RealGM Top 100 List #9

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

tsherkin
Forum Mod - Raptors
Forum Mod - Raptors
Posts: 89,716
And1: 29,664
Joined: Oct 14, 2003
 

Re: RealGM Top 100 List #9 

Post#181 » by tsherkin » Mon Jul 21, 2014 2:56 am

Warspite wrote:KG= GOAT 3rd option role player.

"I love KG so much but cant use video or stats to make him look great. What should I do? i know Ill invent a bunch of adv stats that make him look great."


This is a non-starter post with obvious scoring bias that fails to account for various elements of the game and relative context. Within the bounds of this project, it's a pretty lazy, insulting and low-quality post and that's strange coming from you, W. It's cool to disagree, but show it, don't denigrate these other avenues of analysis, especially without providing something substantive as a counter.
ceiling raiser
Lead Assistant
Posts: 4,531
And1: 3,754
Joined: Jan 27, 2013

Re: RealGM Top 100 List #9 

Post#182 » by ceiling raiser » Mon Jul 21, 2014 3:02 am

Baller2014 wrote:
john248 wrote:Looks like either Bird or Olajuwon will go here. Almost safe to say whoever isn't voted in now will likely be voted in at 10 or 11 at the latest since these 2 are the last of the truly elite players. #11 and #12 will bring a flood of players.


Yeh, almost every single Bird/Hakeem voter has said "this is between Bird and Hakeem for me". Whoever wins, I think we should adopt a lax approach about the discussion in the next thread, and let people talk about candidates up for nomination at the #11 spot a little bit. Otherwise we're going to get 2 days of waiting for Bird/Hakeem to officially win by 20+ votes, and it's going to get deadly dull having to talk about Hakeem/Bird when 90% of voters agree he belongs there.

To be fair though, I think a lot of good stuff is being posted on Bird/Hakeem here.

With regards to the next thread being a given, that's a possibility, but maybe not. Last thread Hakeem made it to a relatively close runoff with Magic, and I think he's down in the voting in this thread. Anything can happen.

The teens aren't as exciting probably, though I am interested in reading posts about a few players. Would love to see some stuff posted on Robinson, Stockton, Barry, Ewing, Baylor. No guarantee they all make it in then (or that any of them do), but I think there's going to be a lot of new research about them coming up in those threads. :)
Now that's the difference between first and last place.
HeartBreakKid
RealGM
Posts: 22,395
And1: 18,823
Joined: Mar 08, 2012
     

Re: RealGM Top 100 List #9 

Post#183 » by HeartBreakKid » Mon Jul 21, 2014 3:05 am

It's not a given that the winner of Bird/Hakeem will be the automatic #10 spot. Let's say Bird wins the head to head, the people who voted Bird might not automatically think Hakeem is worth the #10 spot, so they could vote for Bryant, Robertson etc.

It's already happen a few times in this project.
User avatar
Moonbeam
Forum Mod - Blazers
Forum Mod - Blazers
Posts: 10,214
And1: 5,062
Joined: Feb 21, 2009
Location: Sydney, Australia
     

Re: RealGM Top 100 List #9 

Post#184 » by Moonbeam » Mon Jul 21, 2014 3:12 am

fpliii wrote:
Warspite wrote:KG= GOAT 3rd option role player.

"I love KG so much but cant use video or stats to make him look great. What should I do? i know Ill invent a bunch of adv stats that make him look great."

I think that's incredibly disrespectful to KG and the analysis that some posters have done. Ridge regression is a proven, peer-reviewed mathematical method, and even aside from RAPM, KG excels in other non-box score metrics (APM in particular, but he holds up well in WOWY studies as well). It's your call to choose to subscribe to them or not, but to dismiss the work that's been done like that is unfortuante IMO.

drza had a great breakdown, mixing quantitative WITH qualitative a couple of threads ago:

viewtopic.php?p=40580241#p40580241


Very cool to see ridge regression applied to APM - hadn't seen that before! Do you know if there is a version that incorporates a different type of penalty, like LASSO or elastic net?
Baller2014
Banned User
Posts: 2,049
And1: 519
Joined: May 22, 2014
Location: No further than the thickness of a shadow
     

Re: RealGM Top 100 List #9 

Post#185 » by Baller2014 » Mon Jul 21, 2014 3:12 am

HeartBreakKid wrote:It's not a given that the winner of Bird/Hakeem will be the automatic #10 spot. Let's say Bird wins the head to head, the people who voted Bird might not automatically think Hakeem is worth the #10 spot, so they could vote for Bryant, Robertson etc.

It's already happen a few times in this project.


Yeh, except I've been carefully reading the posts of each voter, and the overwhelming majority of Bird/Hakeem voters think it's a choice between those 2, and say so. I'm voting for Bird, but for me it isn't even close who to vote for next- Hakeem by a mile. But because there are technically 60 voters in the project, it's unlikely either will attain the 31 vote true majority, so we're going to have Hakeem/Bird's margin slowly grow and grow over 2 days. I feel like that might start to get a bit stale. #11 is where this thing is going to explode in 6 different directions, I feel like a little bit more scope to discuss those 6+ guys for #11 might make the 2 days of a Hakeem/Bird walkover more invigorating. I mean, if you haven't been convinced against Bird/Hakeem by this point, it's unlikely anyone is going to be swayed now.
ceiling raiser
Lead Assistant
Posts: 4,531
And1: 3,754
Joined: Jan 27, 2013

Re: RealGM Top 100 List #9 

Post#186 » by ceiling raiser » Mon Jul 21, 2014 3:19 am

Moonbeam wrote:
fpliii wrote:
Warspite wrote:KG= GOAT 3rd option role player.

"I love KG so much but cant use video or stats to make him look great. What should I do? i know Ill invent a bunch of adv stats that make him look great."

I think that's incredibly disrespectful to KG and the analysis that some posters have done. Ridge regression is a proven, peer-reviewed mathematical method, and even aside from RAPM, KG excels in other non-box score metrics (APM in particular, but he holds up well in WOWY studies as well). It's your call to choose to subscribe to them or not, but to dismiss the work that's been done like that is unfortuante IMO.

drza had a great breakdown, mixing quantitative WITH qualitative a couple of threads ago:

viewtopic.php?p=40580241#p40580241


Very cool to see ridge regression applied to APM - hadn't seen that before! Do you know if there is a version that incorporates a different type of penalty, like LASSO or elastic net?

Well, I'm not sure if there's any published version using LASSO, but here's a comment from mystic, who doesn't seem to favor it (I guess it's because it uses an L1 norm, as opposed to ridge regression, which uses an L2 penalty, so maybe the fit is better?):

If you choose lambda to be zero (default is NULL), it should give you the APM results. You can also choose a parameter called alpha, which is the elastic-net parameter, that is basically a different kind of regularization, which is in essence better than LASSO (least absolute shrinkage and selection operator), which is then again a different way of handling such ill-posed problems. Anyway, you should probably start with alpha at 0 (which should be the default value, if I'm not mistaken).


I've been meaning to play around with when I have a chance (if you don't want to bother preparing a dataset, here's a link with a quick tutorial and easy script to get started: http://www.hickory-high.com/how-to-calculate-rapm/). glment in R has an optional alpha parameter. In calculating RAPM there's cross validation of lambda already. With a bigger dataset (I know J.E. on APBRmetrics has a big 14-year set) they can find a nice stable alpha via cv. That's another thing to play around with even in single-season RAPM though. Anyhow, maybe shutupandjam or acrossthecourt has experience fooling around with either.
Now that's the difference between first and last place.
therealbig3
RealGM
Posts: 29,433
And1: 16,019
Joined: Jul 31, 2010

Re: RealGM Top 100 List #9 

Post#187 » by therealbig3 » Mon Jul 21, 2014 3:26 am

For me, Bird doesn't come up on my radar until #12. Fantastic player, and honestly, as much as I like LeBron...LeBron vs Bird is a LEGITIMATE debate for me as far as peak vs peak, and I didn't used to think that way before this project (I had LeBron as the #3 peak prior to the project fwiw...Jordan and Shaq at #1 and #2...those were a clear top 3 for me...but now Bird and LeBron are a toss-up for #3 imo).

But his longevity is just weak compared to someone like KG, as well as Hakeem. 80-88 + 90 is 10 years where he's a super-high impact player. But injuries hamper all of these years. His 90 season in general is a clear step below his first 9 years. His 80 playoffs are marred by the fact that he runs out of gas by the time he hits the playoffs. His 83 playoffs are marred by the flu. His 85 playoffs are marred by a broken hand he suffered during the Finals. His 88 playoffs are marred by bone spurs. He was still a great player in the face of all of these ailments, but was he a better player than a healthy, prime KG? I don't think so.

And on top of that, KG has 10 years of being a super-high impact player imo (99-08), as well as a few AS-caliber seasons in addition to that (97, 98, and 10-13). That's 16 years vs Bird's 10 years.

TBH, I think I would take Karl Malone over Bird as well. Again, has nothing to do with prime Bird, who I think is a top 3-4 player in the history of the game...has everything to do with the fact that prime Bird just wasn't around very long, and when he was, he was still hampered by circumstances that reduced his impact. Injuries really didn't affect Hakeem, KG, or K. Malone all that much (they may have in a couple of instances here and there, but nothing major really comes to mind...10 for KG probably, who was visibly injured throughout the season following his knee injury in 09).

So I don't have Bird coming up until after Hakeem, KG, and K. Malone are in. There are a few other players that could maybe be argued over Bird as well (Oscar, West, and Erving come to mind).
Basketballefan
Banned User
Posts: 2,170
And1: 583
Joined: Oct 14, 2013

Re: RealGM Top 100 List #9 

Post#188 » by Basketballefan » Mon Jul 21, 2014 3:27 am

Warspite wrote:
HeartBreakKid wrote:
Warspite wrote:KG= GOAT 3rd option role player.

"I love KG so much but cant use video or stats to make him look great. What should I do? i know Ill invent a bunch of adv stats that make him look great."


Yes, because it's a giant conspiracy to create stats that happen to make KG look great.


Your first sentence doesn't even make sense. You talk as if Kevin Garnett was never considered a superstar or never considered the best player in the league. I guess everyone watching the NBA and everyone in the NBA must have "made up advance stats too". :crazy: Glad to know he was the third option in Minny as well, I guess Wally Z and Joe Smith was their 1-2 punch.


No KG wasnt a superstar nor best player at his position much less in the NBA. He was a perennial all star and a 2nd or 3rd all NBA team type player. Great player who put up great stats on mediocre teams that never developed.

He is the defensive version of Bob McAdoo.

I'm not sure how to even respond to this flagrant disrespect...Kg was easily a top 5 player for several different seasons..probably the best in 04, he put up the best numbers in the league while being the best defender and came within 2 wins of beating Kobe and Shaq to go to the finals.
Doctor MJ
Senior Mod
Senior Mod
Posts: 52,823
And1: 21,749
Joined: Mar 10, 2005
Location: Cali
     

Re: RealGM Top 100 List #9 

Post#189 » by Doctor MJ » Mon Jul 21, 2014 3:38 am

fpliii wrote:
Doctor MJ wrote:[So this is extremely useful information to have. Do people realize how painstaking this process is for ElGee? Were we granting PhD's in basketball analysis, this type of stuff would be worth the degree right there.

Agree 100%, it's incredibly, incredibly valuable, and in no small part due to the fact that it's orthogonal to box score data (one of the strengths of RAPM, as yourself and others have suggested). I really dig the fact that ElGee reported confidence intervals with his results. Really helps get an idea of a guy's impact insofar as affecting scoring margin is concerned. I really love what he's doing, and hope a team doesn't snatch him up too soon (as well as acrossthecourt, and some of the other brilliant analysts we have here). :wink:

Just wondering Doc, when comparing players, when do you believe the present "era" started? Would you go with 01-02, with the elimination of illegal defense? Or maybe something more recent (Thibs/KG changing how defense is played, Nash/D'Antoni bringing spacing+pacing into the league along with some of what the Spurs/Mavs/Heat have been doing, etc.)?

I ask because with regards to Hakeem/Garnett (for the record, even though the voting results aren't terribly important to me, I can't imagine there are 8 better players better than both guys), is the shotblocking really an advantage? With some of what yourself/drza/ElGee (and others) have posted, I've really been forced to reconsider the value of the vertical game defensively in the present era.

Has there been a study comparing shotblocking defensive anchors in the present era to non-shotblocking defensive anchors? Perhaps not only has KG's not the exception, but KG-"like" defenders (I use "like" very loosely here based on how unique he is) are far more valuable in today's league than shotblockers.


Really good questions. To be honest, I haven't done an analysis trying to quantify these era differences. So the things to look into:

How much (not if) of an advantage did no-hand check it to different types of perimeter player on offense?
Did no-hand check kill off perimeter player defensive impact?
What about big men defensive impact?

Are the huge defensive numbers we're seeing for big men defenders in the late '90s something systematic, or was it due simply to more talented bigs?

With regards to the last question, a detailed eye ball analysis of Dikembe Mutombo would be awesome. In the past he's not someone I feel like we've really focused on here. He's always obviously been a stellar defender, but he's the guy whose defensive RAPM are absolutely off the chart. How would his M.O. be forced to change with newer things (illegal defense, more fast breaks, extreme 3-point spacing, etc), and what of his strengths would be just weakened?

Answering literally your question about era those. The only one I've referred to here is the databall era, which to my mind extends back as far as we've got +/- for when we talk in this context (so '97), but would probably be 2000 based on coaches & GMs being informed.

Clearly '04-05 marks the start of a new era based on no-hand check.
I don't know if I think that the illegal defense rule change warrants being called an era shift. Had you asked me in '01-02 I might have said it was based on how the New Jersey Nets' running was the signature style of that year, but given what i now know of actual offensive efficacy and how meh the Nets were, it hardly seems to be worth calling so revolutionary.
Getting ready for the RealGM 100 on the PC Board

Come join the WNBA Board if you're a fan!
andrewww
General Manager
Posts: 7,989
And1: 2,687
Joined: Jul 26, 2006

Re: RealGM Top 100 List #9 

Post#190 » by andrewww » Mon Jul 21, 2014 3:48 am

Doctor MJ wrote:Really good questions. To be honest, I haven't done an analysis trying to quantify these era differences. So the things to look into:

How much (not if) of an advantage did no-hand check it to different types of perimeter player on offense?
Did no-hand check kill off perimeter player defensive impact?
What about big men defensive impact?

Are the huge defensive numbers we're seeing for big men defenders in the late '90s something systematic, or was it due simply to more talented bigs?

With regards to the last question, a detailed eye ball analysis of Dikembe Mutombo would be awesome. In the past he's not someone I feel like we've really focused on here. He's always obviously been a stellar defender, but he's the guy whose defensive RAPM are absolutely off the chart. How would his M.O. be forced to change with newer things (illegal defense, more fast breaks, extreme 3-point spacing, etc), and what of his strengths would be just weakened?

Answering literally your question about era those. The only one I've referred to here is the databall era, which to my mind extends back as far as we've got +/- for when we talk in this context (so '97), but would probably be 2000 based on coaches & GMs being informed.

Clearly '04-05 marks the start of a new era based on no-hand check.
I don't know if I think that the illegal defense rule change warrants being called an era shift. Had you asked me in '01-02 I might have said it was based on how the New Jersey Nets' running was the signature style of that year, but given what i now know of actual offensive efficacy and how meh the Nets were, it hardly seems to be worth calling so revolutionary.


I'd like to see if there are any trends or stats that might give more insight on how hand checking affected perimeter players.

My first reaction is that point guards and slashing wing players like LeBron or Wade would be the primary beneficiaries, moreso than more perimeter-oriented wing players like TMac or Kobe.
User avatar
SactoKingsFan
Assistant Coach
Posts: 4,236
And1: 2,760
Joined: Mar 15, 2014
       

Re: RealGM Top 100 List #9 

Post#191 » by SactoKingsFan » Mon Jul 21, 2014 3:50 am

Baller2014 wrote:Fplii, that's the expected narrative from Hakeem (we lost, but it wasn't our fault), but the truth is that Hakeem struggled against the Sonics over a prolonged period (both regular season and playoffs), and the reason was largely attributed to their "borderline illegal" zone D. A defensice technique that is completely legal against modern players. We should be looking closer at Hakeem's struggles with the Sonics, not washing over them with Hakeem's biography where he just blames the refs. His own team said at the time it was due to illegal D.


I'm not penalizing a post prime 33 year old Hakeem for struggling in one playoff series against the Sonics (96: 18.3 PTS, 47.5 FG%). He played much better in the playoffs against the Sonics in 97 (21.7 PTS, 57.5 FG%) and the Rockets went on to lose in the WCF in 6 to a superior Jazz team (UTA SRS: 7.97, HOU SRS: 3.85).
User avatar
Moonbeam
Forum Mod - Blazers
Forum Mod - Blazers
Posts: 10,214
And1: 5,062
Joined: Feb 21, 2009
Location: Sydney, Australia
     

Re: RealGM Top 100 List #9 

Post#192 » by Moonbeam » Mon Jul 21, 2014 3:51 am

fpliii wrote:
Moonbeam wrote:
fpliii wrote:I think that's incredibly disrespectful to KG and the analysis that some posters have done. Ridge regression is a proven, peer-reviewed mathematical method, and even aside from RAPM, KG excels in other non-box score metrics (APM in particular, but he holds up well in WOWY studies as well). It's your call to choose to subscribe to them or not, but to dismiss the work that's been done like that is unfortuante IMO.

drza had a great breakdown, mixing quantitative WITH qualitative a couple of threads ago:

viewtopic.php?p=40580241#p40580241


Very cool to see ridge regression applied to APM - hadn't seen that before! Do you know if there is a version that incorporates a different type of penalty, like LASSO or elastic net?

Well, I'm not sure if there's any published version using LASSO, but here's a comment from mystic, who doesn't seem to favor it (I guess it's because it uses an L1 norm, as opposed to ridge regression, which uses an L2 penalty, so maybe the fit is better?):

If you choose lambda to be zero (default is NULL), it should give you the APM results. You can also choose a parameter called alpha, which is the elastic-net parameter, that is basically a different kind of regularization, which is in essence better than LASSO (least absolute shrinkage and selection operator), which is then again a different way of handling such ill-posed problems. Anyway, you should probably start with alpha at 0 (which should be the default value, if I'm not mistaken).


I've been meaning to play around with when I have a chance (if you don't want to bother preparing a dataset, here's a link with a quick tutorial and easy script to get started: http://www.hickory-high.com/how-to-calculate-rapm/). glment in R has an optional alpha parameter. In calculating RAPM there's cross validation of lambda already. With a bigger dataset (I know J.E. on APBRmetrics has a big 14-year set) they can find a nice stable alpha via cv. That's another thing to play around with even in single-season RAPM though. Anyhow, maybe shutupandjam or acrossthecourt has experience fooling around with either.


Thanks for the info. I work with regularization in different contexts (and even have my own humble little R package called ppmlasso which is past due for an update), and I have found LASSO to be generally preferable for the problems I encounter given that it performs variable selection as well as shrinkage. You can get really complicated with this stuff with two-stage LASSO, adaptive LASSO or combine LASSO with ridge regression via the elastic net, but I thought I'd ask if there were any endeavors to use other penalties. :)

For anybody who is interested, Hastie's book The Elements of Statistical Learning is available for free online, and it has a nice section on penalized regression:

http://web.stanford.edu/~hastie/local.f ... print4.pdf
ceiling raiser
Lead Assistant
Posts: 4,531
And1: 3,754
Joined: Jan 27, 2013

Re: RealGM Top 100 List #9 

Post#193 » by ceiling raiser » Mon Jul 21, 2014 4:02 am

Thanks for the response.
Doctor MJ wrote:Really good questions. To be honest, I haven't done an analysis trying to quantify these era differences. So the things to look into:

How much (not if) of an advantage did no-hand check it to different types of perimeter player on offense?
Did no-hand check kill off perimeter player defensive impact?
What about big men defensive impact?

Hand-checking is interesting, and it did impact individual scoring for perimeter players, but I'm (perhaps incorrectly) thinking that a change affecting how bigs play is likely going to change the game more than how one impacting perimeter players is. Though spacing is largely making non-bigs more important, so maybe I'm wrong.

Hand-checking was also curtailed going into 78-79 and 94-95, so I wonder if those had similar effects?

Are the huge defensive numbers we're seeing for big men defenders in the late '90s something systematic, or was it due simply to more talented bigs?

Great, great question. That's what I'm hoping to learn.

With regards to the last question, a detailed eye ball analysis of Dikembe Mutombo would be awesome. In the past he's not someone I feel like we've really focused on here. He's always obviously been a stellar defender, but he's the guy whose defensive RAPM are absolutely off the chart. How would his M.O. be forced to change with newer things (illegal defense, more fast breaks, extreme 3-point spacing, etc), and what of his strengths would be just weakened?

Mutombo is a tremendous study here with his severe outliers in terms of dRAPM. I'm not sure how to feel about him. Was he that dominant? Or as you said above, was it something systematic? Depending on the answer, I might have to change how I think about defense entirely.

Answering literally your question about era those. The only one I've referred to here is the databall era, which to my mind extends back as far as we've got +/- for when we talk in this context (so '97), but would probably be 2000 based on coaches & GMs being informed.

Good point. So you'd go with 2000? Is that around when analytics departments/consultants started popping up (is there a timeline in terms of adaptation? Unfortunately for both of us, I believe the Lakers are still very much behind the ball in this regard)? Or just based on the availability/knowledge of their existence around the league?

Having the capacity to make informed decisions is a huge deal. It's still on the decision-makers to use the information they have obviously, but without the data existing this point is a non-starter.

I don't think it's a coincidence that around that time, the league became saturated with international players:

Image

The databall era likely had a huge impact on scouting/drafting immediately.

Clearly '04-05 marks the start of a new era based on no-hand check.
I don't know if I think that the illegal defense rule change warrants being called an era shift. Had you asked me in '01-02 I might have said it was based on how the New Jersey Nets' running was the signature style of that year, but given what i now know of actual offensive efficacy and how meh the Nets were, it hardly seems to be worth calling so revolutionary.

Good point. Some of what drza posted about Minny playing KG in a weird zone defensive scheme (something like four corners with KG roaming a ton?) early on the first year after the change is another interesting development.

Is there any way to quantify spacing? Maybe someone could generate something from the play-by-play, but just going by the box score:

Code: Select all

Season   3PA/FGA   3PA/TSA
2013-14   25.9%   24.7%
2012-13   24.4%   23.7%
2011-12   22.6%   22.1%
2010-11   22.2%   21.4%
2009-10   22.2%   21.3%
2008-09   22.4%   21.7%
2007-08   22.2%   21.4%
2006-07   21.2%   20.5%
2005-06   20.3%   19.9%
2004-05   19.7%   19.2%
2003-04   18.7%   18.4%
2002-03   18.2%   17.7%
2001-02   18.1%   17.7%
2000-01   17.0%   16.5%
1999-00   16.7%   15.8%
1998-99   16.9%   16.8%
1997-98   15.9%   15.6%
1996-97   21.2%   20.9%
1995-96   20.0%   19.3%
1994-95   18.8%   17.9%   <--Four-year college players from here on had the 3 each year in HS+college
1993-94   11.7%   10.7%
1992-93   10.5%   9.3%
1991-92   8.7%   7.8%
1990-91   8.1%   7.2%   <--Four-year college players from here on had the 3 each year in college
1989-90   7.6%   6.7%
1988-89   7.4%   6.4%
1987-88   5.7%   4.9%   <--High Schools adapt three
1986-87   5.3%   4.5%   <--NCAA adapts three
1985-86   3.7%   3.2%
1984-85   3.5%   2.9%
1983-84   2.7%   2.3%
1982-83   2.6%   2.2%
1981-82   2.6%   2.2%
1980-81   2.3%   1.9%
1979-80   3.1%   2.6%   <--NBA adapts three


Looks like it crossed 20% two years after the latest hand-checking rule change (three years in terms of 3PA/TSA, which might be more telling since players are rarely fouled on threes), though it's increasing in prevalence even further in the past couple of seasons.
Now that's the difference between first and last place.
ceiling raiser
Lead Assistant
Posts: 4,531
And1: 3,754
Joined: Jan 27, 2013

Re: RealGM Top 100 List #9 

Post#194 » by ceiling raiser » Mon Jul 21, 2014 4:03 am

Moonbeam wrote:Thanks for the info. I work with regularization in different contexts (and even have my own humble little R package called ppmlasso which is past due for an update), and I have found LASSO to be generally preferable for the problems I encounter given that it performs variable selection as well as shrinkage. You can get really complicated with this stuff with two-stage LASSO, adaptive LASSO or combine LASSO with ridge regression via the elastic net, but I thought I'd ask if there were any endeavors to use other penalties. :)

For anybody who is interested, Hastie's book The Elements of Statistical Learning is available for free online, and it has a nice section on penalized regression:

http://web.stanford.edu/~hastie/local.f ... print4.pdf

Very cool! Thanks for sharing. :)
Now that's the difference between first and last place.
Baller2014
Banned User
Posts: 2,049
And1: 519
Joined: May 22, 2014
Location: No further than the thickness of a shadow
     

Re: RealGM Top 100 List #9 

Post#195 » by Baller2014 » Mon Jul 21, 2014 4:06 am

SactoKingsFan wrote:
Baller2014 wrote:Fplii, that's the expected narrative from Hakeem (we lost, but it wasn't our fault), but the truth is that Hakeem struggled against the Sonics over a prolonged period (both regular season and playoffs), and the reason was largely attributed to their "borderline illegal" zone D. A defensice technique that is completely legal against modern players. We should be looking closer at Hakeem's struggles with the Sonics, not washing over them with Hakeem's biography where he just blames the refs. His own team said at the time it was due to illegal D.


I'm not penalizing a post prime 33 year old Hakeem for struggling in one playoff series against the Sonics (96: 18.3 PTS, 47.5 FG%). He played much better in the playoffs against the Sonics in 97 (21.7 PTS, 57.5 FG%) and the Rockets went on to lose in the WCF in 6 to a superior Jazz team (UTA SRS: 7.97, HOU SRS: 3.85).


It's not just one series Hakeem struggled in, he was well documented at the time to struggle against the Sonics zone D. I linked an article about this in an earlier thread that I'm happy to show you. The Rockets regular season record against the Sonics from 93-96 was 3-13, and in 93 and 96 the Rockets lost to the Sonics. Many people think Hakeem was extraordinarily lucky not to have to play the Sonics in 94 or 95, since he always seemed to do poorly against them.
therealbig3
RealGM
Posts: 29,433
And1: 16,019
Joined: Jul 31, 2010

Re: RealGM Top 100 List #9 

Post#196 » by therealbig3 » Mon Jul 21, 2014 4:31 am

Except literally the only time that Hakeem struggled against the Sonics was that 96 series. He was great every other time he faced off with them.
User avatar
SactoKingsFan
Assistant Coach
Posts: 4,236
And1: 2,760
Joined: Mar 15, 2014
       

Re: RealGM Top 100 List #9 

Post#197 » by SactoKingsFan » Mon Jul 21, 2014 4:33 am

Baller2014 wrote:
SactoKingsFan wrote:
Baller2014 wrote:Fplii, that's the expected narrative from Hakeem (we lost, but it wasn't our fault), but the truth is that Hakeem struggled against the Sonics over a prolonged period (both regular season and playoffs), and the reason was largely attributed to their "borderline illegal" zone D. A defensice technique that is completely legal against modern players. We should be looking closer at Hakeem's struggles with the Sonics, not washing over them with Hakeem's biography where he just blames the refs. His own team said at the time it was due to illegal D.


I'm not penalizing a post prime 33 year old Hakeem for struggling in one playoff series against the Sonics (96: 18.3 PTS, 47.5 FG%). He played much better in the playoffs against the Sonics in 97 (21.7 PTS, 57.5 FG%) and the Rockets went on to lose in the WCF in 6 to a superior Jazz team (UTA SRS: 7.97, HOU SRS: 3.85).


It's not just one series Hakeem struggled in, he was well documented at the time to struggle against the Sonics zone D. I linked an article about this in an earlier thread that I'm happy to show you. The Rockets regular season record against the Sonics from 93-96 was 3-13, and in 93 and 96 the Rockets lost to the Sonics. Many people think Hakeem was extraordinarily lucky not to have to play the Sonics in 94 or 95, since he always seemed to do poorly against them.


Hakeem's struggles against the Sonics from 93-96 were overblown.

93 RS vs SEA: 23.0 PTS, .481 FG%, 14.5 RB, 1.8 AST, 2.0 STL, 6.0 BLK, 4.5 TOV

94 RS vs SEA: 28.0 PTS, .686 FG%, 8.5 RB, 3.3 AST, 1.75 STL, 4.5 BLK, 2.25 TOV

95 RS vs SEA: 23 PTS, .481 FG%, 9.7 RB, 2.3 AST, 1.7 STL, 2.7 BLK, 2.7 TOV

96 RS vs SEA: 27.3 PTS, .571 FG%, 10.7 RB, 4.0 AST, 1.7 STL, 1.33 BLK, 3.7 TOV
90sAllDecade
Starter
Posts: 2,263
And1: 818
Joined: Jul 09, 2012
Location: Clutch City, Texas
   

Re: RealGM Top 100 List #9 

Post#198 » by 90sAllDecade » Mon Jul 21, 2014 4:44 am

I'm not penalizing Hakeem either, he played great without nearly the same team support as Bird, who struggled mightily with Dr. J. Especially with Bird's long list of other playoff failures with stacked teams and HCA.

By that logic (which I disagree with) Bird lucked out not playing prime Julius Erving in the 70s, but let's look at when they did play before age 34 and he declined.

When Dr. J was in his prime and they played each other from 79-82 in the postseason, he usually outplayed Larry and won 2-1 in playoff series, having Bird's number in two of them. In the only series Bird dominated and won in Erving's prime, the series went 7 games. Even in the last series when the Celtics clearly won, Bird struggled against old Dr. J in the playoffs.

Bird was young, but 2nd year Hakeem comparatively outplayed Kareem in the playoffs, playing PF and dominated another HOFer in Parish at center.
NBA TV Clutch City Documentary Trailer:
https://vimeo.com/134215151
JordansBulls
RealGM
Posts: 60,466
And1: 5,344
Joined: Jul 12, 2006
Location: HCA (Homecourt Advantage)

Re: RealGM Top 100 List #9 

Post#199 » by JordansBulls » Mon Jul 21, 2014 4:51 am

Here are the players that have a shot at this spot and there records while having the HCA.

Code: Select all

 vs 50 win teams/non-50 win teams 
Bird:     10-6 (63%)/  14-1 (93%)
Olajuwon: 4-0 (100%)/  5-2 (71%)
Kobe       18-2 (90%) / 7-0  (100%)


Bird 24-7 with HCA
Hakeem 9-2 with HCA (amazing he only had so few series with HCA)
Kobe 25-2 with HCA



Here are those guys without HCA

Code: Select all

 
              Road(50+)/non-50   
Olajuwon:     7-8 / 1-1
Bird:         0-4 / 0-0
Kobe:         5-5 / 0-0
Image
"Talent wins games, but teamwork and intelligence wins championships."
- Michael Jordan
90sAllDecade
Starter
Posts: 2,263
And1: 818
Joined: Jul 09, 2012
Location: Clutch City, Texas
   

Re: RealGM Top 100 List #9 

Post#200 » by 90sAllDecade » Mon Jul 21, 2014 5:30 am

When you consider the heavy team support advantage, it helps put those team based statistics in perspective.

With his minimal in comparison to the other player's team support, Hakeem was better against 50 win teams with HCA and Bird never won without HCA. Bird very much needed that team support to help him succeed in comparison to what Hakeem had for their careers.

These numbers are also skewed to help players who had better team support as well. Having higher ranked all time players & coaching on your roster will naturally increase your odds of team success, which puts stats like this in context.

It doesn't analyze the differences in teammate quality and coaching each player had to get these team based statistics.
NBA TV Clutch City Documentary Trailer:
https://vimeo.com/134215151

Return to Player Comparisons