FT/FTA to each zone would create a TS% for each.
And TO/FGA from each zone would give us an ORtg for each. Maybe subtract "heaves" -- beyond halfcourt shots.
(I see Jokic is 2/22 from super-long, bringing his 3FG% down by .024).
An interesting study would be the evolution of relative efficiencies, for league or for players. Harden FTr ranges from .42 to .65 -- correlation with 3PAr is -.63
Guessing that relatively few of each are outside the arc, but by assigning some % of FT/FTA to each zone would create a TS% for each.
And TO/FGA from each zone would give us an ORtg for each. Maybe subtract "heaves" -- beyond halfcourt shots.
(I see Jokic is 2/22 from super-long, bringing his 3FG% down by .024).
An interesting study would be the evolution of relative efficiencies, for league or for players. Harden FTr ranges from .42 to .65 -- correlation with 3PAr is -.63
Re: Sloan'25
Probably 15-20 years ago but fwiw:
https://www.82games.com/locations.htm
Ability to draw a shooting foul in the low paint would separate those takes from much of the rest of short 2s NDNL. The greater threat of a dunk or layup.
https://www.82games.com/locations.htm
Ability to draw a shooting foul in the low paint would separate those takes from much of the rest of short 2s NDNL. The greater threat of a dunk or layup.
Re: Sloan'25
Well, a season-by-season "prediction" of James Harden's FTA, based only on his 2FGA and 3FGA is:
FTA = .333*3FGA + .667*2FGA
So 100 FGA inside the arc creates twice as many FT as 100 FGA outside the arc.
For his first year and most of his later years, this formula overestimates his FTA.
For most of his Houston years, he shot more FTA relative to his FGA.
FTA = .333*3FGA + .667*2FGA
So 100 FGA inside the arc creates twice as many FT as 100 FGA outside the arc.
For his first year and most of his later years, this formula overestimates his FTA.
For most of his Houston years, he shot more FTA relative to his FGA.
Re: Sloan'25
I think there's another point to make note of regarding the change in shot selection between 2002 (again, chosen as that is the first year that B-R.com includes layup data) and today.
I would like to think it is now, at long last, generally recognized how terrible the NBA was in how slow it was in adopting the three-point shot, "inexplicably" refusing to shed low-expected-value 2PAs for higher ones beyond the arc.
However, if the numbers I recently shared are correct (i.e. the implied FG% of those relocated 2PAs at various distances) it suggests that even in a two-point-only context, there were counterfactual gains that could have been had by moving non-dunk/non-layup attempts from within 10 feet to farther out but still within the arc.
The 82games article that Crow linked to isn't dated (as I see it) but it is part of their game-charting project so probably pertained to data from about 2006. And I think it's fair for these numbers to be understood as also applicable to the 2002 season.
Anyway, there is no credible reading of those FTA/FGA by distance data that would overturn the conclusion that the worst shots on the court were those NDNL within 10 feet and marginal gains could have been had by taking them farther out, even inside the three-point line.
To restate the point, in 2002, twenty plus years after the adoption of the three-point shot, the NBA wasn't playing optimally even in a 2PA context.
Though this is worth saying on its own, if only for the fun of kicking around a mostly bygone NBA establishment, it is in a sense a point of current relevance.
For those who aren't happy with the modern NBA game (one of shots either behind the arc or around the basket) one should contemplate what the NBA would look like without a three-point shot. It cannot be assumed that the effect would be a return to the shot distribution of some bygone era. If NDNL FGA within 10 feet are in fact the worst shots on the court, then one would expect the adaption to result in the aesthetic pleasure of more "long twos" than ever before. And that would be something to behold...
I would like to think it is now, at long last, generally recognized how terrible the NBA was in how slow it was in adopting the three-point shot, "inexplicably" refusing to shed low-expected-value 2PAs for higher ones beyond the arc.
However, if the numbers I recently shared are correct (i.e. the implied FG% of those relocated 2PAs at various distances) it suggests that even in a two-point-only context, there were counterfactual gains that could have been had by moving non-dunk/non-layup attempts from within 10 feet to farther out but still within the arc.
The 82games article that Crow linked to isn't dated (as I see it) but it is part of their game-charting project so probably pertained to data from about 2006. And I think it's fair for these numbers to be understood as also applicable to the 2002 season.
Anyway, there is no credible reading of those FTA/FGA by distance data that would overturn the conclusion that the worst shots on the court were those NDNL within 10 feet and marginal gains could have been had by taking them farther out, even inside the three-point line.
To restate the point, in 2002, twenty plus years after the adoption of the three-point shot, the NBA wasn't playing optimally even in a 2PA context.
Though this is worth saying on its own, if only for the fun of kicking around a mostly bygone NBA establishment, it is in a sense a point of current relevance.
For those who aren't happy with the modern NBA game (one of shots either behind the arc or around the basket) one should contemplate what the NBA would look like without a three-point shot. It cannot be assumed that the effect would be a return to the shot distribution of some bygone era. If NDNL FGA within 10 feet are in fact the worst shots on the court, then one would expect the adaption to result in the aesthetic pleasure of more "long twos" than ever before. And that would be something to behold...
Re: Sloan'25
I wonder how many 3-10' shots were drives toward the basket that didn't go quite as planned, and the driver had to put up a less than optimal shot. A 32% FGA is better than a travel or a charge or passing turnover, especially if you might get the O-Reb.
Didn't we at one time assume that play-by-play or some refinement of it would tell us how many of a player's shots in a zone are O-Reb'd, fouled, or etc? Maybe it's out there for a subscription fee?
The best scorers have always been those who are dangerous from anywhere on the court. Luka, Kawhi, Steph, Jokic, Shai, LeBron, Harden, MJ, Bird ...
Didn't we at one time assume that play-by-play or some refinement of it would tell us how many of a player's shots in a zone are O-Reb'd, fouled, or etc? Maybe it's out there for a subscription fee?
The best scorers have always been those who are dangerous from anywhere on the court. Luka, Kawhi, Steph, Jokic, Shai, LeBron, Harden, MJ, Bird ...