2019-2020 Player of the Year Discussion Thread

Moderators: penbeast0, trex_8063, PaulieWal, Doctor MJ, Clyde Frazier

HeartBreakKid
RealGM
Posts: 22,395
And1: 18,813
Joined: Mar 08, 2012
     

Re: 2019-2020 Player of the Year Discussion Thread 

Post#61 » by HeartBreakKid » Sun Feb 16, 2020 8:31 pm

E-Balla wrote:This whole 3 point defense is luck crap is annoying. Are defensive midrange shooting percentages luck? Do we need to adjust for that?

Mid range shots are more contested than 3 point shots.
User avatar
MyUniBroDavis
General Manager
Posts: 7,730
And1: 4,844
Joined: Jan 14, 2013
   

Re: 2019-2020 Player of the Year Discussion Thread 

Post#62 » by MyUniBroDavis » Mon Feb 17, 2020 1:54 am

bondom34 wrote:
MyUniBroDavis wrote:
bondom34 wrote:I don't think so, think it's just luck adjusted on/off w/ a box prior.


Wow well screw that lmao

On the luck adjusted rapm, is it the same data as others? Kobes 2013 wasnt third in offensive rapm in the other one I found

Also one thing about luck adjustments that bother me, it assumes that a player cant effect team three point percentage and idk if I agree with that, idk if it adjusts for openness but even in terms of delivering it to the shot pocket and stuff like that, ditto on defense in terms of a teams defensive scheme possibly changing with certain players being on or off the court

If it doesnt adjust for openness (and idt it using any tracking data, altho tracking data itself is flawed, also qrong) then idk about that part of luck adjustment (i totally like the ft portion of it though).

Like giannis for example i feel isnt the best at kicking it to that shot pocket, and in terms of mvp giannis luck adjustment boosts him an insane amount

I'd disagree, there have been a few studies on it and its been a lot more luck effected than anything else. I don't think Nikola Vucevic is making an opponent shoot 10% worse on 3s some seasons for example. I also know offhand this year Kanter is benefiting a ton from it, and Schroder's DRAPM is really inflated. Last I checked he's been well ahead of CP3 in some metrics on defense because there's about a 10% gap in 3 point shooting (and a big FT% difference) depending on who's on court.

Also tracking data on 3s has been.....very spotty? at best. Partnow's already said basically the data on players being "guarded by" someone has been off. Something akin to if you're being guarded by Robert Covington but drive past him and PJ Tucker contests the shot, NBA.com classifies it as being guarded by Tucker, even though he wasn't at fault. The only real spot that data works is maybe for rim protection, but:

https://fansided.com/2018/01/29/nylon-calculus-3-point-team-defense-closure/

Let’s get into the meat of the issue. The core problem is that most 3-pointers are open shots, and open shots inherently can’t be influenced by the defense. They’re open. It’s the same reason why we shouldn’t use individual defensive field goal percentage — the open shot data tells you nothing. Likewise, rim protection stats are actually useful because most of those shots are not open and, thus, can be influenced by the defender. It’s a basic law of shot defense: the closer you are, the more you can influence the shot.



And the original idea was from Pomeroy (college ball, but the point holds), which I'd generally agree w/ too:

https://theathletic.com/720048/2018/12/18/kenpom-shooting-and-defending-the-3-is-random-business-i-e-relax-its-early/



We can get a little more complex by constructing a regression model to predict a team’s 3-point defense the rest of the season based on its first nine games. Taking the last three seasons collectively, the result is that early season 3-point defense is not completely inconsequential, but nearly so. For every 1 percent that a team’s early season 3-point defense improves, its defense the rest of the season improves by 0.10 percent. Do not read that as 10 percent, but one-10th of a percent. Not nothing, but close to it. To illustrate, here’s a plot of a team’s 3-point defense over the first nine games and then the remainder of the season. It’s not a completely random scattering of data, but it’s close. There’s just very little predictive value in a team’s early 3-point defense numbers.



Spoiler:
Image


This isn’t to say that 3-point defense doesn’t exist. If a team made its only priority defending the rim, say, by keeping all five defenders in the paint on every possession, opponents would shoot an impressive percentage from beyond the arc. But assuming a realistic defensive approach, shooters are going to avoid taking 3s with somebody draped all over them. Thus, defenses have much more influence on whether a 3 will be taken at all than whether a shot goes in once it’s airborne.The results are very similar. In fact, there’s slightly more utility to the 2-point defense data. Play a team with a very good 2-point defense and you are liable to struggle on 3-point shooting as well. Likewise, if a team plays an opponent that has allowed a high 2-point percentage, its 3-point shooting is also likely to improve. In order to get an idea of how a team might shoot from 3 on a particular night, its opponent’s 2-point defense is a slightly more useful indicator than its opponent’s 3-point defense. In part, that’s because 3-point defense is subjected to more randomness.


Basically it's not all luck, but there's a fair amount at times, and going off of no luck is assuming that contesting a 3 really has that big an effect. Know I've gone back and forth w/ a few at times on this a while back, but basically I think its a lot more luck than some do, and don't at all trust or believe most of the tracking defense other than maybe for "wide open" 3s or rim protection, other than that it's basically useless because these guys can shoot if you're still somewhat close to them.


Nah i def agree that luck does play a part in it, like i remember kawhi got absolutely schmacked because teams were hella hot from three with him in the floor. Its just difficult because a players presence can effect how good a team are at guarding the three, but tracking fata just isnt reliable enough to adjust for it. Wasnt drapm without luck adjustment more predictive/accurate than drapm with lucj adjustment anyw?
iggymcfrack wrote: I have Bird #19 and Kobe #20 on my all-time list and both guys will probably get passed by Jokic by the end of this season.


^^^^ posted January 8 2023 :banghead: :banghead:
User avatar
bondom34
Retired Mod
Retired Mod
Posts: 66,580
And1: 50,199
Joined: Mar 01, 2013

Re: 2019-2020 Player of the Year Discussion Thread 

Post#63 » by bondom34 » Mon Feb 17, 2020 2:02 am

MyUniBroDavis wrote:
bondom34 wrote:
MyUniBroDavis wrote:
Wow well screw that lmao

On the luck adjusted rapm, is it the same data as others? Kobes 2013 wasnt third in offensive rapm in the other one I found

Also one thing about luck adjustments that bother me, it assumes that a player cant effect team three point percentage and idk if I agree with that, idk if it adjusts for openness but even in terms of delivering it to the shot pocket and stuff like that, ditto on defense in terms of a teams defensive scheme possibly changing with certain players being on or off the court

If it doesnt adjust for openness (and idt it using any tracking data, altho tracking data itself is flawed, also qrong) then idk about that part of luck adjustment (i totally like the ft portion of it though).

Like giannis for example i feel isnt the best at kicking it to that shot pocket, and in terms of mvp giannis luck adjustment boosts him an insane amount

I'd disagree, there have been a few studies on it and its been a lot more luck effected than anything else. I don't think Nikola Vucevic is making an opponent shoot 10% worse on 3s some seasons for example. I also know offhand this year Kanter is benefiting a ton from it, and Schroder's DRAPM is really inflated. Last I checked he's been well ahead of CP3 in some metrics on defense because there's about a 10% gap in 3 point shooting (and a big FT% difference) depending on who's on court.

Also tracking data on 3s has been.....very spotty? at best. Partnow's already said basically the data on players being "guarded by" someone has been off. Something akin to if you're being guarded by Robert Covington but drive past him and PJ Tucker contests the shot, NBA.com classifies it as being guarded by Tucker, even though he wasn't at fault. The only real spot that data works is maybe for rim protection, but:

https://fansided.com/2018/01/29/nylon-calculus-3-point-team-defense-closure/

Let’s get into the meat of the issue. The core problem is that most 3-pointers are open shots, and open shots inherently can’t be influenced by the defense. They’re open. It’s the same reason why we shouldn’t use individual defensive field goal percentage — the open shot data tells you nothing. Likewise, rim protection stats are actually useful because most of those shots are not open and, thus, can be influenced by the defender. It’s a basic law of shot defense: the closer you are, the more you can influence the shot.



And the original idea was from Pomeroy (college ball, but the point holds), which I'd generally agree w/ too:

https://theathletic.com/720048/2018/12/18/kenpom-shooting-and-defending-the-3-is-random-business-i-e-relax-its-early/



We can get a little more complex by constructing a regression model to predict a team’s 3-point defense the rest of the season based on its first nine games. Taking the last three seasons collectively, the result is that early season 3-point defense is not completely inconsequential, but nearly so. For every 1 percent that a team’s early season 3-point defense improves, its defense the rest of the season improves by 0.10 percent. Do not read that as 10 percent, but one-10th of a percent. Not nothing, but close to it. To illustrate, here’s a plot of a team’s 3-point defense over the first nine games and then the remainder of the season. It’s not a completely random scattering of data, but it’s close. There’s just very little predictive value in a team’s early 3-point defense numbers.



Spoiler:
Image


This isn’t to say that 3-point defense doesn’t exist. If a team made its only priority defending the rim, say, by keeping all five defenders in the paint on every possession, opponents would shoot an impressive percentage from beyond the arc. But assuming a realistic defensive approach, shooters are going to avoid taking 3s with somebody draped all over them. Thus, defenses have much more influence on whether a 3 will be taken at all than whether a shot goes in once it’s airborne.The results are very similar. In fact, there’s slightly more utility to the 2-point defense data. Play a team with a very good 2-point defense and you are liable to struggle on 3-point shooting as well. Likewise, if a team plays an opponent that has allowed a high 2-point percentage, its 3-point shooting is also likely to improve. In order to get an idea of how a team might shoot from 3 on a particular night, its opponent’s 2-point defense is a slightly more useful indicator than its opponent’s 3-point defense. In part, that’s because 3-point defense is subjected to more randomness.


Basically it's not all luck, but there's a fair amount at times, and going off of no luck is assuming that contesting a 3 really has that big an effect. Know I've gone back and forth w/ a few at times on this a while back, but basically I think its a lot more luck than some do, and don't at all trust or believe most of the tracking defense other than maybe for "wide open" 3s or rim protection, other than that it's basically useless because these guys can shoot if you're still somewhat close to them.


Nah i def agree that luck does play a part in it, like i remember kawhi got absolutely schmacked because teams were hella hot from three with him in the floor. Its just difficult because a players presence can effect how good a team are at guarding the three, but tracking fata just isnt reliable enough to adjust for it. Wasnt drapm without luck adjustment more predictive/accurate than drapm with lucj adjustment anyw?

I don't remember anything being studied on it. From all I've read the only real way to drastically effect it is to just not allow shots, most threes are fairly open, and given that there's a lot of variance, Schroeder this year has been a big benefactor. Same for Kanter I believe and both take a hot in luck adjustment. Ft shooting too, but to a lesser extent.
MyUniBroDavis wrote: he was like YALL PEOPLE WHO DOUBT ME WILL SEE YALLS STATS ARE WRONG I HAVE THE BIG BRAIN PLAYS MUCHO NASTY BIG BRAIN BIG CHUNGUS BRAIN YOU BOYS ON UR BBALL REFERENCE NO UNDERSTANDO
User avatar
E-Balla
RealGM
Posts: 35,089
And1: 24,396
Joined: Dec 19, 2012
Location: The Poster Formerly Known As The Gotham City Pantalones
   

Re: 2019-2020 Player of the Year Discussion Thread 

Post#64 » by E-Balla » Mon Feb 17, 2020 3:23 am

HeartBreakKid wrote:
E-Balla wrote:This whole 3 point defense is luck crap is annoying. Are defensive midrange shooting percentages luck? Do we need to adjust for that?

Mid range shots are more contested than 3 point shots.

Yeah but contesting 3 point shots lowers the FG% of them way more than contesting midrange shots does.

The highest team by midrange percentage on open shots is 49% while the lowest is 37% with a 43% median. Meanwhile the highest percentage on tight shots is 46% while the lowest is 33% and the median is 38%. On wide open shots the highest percentage is 52%, the lowest is 39%, and the median is 45%.

For 3 point shots the highest percentage on open looks is 39% while the lowest is 29% with a 34% median. The highest percentage on tight shots is 35% the lowest is 25% and the median is 31%. On wide open shots the highest percentage is 46%, the lowest is 28%, and the median is 38%.

Seems to me like the raw percentages are effected the same which means 3 point efficiency is effected 50% more by being guarded.
User avatar
bondom34
Retired Mod
Retired Mod
Posts: 66,580
And1: 50,199
Joined: Mar 01, 2013

Re: 2019-2020 Player of the Year Discussion Thread 

Post#65 » by bondom34 » Mon Feb 17, 2020 3:26 am

LARAPM would adjust midrange too. It would do it for any shot. And generally, NBA.com's definitions are pretty shaky. "Open" for them on a 3 is really pretty much a normal 3, and tight ones aren't taken often in general iirc. Partnow did a really good explainer on this a few months back.
MyUniBroDavis wrote: he was like YALL PEOPLE WHO DOUBT ME WILL SEE YALLS STATS ARE WRONG I HAVE THE BIG BRAIN PLAYS MUCHO NASTY BIG BRAIN BIG CHUNGUS BRAIN YOU BOYS ON UR BBALL REFERENCE NO UNDERSTANDO
Statlanta
RealGM
Posts: 12,448
And1: 9,097
Joined: Mar 06, 2016

Re: 2019-2020 Player of the Year Discussion Thread 

Post#66 » by Statlanta » Mon Feb 17, 2020 8:09 pm

70sFan wrote:The only way to rate defense is to watch tons of footage with paying attention to everything and trying to find good and bad tendencies. It's still very flawed because everyone has subjective feeling on what is positive and what not, but I think that you can't do that any better with defense in basketball.


Just watch the last 5 minutes of a 4th quarter in any game, not a big sample size and horrendous for statistical analysis but enough to get the job done. Usually when both teams are giving it there best on both sides.

People have been calling Davis a clutch defender the entire season. I'd take that over his bad DRPM numbers, which are muddled because of the varying levels of effort teams put in, his own varying effort, and his transition to a new team and new role.

I think DRPM personally isn't bad when you account for role, ex: LBJ is consistently a good defender in his prime(2009-2016) and with a defensive coach, 17 years of experience and 3 of the top shot blockers in the last 7 years, and being the only big wing level defender on his team its not hard to see why his role's numbers are high compared to say Gobert who is dealing with worse defensive personnel and a different role.

But LBJ is not the best defensive player on his team. His team just requires his role more because the personnel Bradley/KCP(too small) and Kuzma(too inexperienced) cannot fulfill that role.
East #1 Draft Picks: Fultz, Banchero, Wiggins, Cuninigham
West #1 Draft Picks: Edwards, WIlliamson, Ayton, Towns
User avatar
bondom34
Retired Mod
Retired Mod
Posts: 66,580
And1: 50,199
Joined: Mar 01, 2013

Re: 2019-2020 Player of the Year Discussion Thread 

Post#67 » by bondom34 » Mon Feb 17, 2020 8:15 pm

I've always kind of felt eye test can slant a view though as well. I definitely agree there's a need for it in some fashion, and an understanding of what you're looking at in numbers too, but there's way too much to defense (off ball especially) to really just get it in eye test for me.
MyUniBroDavis wrote: he was like YALL PEOPLE WHO DOUBT ME WILL SEE YALLS STATS ARE WRONG I HAVE THE BIG BRAIN PLAYS MUCHO NASTY BIG BRAIN BIG CHUNGUS BRAIN YOU BOYS ON UR BBALL REFERENCE NO UNDERSTANDO
User avatar
Baski
Lead Assistant
Posts: 4,532
And1: 3,950
Joined: Feb 09, 2017
   

Re: 2019-2020 Player of the Year Discussion Thread 

Post#68 » by Baski » Sun Feb 23, 2020 8:23 am

bondom34 wrote:
MyUniBroDavis wrote:Wait whats the difference between luck adjusted rapm and pipm?

Pretty sure PIPM has some box score stuff in it too, and player tracking. I'm not a fan of the tracking data in anything (I don't like the new RPM or RAPTOR much at all, definitely defensively). LA-RAPM is RAPM with these changes from Ryan Davis's twitter (and hes a good follow too):

Read on Twitter

I'm a little late here but I wanna ask, and I'm probably not understanding that statement well but bear with me: if you change opponents PPS to their career average, doesn't that completely eliminate the effect of the defense and other factors?
And wouldn't that mean the player would always look good against bad shooters and bad against good shooters?
User avatar
bondom34
Retired Mod
Retired Mod
Posts: 66,580
And1: 50,199
Joined: Mar 01, 2013

Re: 2019-2020 Player of the Year Discussion Thread 

Post#69 » by bondom34 » Sun Feb 23, 2020 8:31 am

Baski wrote:
bondom34 wrote:
MyUniBroDavis wrote:Wait whats the difference between luck adjusted rapm and pipm?

Pretty sure PIPM has some box score stuff in it too, and player tracking. I'm not a fan of the tracking data in anything (I don't like the new RPM or RAPTOR much at all, definitely defensively). LA-RAPM is RAPM with these changes from Ryan Davis's twitter (and hes a good follow too):

Read on Twitter

I'm a little late here but I wanna ask, and I'm probably not understanding that statement well but bear with me: if you change opponents PPS to their career average, doesn't that completely eliminate the effect of the defense and other factors?
And wouldn't that mean the player would always look good against bad shooters and bad against good shooters?

It's only done on 3s. The "luck" part is the idea that 3 point shooting is largely luck/variation based and everyone would shoot about their average from 3 on all shots.
MyUniBroDavis wrote: he was like YALL PEOPLE WHO DOUBT ME WILL SEE YALLS STATS ARE WRONG I HAVE THE BIG BRAIN PLAYS MUCHO NASTY BIG BRAIN BIG CHUNGUS BRAIN YOU BOYS ON UR BBALL REFERENCE NO UNDERSTANDO
User avatar
Baski
Lead Assistant
Posts: 4,532
And1: 3,950
Joined: Feb 09, 2017
   

Re: 2019-2020 Player of the Year Discussion Thread 

Post#70 » by Baski » Sun Feb 23, 2020 8:44 am

bondom34 wrote:
Baski wrote:
bondom34 wrote:Pretty sure PIPM has some box score stuff in it too, and player tracking. I'm not a fan of the tracking data in anything (I don't like the new RPM or RAPTOR much at all, definitely defensively). LA-RAPM is RAPM with these changes from Ryan Davis's twitter (and hes a good follow too):

Read on Twitter

I'm a little late here but I wanna ask, and I'm probably not understanding that statement well but bear with me: if you change opponents PPS to their career average, doesn't that completely eliminate the effect of the defense and other factors?
And wouldn't that mean the player would always look good against bad shooters and bad against good shooters?

It's only done on 3s. The "luck" part is the idea that 3 point shooting is largely luck/variation based and everyone would shoot about their average from 3 on all shots.

Well yeah but is my understanding of it right? That seems to really take out a lot of context of what happens in-game.
User avatar
bondom34
Retired Mod
Retired Mod
Posts: 66,580
And1: 50,199
Joined: Mar 01, 2013

Re: 2019-2020 Player of the Year Discussion Thread 

Post#71 » by bondom34 » Sun Feb 23, 2020 6:18 pm

Baski wrote:
bondom34 wrote:
Baski wrote:I'm a little late here but I wanna ask, and I'm probably not understanding that statement well but bear with me: if you change opponents PPS to their career average, doesn't that completely eliminate the effect of the defense and other factors?
And wouldn't that mean the player would always look good against bad shooters and bad against good shooters?

It's only done on 3s. The "luck" part is the idea that 3 point shooting is largely luck/variation based and everyone would shoot about their average from 3 on all shots.

Well yeah but is my understanding of it right? That seems to really take out a lot of context of what happens in-game.

Not sure, the idea is that 3 point shooting is largely not dependent on how contested the shot is (there have been studies done), so removing that variance and 3 point variance.

Edit: Nylon Calculus looked into it years ago iirc. And Ken Pomeroy has looked into it in NCAA too.

So like teams shooting 10% worse on 3s while Enes Kanter is on court isn't really saying Kanter is a good defender. Similar happens with some perimeter guys (Schroder was getting this and I think still is).
MyUniBroDavis wrote: he was like YALL PEOPLE WHO DOUBT ME WILL SEE YALLS STATS ARE WRONG I HAVE THE BIG BRAIN PLAYS MUCHO NASTY BIG BRAIN BIG CHUNGUS BRAIN YOU BOYS ON UR BBALL REFERENCE NO UNDERSTANDO
Dr Spaceman
General Manager
Posts: 8,575
And1: 11,205
Joined: Jan 16, 2013
   

Re: 2019-2020 Player of the Year Discussion Thread 

Post#72 » by Dr Spaceman » Sun Feb 23, 2020 6:44 pm

bondom34 wrote:
Baski wrote:
bondom34 wrote:It's only done on 3s. The "luck" part is the idea that 3 point shooting is largely luck/variation based and everyone would shoot about their average from 3 on all shots.

Well yeah but is my understanding of it right? That seems to really take out a lot of context of what happens in-game.

Not sure, the idea is that 3 point shooting is largely not dependent on how contested the shot is (there have been studies done), so removing that variance and 3 point variance.

Edit: Nylon Calculus looked into it years ago iirc. And Ken Pomeroy has looked into it in NCAA too.

So like teams shooting 10% worse on 3s while Enes Kanter is on court isn't really saying Kanter is a good defender. Similar happens with some perimeter guys (Schroder was getting this and I think still is).


Then what is it dependent on? Just because a few statisticians have decided the answer is “pure luck” doesn’t mean they’re right or that we should accept this. Throwing out a bunch of useful data because you don’t know what it’s useful for makes no sense.

There are easy , intuitive hypotheses for how defenses can control opponent three point shooting.
“I’m not the fastest guy on the court, but I can dictate when the race begins.”
User avatar
bondom34
Retired Mod
Retired Mod
Posts: 66,580
And1: 50,199
Joined: Mar 01, 2013

Re: 2019-2020 Player of the Year Discussion Thread 

Post#73 » by bondom34 » Sun Feb 23, 2020 7:07 pm

Dr Spaceman wrote:
bondom34 wrote:
Baski wrote:Well yeah but is my understanding of it right? That seems to really take out a lot of context of what happens in-game.

Not sure, the idea is that 3 point shooting is largely not dependent on how contested the shot is (there have been studies done), so removing that variance and 3 point variance.

Edit: Nylon Calculus looked into it years ago iirc. And Ken Pomeroy has looked into it in NCAA too.

So like teams shooting 10% worse on 3s while Enes Kanter is on court isn't really saying Kanter is a good defender. Similar happens with some perimeter guys (Schroder was getting this and I think still is).


Then what is it dependent on? Just because a few statisticians have decided the answer is “pure luck” doesn’t mean they’re right or that we should accept this. Throwing out a bunch of useful data because you don’t know what it’s useful for makes no sense.

There are easy , intuitive hypotheses for how defenses can control opponent three point shooting.

More on volume and on who's shooting. Which can be controlled.

Like stated, I don't think Schroder is why opponents are shooting 29% on 3s while he's on court w/o Chris Paul, while they're suddenly shooting 40% when Paul is on w/o Schroder.
MyUniBroDavis wrote: he was like YALL PEOPLE WHO DOUBT ME WILL SEE YALLS STATS ARE WRONG I HAVE THE BIG BRAIN PLAYS MUCHO NASTY BIG BRAIN BIG CHUNGUS BRAIN YOU BOYS ON UR BBALL REFERENCE NO UNDERSTANDO
User avatar
Baski
Lead Assistant
Posts: 4,532
And1: 3,950
Joined: Feb 09, 2017
   

Re: 2019-2020 Player of the Year Discussion Thread 

Post#74 » by Baski » Sun Feb 23, 2020 7:31 pm

bondom34 wrote:
Dr Spaceman wrote:
bondom34 wrote:Not sure, the idea is that 3 point shooting is largely not dependent on how contested the shot is (there have been studies done), so removing that variance and 3 point variance.

Edit: Nylon Calculus looked into it years ago iirc. And Ken Pomeroy has looked into it in NCAA too.

So like teams shooting 10% worse on 3s while Enes Kanter is on court isn't really saying Kanter is a good defender. Similar happens with some perimeter guys (Schroder was getting this and I think still is).


Then what is it dependent on? Just because a few statisticians have decided the answer is “pure luck” doesn’t mean they’re right or that we should accept this. Throwing out a bunch of useful data because you don’t know what it’s useful for makes no sense.

There are easy , intuitive hypotheses for how defenses can control opponent three point shooting.

More on volume and on who's shooting. Which can be controlled.

Like stated, I don't think Schroder is why opponents are shooting 29% on 3s while he's on court w/o Chris Paul, while they're suddenly shooting 40% when Paul is on w/o Schroder.

Something about declaring it "luck" just feels off. And even so, the solution to "luck" being discarding it and then replacing it with past data just screams wrong to me. Too many questions immediately pop up for it to be accepted so easily.
ShotCreator
Analyst
Posts: 3,479
And1: 2,328
Joined: May 18, 2014
Location: CF
     

Re: 2019-2020 Player of the Year Discussion Thread 

Post#75 » by ShotCreator » Sun Feb 23, 2020 7:41 pm

bondom34 wrote:
Dr Spaceman wrote:
bondom34 wrote:Not sure, the idea is that 3 point shooting is largely not dependent on how contested the shot is (there have been studies done), so removing that variance and 3 point variance.

Edit: Nylon Calculus looked into it years ago iirc. And Ken Pomeroy has looked into it in NCAA too.

So like teams shooting 10% worse on 3s while Enes Kanter is on court isn't really saying Kanter is a good defender. Similar happens with some perimeter guys (Schroder was getting this and I think still is).


Then what is it dependent on? Just because a few statisticians have decided the answer is “pure luck” doesn’t mean they’re right or that we should accept this. Throwing out a bunch of useful data because you don’t know what it’s useful for makes no sense.

There are easy , intuitive hypotheses for how defenses can control opponent three point shooting.

More on volume and on who's shooting. Which can be controlled.

Like stated, I don't think Schroder is why opponents are shooting 29% on 3s while he's on court w/o Chris Paul, while they're suddenly shooting 40% when Paul is on w/o Schroder.

CP3 hasn’t contested 3’s all year and is helping off shooters a lot with mixed results.

He also hasn’t fought through and around a screens much other than very early on and like January.

Meh defensive year for him. Not bad but very meh.
Swinging for the fences.
User avatar
bondom34
Retired Mod
Retired Mod
Posts: 66,580
And1: 50,199
Joined: Mar 01, 2013

Re: 2019-2020 Player of the Year Discussion Thread 

Post#76 » by bondom34 » Sun Feb 23, 2020 7:47 pm

Baski wrote:
bondom34 wrote:
Dr Spaceman wrote:
Then what is it dependent on? Just because a few statisticians have decided the answer is “pure luck” doesn’t mean they’re right or that we should accept this. Throwing out a bunch of useful data because you don’t know what it’s useful for makes no sense.

There are easy , intuitive hypotheses for how defenses can control opponent three point shooting.

More on volume and on who's shooting. Which can be controlled.

Like stated, I don't think Schroder is why opponents are shooting 29% on 3s while he's on court w/o Chris Paul, while they're suddenly shooting 40% when Paul is on w/o Schroder.

Something about declaring it "luck" just feels off. And even so, the solution to "luck" being discarding it and then replacing it with past data just screams wrong to me. Too many questions immediately pop up for it to be accepted so easily.

Calling it "luck" is just wording, it's basically "variance adjusted". I mean, it's been looked into, and the tracking data has backed it up a fair amount, and teams generally revert to the mean. I'd trust it a good bit over other APM's personally, but if others don't that's fine too.

ShotCreator wrote:CP3 hasn’t contested 3’s all year and is helping off shooters a lot with mixed results.

He also hasn’t fought through and around a screens much other than very early on and like January.

Meh defensive year for him. Not bad but very meh.

Maybe, but I also don't think Schroder is really like a DPOY level point guard.

It does help more for bigs especially, ie Kanter going from 123rd in DRAPM to 440 in LA. He's not really the reason for opponents shooting poorly from distance.
MyUniBroDavis wrote: he was like YALL PEOPLE WHO DOUBT ME WILL SEE YALLS STATS ARE WRONG I HAVE THE BIG BRAIN PLAYS MUCHO NASTY BIG BRAIN BIG CHUNGUS BRAIN YOU BOYS ON UR BBALL REFERENCE NO UNDERSTANDO
User avatar
Baski
Lead Assistant
Posts: 4,532
And1: 3,950
Joined: Feb 09, 2017
   

Re: 2019-2020 Player of the Year Discussion Thread 

Post#77 » by Baski » Mon Feb 24, 2020 7:05 am

bondom34 wrote:Calling it "luck" is just wording, it's basically "variance adjusted". I mean, it's been looked into, and the tracking data has backed it up a fair amount, and teams generally revert to the mean. I'd trust it a good bit over other APM's personally, but if others don't that's fine too.

The point remains the same regardless of wording though. Whatever they're doing is basically saying "even though the ball went into the net it shouldn't have so it didn't" and the converse.
I'm glad this is getting significant pushback. Can't even begin to wrap my head around it as a valid approach. But yeah agree to disagree
User avatar
bondom34
Retired Mod
Retired Mod
Posts: 66,580
And1: 50,199
Joined: Mar 01, 2013

Re: 2019-2020 Player of the Year Discussion Thread 

Post#78 » by bondom34 » Mon Feb 24, 2020 7:08 am

Baski wrote:
bondom34 wrote:Calling it "luck" is just wording, it's basically "variance adjusted". I mean, it's been looked into, and the tracking data has backed it up a fair amount, and teams generally revert to the mean. I'd trust it a good bit over other APM's personally, but if others don't that's fine too.

The point remains the same regardless of wording though. Whatever they're doing is basically saying "even though the ball went into the net it shouldn't have so it didn't" and the converse.
I'm glad this is getting significant pushback. Can't even begin to wrap my head around it as a valid approach. But yeah agree to disagree

That's not really what it's saying, but OK. There's actually information behind it, and it's gotten quite a lot of support including that from people who've worked for teams (I know Partnow is a proponent of it from twitter and his articles), 3 point shooting doesn't depend as much on the actual defense of the shot, but (more) on volume of shots. After reading and knowing it's good enough for people who work for teams, I'd say it makes sense.

That plus there's actual data behind it, so the pushback is sort of odd in that sense to me. Bigs especially aren't effecting opponent 3 point shooting, its correcting for that and tbh I'm not too sure why that seems odd, a guy like Kanter or Vuc isn't responsible for an opponent's 3 point shooting at all, and even perimeter defense on it has shown to generally revert to near average. It's pretty logical, not really sure why it's hard to understand in some way. Agree to disagree.

And to be clear, I'd see it as an upgraded RAPM, which I wouldn't use as some all knowing measure either, it's another tool that I'd take over other APMs but not as a final "rank" or anything.
MyUniBroDavis wrote: he was like YALL PEOPLE WHO DOUBT ME WILL SEE YALLS STATS ARE WRONG I HAVE THE BIG BRAIN PLAYS MUCHO NASTY BIG BRAIN BIG CHUNGUS BRAIN YOU BOYS ON UR BBALL REFERENCE NO UNDERSTANDO
User avatar
Baski
Lead Assistant
Posts: 4,532
And1: 3,950
Joined: Feb 09, 2017
   

Re: 2019-2020 Player of the Year Discussion Thread 

Post#79 » by Baski » Mon Feb 24, 2020 7:15 am

bondom34 wrote:
Baski wrote:
bondom34 wrote:Calling it "luck" is just wording, it's basically "variance adjusted". I mean, it's been looked into, and the tracking data has backed it up a fair amount, and teams generally revert to the mean. I'd trust it a good bit over other APM's personally, but if others don't that's fine too.

The point remains the same regardless of wording though. Whatever they're doing is basically saying "even though the ball went into the net it shouldn't have so it didn't" and the converse.
I'm glad this is getting significant pushback. Can't even begin to wrap my head around it as a valid approach. But yeah agree to disagree

That's not really what it's saying, but OK. There's actually information behind it, and it's gotten quite a lot of support including that from people who've worked for teams (I know Partnow is a proponent of it from twitter and his articles), 3 point shooting doesn't depend as much on the actual defense of the shot, but (more) on volume of shots. After reading and knowing it's good enough for people who work for teams, I'd say it makes sense.

That plus there's actual data behind it, so the pushback is sort of odd in that sense to me. Bigs especially aren't effecting opponent 3 point shooting, its correcting for that. Agree to disagree.

Then maybe I'm not understanding it right. Lemme lay it out how I understand it: Schroeder's opponents shoot 10% lower than normal when he's on and CP3's shoot 10% higher when he's on. So LA tries to raise Schroeder's opp. Shooting % and lower CP3's. Do I have that right?

Also, I do think there's validity in the claim that high variance can overrate some players and underrate other in terms of how they affect opponents Shooting. I doubt there'd be any pushback to that particular point.
I can't speak for everyone but my issue is with this particular method of correcting that, or that it even has to be corrected in the first place.
User avatar
bondom34
Retired Mod
Retired Mod
Posts: 66,580
And1: 50,199
Joined: Mar 01, 2013

Re: 2019-2020 Player of the Year Discussion Thread 

Post#80 » by bondom34 » Mon Feb 24, 2020 7:18 am

Baski wrote:
bondom34 wrote:
Baski wrote:The point remains the same regardless of wording though. Whatever they're doing is basically saying "even though the ball went into the net it shouldn't have so it didn't" and the converse.
I'm glad this is getting significant pushback. Can't even begin to wrap my head around it as a valid approach. But yeah agree to disagree

That's not really what it's saying, but OK. There's actually information behind it, and it's gotten quite a lot of support including that from people who've worked for teams (I know Partnow is a proponent of it from twitter and his articles), 3 point shooting doesn't depend as much on the actual defense of the shot, but (more) on volume of shots. After reading and knowing it's good enough for people who work for teams, I'd say it makes sense.

That plus there's actual data behind it, so the pushback is sort of odd in that sense to me. Bigs especially aren't effecting opponent 3 point shooting, its correcting for that. Agree to disagree.

Then maybe I'm not understanding it right. Lemme lay it out how I understand it: Schroeder's opponents shoot 10% lower than normal when he's on and CP3's shoot 10% higher when he's on. So LA tries to raise Schroeder's opp. Shooting % and lower CP3's. Do I have that right?

See my edit, but mostly. There have been multiple studies that opponent shooting regresses close to the mean. This corrects for that. 3 point defense is more about volume of shots given up (and its even more a correction for a big man, like using the Kanter example). Enes would have no effect at all on opponent 3 point shooting and it shows a ton in DRAPM. It's not some decisive rank just like RPM/PIPM/BPM/anything else, but I'd see it as an upgrade to regular RAPM.

Edit: This was Partnow's original piece I think.

https://fansided.com/2015/02/06/defense-variance-luck/

I don't just blindly accept it, read the logic behind it and to me it makes sense, I'll agree to disagree though.
MyUniBroDavis wrote: he was like YALL PEOPLE WHO DOUBT ME WILL SEE YALLS STATS ARE WRONG I HAVE THE BIG BRAIN PLAYS MUCHO NASTY BIG BRAIN BIG CHUNGUS BRAIN YOU BOYS ON UR BBALL REFERENCE NO UNDERSTANDO

Return to Player Comparisons