Why were Kobe's RAPM numbers better in 08-10 than 02-04?

[TheChosen618]

People always say that Kobe was overrated defensively after Shaq left because he would coast and not play much defense. People do acknowledge that Kobe would play defense back when he was with Shaq though but they say it because he was a side. Why are Kobe's RAPM numbers better from 08-10 than 02-04?

https://sites.google.com/site/rapmstats/2010-rapm

Kobe 2002: +0.1
Kobe 2003: +0.1
Kobe 2004: -0.6

Kobe 2008: +0.5
Kobe 2009: +0.4
Kobe 2010: +1.3

Keep in mind that I am talking about RAPM here and not XRAPM. I am curious about the context of this stat more than anything. I understand that no stat is perfect but this stat supposedly adjusts everything and makes it act as if he was playing with 4 average players, what could be the reason for why the stats are what they are?

Could it be because Kobe was actually a better defender in 08-10? or.....?

[RayBan-Sematra]

Confusing.
I thought Kobe was quite good defensively in 08 & 09 while I thought he was terrible defensively in 2010.

In 02 I thought he was still a good (not elite) defender.
In terms of 03 & 04 I didn't think he was great defensively those years.

So if I had to rank those seasons in terms of defense.

1. 2008
2. 2009
3. 2002
4. 2003
5. 2004
6. 2010

Hard to answer your question but I would say his 08-10 stretch was better defensively then his 02-04 stretch.
08-10 = two very good years + one horrible year.
02-04 = one good year + two neutral/average years.

[trainwreckog]

rapm is measuring the same thing apm measures, but is just doing a better job of it right?

if so, then kobe's rapm figure is a numerical value that represents the portion of the team's differential that kobe is responsible for (not really the "team's" differential, but various 5-man units that kobe is a part of).

so if the point differential between a 5-man unit consisting of kobe and 4 other players is say, 10 points when that 5 goes up against another 5 guys, then the rapm figure shows what portion of that 10 points kobe is responsible for... or at least the rapm score is calculated after deducing this kobe proportion from empirical data.

if you look at the lakers point differential as a team in the years 2008-2010, it is much higher then their point differential in 2002-2004, so there is a bigger pie to calculate kobe's slice from in 2008-2010.

this could easily not be the reason.. i'm just taking a shot at it... i had never looked up rapm before.. from what i can tell, there is a ton of error without proper sample sizes, which can take years to accumulate (since there is often not big sample sizes of minutes for specific 5-man units vs. other specific 5-man units).

[trainwreckog]

actually rapm must account for changing point differentials..

don't know..

[colts18]

His RAPM's were better late in his career.


Year Year Rank Name Offense per 100 Defense per 100 Off+Def per 200
2012 55 Kobe Bryant 1.3 0.2 1.5
2011 32 Kobe Bryant 3.6 -0.9 2.7
2010 4 Kobe Bryant 4.3 1.3 5.6
2009 5 Kobe Bryant 5.8 0.4 6.1
2008 6 Kobe Bryant 5.7 0.5 6.1
2007 8 Kobe Bryant 6 -0.5 5.5
2006 6 Kobe Bryant 5.9 -1.2 4.8
2006 NPI 24 Kobe Bryant 3.6 -1.1 2.5
2005 123 Kobe Bryant 1.8 -1.1 0.7
2005 NPI 95 Kobe Bryant 1.8 -1.1 0.7
2004 55 Kobe Bryant 2.2 -0.6 1.6
2004 NPI 29 Kobe Bryant 2.5 -0.4 2.2
2003 18 Kobe Bryant 2.3 0.1 2.4
2003 NPI 19 Kobe Bryant 2.9 -0.2 2.8
2002 18 Kobe Bryant 1.8 0.1 1.9
2002 NPI 26 Kobe Bryant 3.5 -1 2.5
2001 NPI 8 Kobe Bryant 3.2 0.9 4.1
08-11 15 Kobe Bryant 4.7 0.1 4.8
06-11 13 Kobe Bryant 4.9 -0.3 4.6
02-11 5 Kobe Bryant 6.6 -0.5 6.1

[Durins Baynes]

Dissent as appropriate, but do it in a productive manner.

[tsherkin]

Keep your replies to productive debate/commentary, people; if you're just casually dismissing something, don't post.

[bondom34]

I'm not exactly sure of the RAPM calculation, but trainwreckdog may be close to the reason I think. If you look at the Lakers as a team, the 02-04 range had a better overall D Rating , but ranked much worse comparitively in the league (they were around 5th in the league in 08-10). Considering the team as a whole was ranked higher defensively, I would think player's numbers in RAPM would see a bump as well, considering it takes lineups into account. Even if he individually wasn't better it could possibly do something when compared to league average numbers.

[TheChosen618]

The purpose of RAPM is to adjust though and adjusts lineups and act as if that player were replacing an average player. 0 is the number for the average player. The way people talk about this stat on here, you would think it is flawless but I know it isn't because there is no stat that is flawless and I'd like to hear an expert on RAPM (Doctor MJ, SideshowBob) to explain it.

[bondom34]

Hmm..I'd be interested myself now that you mention it, it would be a good lesson from one of those guys as well.

[emotional]

Kobe is the most enigmatic player ever imo. Narrative has often muddled the truth and different stats even duke it out.


Also, what is the difference between RAPM and XRAPM. Why do people use the former if the guy who created it abandoned it for the newer modification?

[TheChosen618]

Kobe is the most enigmatic player ever imo. Narrative has often muddled the truth and different stats even duke it out.


Also, what is the difference between RAPM and XRAPM. Why do people use the former if the guy who created it abandoned it for the newer modification?

xRAPM adds boxscore data to the numbers, so it factors in many irrelevant things.

Kobe's xRAPM is pretty much average to below average from 00 to 10 according to xRAPM though. It was very steady but xRAPM is not reliable.

[tsherkin]

APM in general is a summary of the efficiency differentials for player matchups over a whole season, weighted by number of possessions. Consequently, sample size IS very important, even for derivatives like RAPM and xRAPM. Most of the good ones are multi-year samples, the larger the better (though even then, you run into issues like injury and decline phase/age, etc). RAPM helps control for some of the various issues with APM through ridge regression, but that accuracy comes with a prerequisite of ~ 3 years of data. Basically, prior-informed RAPM helps smooth things out by using a statistical "laugh test" to ensure that things stay within a reasonable range of probable variation, to my understanding.

IIRC, xRAPM is essentially regularized Statistical Plus/Minus, no? Isn't that Rosenbaum's methodology?

[PCProductions]

Isn't RAPM based on previous data? I'm guessing the influence of having years of his data prior to the 08-10 years is part of the reason why it beats the 02-04 years where they only had a few years of "prediction" or whatever.

[acrossthecourt]

Kobe is the most enigmatic player ever imo. Narrative has often muddled the truth and different stats even duke it out.


Also, what is the difference between RAPM and XRAPM. Why do people use the former if the guy who created it abandoned it for the newer modification?

xRAPM adds boxscore data to the numbers, so it factors in many irrelevant things.

Kobe's xRAPM is pretty much average to below average from 00 to 10 according to xRAPM though. It was very steady but xRAPM is not reliable.

If it offers a better out of sample prediction, how is it less reliable?

[nikomCH]

RAPM data isn't exactly comparable across all seasons because of sample size. It's not until like 2004 at the earliest when it actually makes sense to look at the numbers that way.

[Dr Positivity]

A RAPM gap that small is basically meaningless, especially across different seasons. Eg. It could be caused by the Odom Gasol frontcourt starting with Kobe, where Lakers success correlated most

[giberish]

There's some noise on year to year RAPM numbers, and you're not showing large changes. Though there is a solid case that he wasn't better in 02-04.

It does agree with the theory that Kobe has always coasted significantly on D (why he always rates near average despite his name recognition and 1 on 1 D getting him All-D team honors). There's also a solid case that Pau/Odom/Bynum combined to be as big or bigger offensive force than Shaq by himself so that Kobe was expending no more (and perhaps a bit less) energy on offense in 08-10 than he was in 02-04.

[nikomCH]

There's also a solid case that Pau/Odom/Bynum combined to be as big or bigger offensive force than Shaq by himself so that Kobe was expending no more (and perhaps a bit less) energy on offense in 08-10 than he was in 02-04.

Don't see this at all. Prime Shaq's offensive RAPM numbers (using xRAPM instead since classic RAPM is incomplete) completely demolish Pau+Odom+Bynum's even when combined.

[TheChosen618]

A RAPM gap that small is basically meaningless, especially across different seasons. Eg. It could be caused by the Odom Gasol frontcourt starting with Kobe, where Lakers success correlated most

But doesn't RAPM adjust for the quality of their teammates and opponents though? If so, why would those players play a factor into it?

There's some noise on year to year RAPM numbers, and you're not showing large changes. Though there is a solid case that he wasn't better in 02-04.

It does agree with the theory that Kobe has always coasted significantly on D (why he always rates near average despite his name recognition and 1 on 1 D getting him All-D team honors). There's also a solid case that Pau/Odom/Bynum combined to be as big or bigger offensive force than Shaq by himself so that Kobe was expending no more (and perhaps a bit less) energy on offense in 08-10 than he was in 02-04.

What do you mean by noise?

Also, while his 08-10 front-court may have been better than Shaq alone, Kobe still gave more effort defensively under Shaq because he wasn't the #1 on the team with Shaq like he was in 08-10.