What does RAPM say about Kobe's career?

[IdolW0rm]

Do his RAPM values back up the fact that he should be viewed as a Top 10, possibly Top 5 ATG?
What about Jordan v Lebron?

I apologyze for asking this because I'm a total noob at these advanced stats and I don't know the metrics of RAPM all too much nor did I find it anywhere, but I get the sense that the community on this board seems to have RAPM in pretty good consideration.

Hoping anyone can shed some light for me on this!

[giberish]

From memory, he looks good but not as great as his fans think he was. So it doesn't back up the opinion of him being a top-10 all time player.

Generally only a couple of top-5 seasons but a ton of top-20 seasons.

Again, from memory so may not be correct (and depends some of which RAPM site is being used).

[Odinn21]

RAPM is only available for 1996-97 and onwards because it's the mark for league wide +/- tracking.

There are two types RAPM;
- PI, prior informed; this helps with overall regression / reliability of results. It basically gives a head start to coefficients, instead of starting from scratch. But it also causes a carrying effect.
- NPI, non-prior informed; lower reliability ratings but no carrying effect.
In general, PI-RAPM is the better option between the two.

Some reliable RAPM sources for Bryant's career;
https://docs.google.com/spreadsheets/d/1ZW5UBk5BwcrEhh4mEsWl_H1Y9QGTaMPaT4v6dSEIBIo/edit#gid=0 (create a copy from the top left side under File section to browse through seasons and players with filters / also these numbers have rs + ps play under consideration, not just rs)
https://public.tableau.com/profile/dsmok1#!/vizhome/BPMvs_RAPM/BoxPlusMinusvs_14YearRAPM (I think Meyers used only rs numbers of Engelmann in these)

http://nbashotcharts.com/rapm?id=-2146555570 (just rs consideration)
This is for numbers after 2009.

All these sources have PI-RAPM.

In short;
Kobe was arguably a top 5 offensive player in the league since 1996-97. There's LeBron, Nash, Dirk, Shaq, Steph, Wade, Paul, Durant, Harden as better or similar level players.
When we add defensive end to the mix, Kobe stays in the top 10 but instead of a good argument for top 5, he probably is in 6-10 range as LeBron and Shaq were more impactful defenders than Kobe and there's also defensive savants as Timmy and KG.

The most important thing about all these, it's that this is only since 1996-97. Kobe's stature as an icon is top 5 material but as a player, he doesn't come close enough to be top 5 ever. He's barely top 5 in the last 25 years of ball.

[Colbinii]

RAPM is only available for 1996-97 and onwards because it's the mark for league wide +/- tracking.

There are two types RAPM;
- PI, prior informed; this helps with overall regression / reliability of results. It basically gives a head start to coefficients, instead of starting from scratch. But it also causes a carrying effect.
- NPI, non-prior informed; lower reliability ratings but no carrying effect.
In general, PI-RAPM is the better option between the two.

Some reliable RAPM sources for Bryant's career;
https://docs.google.com/spreadsheets/d/1ZW5UBk5BwcrEhh4mEsWl_H1Y9QGTaMPaT4v6dSEIBIo/edit#gid=0 (create a copy from the top left side under File section to browse through seasons and players with filters / also these numbers have rs + ps play under consideration, not just rs)
https://public.tableau.com/profile/dsmok1#!/vizhome/BPMvs_RAPM/BoxPlusMinusvs_14YearRAPM (I think Meyers used only rs numbers of Engelmann in these)

http://nbashotcharts.com/rapm?id=-2146555570 (just rs consideration)
This is for numbers after 2009.

All these sources have PI-RAPM.

In short;
Kobe was arguably a top 5 offensive player in the league since 1996-97. There's LeBron, Nash, Dirk, Shaq, Steph, Wade, Paul, Durant, Harden as better or similar level players.
When we add defensive end to the mix, Kobe stays in the top 10 but instead of a good argument for top 5, he probably is in 6-10 range as LeBron and Shaq were more impactful defenders than Kobe and there's also defensive savants as Timmy and KG.

The most important thing about all these, it's that this is only since 1996-97. Kobe's stature as an icon is top 5 material but as a player, he doesn't come close enough to be top 5 ever. He's barely top 5 in the last 25 years of ball.

Good post.

Also remember there are other statistics that may match your eye test better than RAPM. Each statistic is a tool and if making a GOAT ranking is equivalent to building a home, you don't want to be cutting studs with a hammer.

[Dutchball97]

I don't really get the appeal of RAPM, at least as a tool for single season ratings I find it to be pretty lacking. It is important to use as many stats as possible but I rarely find myself agreeing with RAPM.

[kayess]

I don't really get the appeal of RAPM, at least as a tool for single season ratings I find it to be pretty lacking. It is important to use as many stats as possible but I rarely find myself agreeing with RAPM.

And why is that?

[Owly]

Caveats about noise, contextual value rather than necessarily pure player goodness etc would have to prefix any conclusions ... Also there isn't one definitive RAPM, others knowing more than me can dive deeper if you/they desire.

https://sites.google.com/site/rapmstats/97-14-rapm-2
https://sites.google.com/site/rapmstats/97-14-rapm

The above covers most of Kobe's prime and all of Kobe's useful years. It covers his same generation guys (Duncan, Garnett, Nowitzki) and adjacent with substantial crossover guys (Shaq, LeBron). I believe it to be reliable (but am open to being wrong on this).

If you were to heavily buy into it (this version of RAPM) as your sole or primary (with a substantial weight) player evaluation tool, then personally I think it would kill the idea of Kobe top 5 all time (and substantially hurt top 10 all time), placing, as it does, Kobe outside the top 5 (6th, below the guys mentioned above) in value added above average in this span, and rate wise significantly lower (so for instance if you think Stockton kept up or increased his impact in earlier years, his career would come out ahead). You can eke him ahead of Shaq (within that span, note that this is ignoring Shaq's first 3 years, Kobe's last two, not necessarily a fair swap) in value above replacement (people may differ on the value being "above replacement"). The above idea doesn't lock him out of top 5 in era, but I say, if heavily weighted it would be very tricky to get him top 5 all time (and as I say hard for top 10 as, with these numbers as your baseline, it might be bullish to see him as even with Nowitzki, or put another way there are a lot of good players with the bulk or entirity of their career outside this window).

People's interpretations may differ, few will put that level of stock into one number and people's criteria for GOAT listings differ (longevity of quality/championship probability added versus peak; playoff weighting; interpreting eras; level of play versus narrative significance and ring; trust in accolades as a measure of level of play etc). Other calculations of RAPM may look different. (I wouldn't know what to do with it, but fwiw, his raw playoff on-off numbers are better than RS and you'd think the impact family in general is kinder in the playoffs - though this version of playoff RAPM - can't speak to reputational accuracy - isn't too bullish
https://docs.google.com/spreadsheets/d/e/2PACX-1vQdG8Zv84zqKEzETDjd8KPsClcw9bPETX9v_x_KEAxjv9NrFaWikOoiSaciy1jbMiygg2D-V8DUQn0O/pubhtml?gid=112475182&single=true
cf: viewtopic.php?f=64&p=76043070, https://www.reddit.com/r/nba/comments/db5hpt/19982019_playoffs_rapm_1_lebron_james_1_draymond/)

RAPM offers little on LeBron versus Jordan as RAPM only exists for the last 2 years of the second 3-peat and Wizards Jordan.

That's a pretty superficial, surface level take. I imagine others who understand it all better could tell you more.

[Colbinii]

I don't really get the appeal of RAPM, at least as a tool for single season ratings I find it to be pretty lacking. It is important to use as many stats as possible but I rarely find myself agreeing with RAPM.

And why is that?

Because he values his own opinion over the statistic, which makes sense given his openness to admit his limitations to understanding the statistic.

[Dutchball97]

I don't really get the appeal of RAPM, at least as a tool for single season ratings I find it to be pretty lacking. It is important to use as many stats as possible but I rarely find myself agreeing with RAPM.

And why is that?

Because he values his own opinion over the statistic, which makes sense given his openness to admit his limitations to understanding the statistic.

Wow, you've been holding that in for a while or something? Or did something happen in your daily life you feel like you have to take out on some guy you don't even know?

You literally just said in your earlier post that it is important to use as many statistics as possible but that RAPM might not be everyone's cup of tea. I then say use as many statistics as possible but RAPM isn't personally my cup of tea and then you choose violence? Are you sure you're not just really fond of RAPM and you can't handle others not liking it after all?

[Colbinii]

And why is that?

Because he values his own opinion over the statistic, which makes sense given his openness to admit his limitations to understanding the statistic.

Wow, you've been holding that in for a while or something? Or did something happen in your daily life you feel like you have to take out on some guy you don't even know?

Nah, life has been great! I just got home from the farmers market!

If you don't agree with RAPM then its quite clear you don't understand what RAPM is capturing as a statistic.

I like the analogy of not agreeing with RAPM as an equivalent to saying you don't agree Steph Curry shot 42.1% from 3 while Joe Harris shot 47.5% from 3 this season.

You literally just said in your earlier post that it is important to use as many statistics as possible but that RAPM might not be everyone's cup of tea. I then say use as many statistics as possible but RAPM isn't personally my cup of tea and then you choose violence? Are you sure you're not just really fond of RAPM and you can't handle others not liking it after all?

I'm not being violent--abrasive? I'll concede that point.

I don't necessarily value RAPM over other all-in-one statistics and I'm not particularly fond of it as a yearly measuring stick.

[Dutchball97]

Because he values his own opinion over the statistic, which makes sense given his openness to admit his limitations to understanding the statistic.

Wow, you've been holding that in for a while or something? Or did something happen in your daily life you feel like you have to take out on some guy you don't even know?

Nah, life has been great! I just got home from the farmers market!

If you don't agree with RAPM then its quite clear you don't understand what RAPM is capturing as a statistic.

I like the analogy of not agreeing with RAPM as an equivalent to saying you don't agree Steph Curry shot 42.1% from 3 while Joe Harris shot 47.5% from 3 this season.

You literally just said in your earlier post that it is important to use as many statistics as possible but that RAPM might not be everyone's cup of tea. I then say use as many statistics as possible but RAPM isn't personally my cup of tea and then you choose violence? Are you sure you're not just really fond of RAPM and you can't handle others not liking it after all?

I'm not being violent--abrasive? I'll concede that point.

I don't necessarily value RAPM over other all-in-one statistics and I'm not particularly fond of it as a yearly measuring stick.

So you agree it isn't a good yearly measuring stick. I'm not saying I don't agree with the outcome of the formula, that is indeed impossible. It was always about how it is an inaccurate stat for yearly catchall rankings.

If someone says they don't think WS is a representative ranking of players would you also say they value their opinion over statistics?

[homecourtloss]

I don't really get the appeal of RAPM, at least as a tool for single season ratings I find it to be pretty lacking. It is important to use as many stats as possible but I rarely find myself agreeing with RAPM.

RAPM isn’t a subjective opinion one “agrees with” or “disagrees with”; it’s NOT a ranking of players. It’s an objective statistical output (though subjectivity may lie in the priors selected if using Prior Informed RAPM) but those priors are at least created via objective data. There are slight variances in the actual final numbers depending on parameters but the end results are consistently similar.

Now, this may seem semantic, but I’m not sure with your choice of diction whether you mean to say “I don’t agree with RAPM methodology,” or “I sometimes don’t agree with the results.” If the latter, there’s nothing to “disagree” about since it’s a statistical output. We can still contextualize, though, but RAPM data in part does that and gives us a starting point to talk about how much a player is contributing to winning margins.

RAPM data isn’t a way to rank players, but it does tell us, especially over larger and larger sets of data, a player’s contributions towards winning margins.

Look at the PI RAPM Odinn21 linked.

https://docs.google.com/spreadsheets/d/1ZW5UBk5BwcrEhh4mEsWl_H1Y9QGTaMPaT4v6dSEIBIo/edit#gid=0

You have 10,500 player seasons there and you see a few names over and over again at the top. The chances that these players are there randomly is close to literally zero given this much data. KG with the top two, 3 of the top 7; LeBron with 3 of the top 6, 6 of the top 11. These are absurd results that they would occupy these spots out of all the data and all the primes and peaks of so many players. But this DOESN’T mean KG or James are necessarily “the best,” or “the greatest,” since those are subjective terms that need to be defined. This output, however, is an objective output of impact contribution towards winning margins. What we do with this data is the contextualization and that contextualization leads to lists of “goodness,” “greatness,” etc., we create.

Mookie, one of the most underrated players ever, has an RAPM number 46th and 58th on this list. Are those seasons “greater than” or “better than” Kobe’s seasons? RAPM does NOT make that argument or ANY argument; it’s simply measuring Mookie’s impact on final margins trying to isolate from teammates’ impact in an objective manner that’s not perfect.

[IdolW0rm]

RAPM is only available for 1996-97 and onwards because it's the mark for league wide +/- tracking.

There are two types RAPM;
- PI, prior informed; this helps with overall regression / reliability of results. It basically gives a head start to coefficients, instead of starting from scratch. But it also causes a carrying effect.
- NPI, non-prior informed; lower reliability ratings but no carrying effect.
In general, PI-RAPM is the better option between the two.

Some reliable RAPM sources for Bryant's career;
https://docs.google.com/spreadsheets/d/1ZW5UBk5BwcrEhh4mEsWl_H1Y9QGTaMPaT4v6dSEIBIo/edit#gid=0 (create a copy from the top left side under File section to browse through seasons and players with filters / also these numbers have rs + ps play under consideration, not just rs)
https://public.tableau.com/profile/dsmok1#!/vizhome/BPMvs_RAPM/BoxPlusMinusvs_14YearRAPM (I think Meyers used only rs numbers of Engelmann in these)

http://nbashotcharts.com/rapm?id=-2146555570 (just rs consideration)
This is for numbers after 2009.

All these sources have PI-RAPM.

In short;
Kobe was arguably a top 5 offensive player in the league since 1996-97. There's LeBron, Nash, Dirk, Shaq, Steph, Wade, Paul, Durant, Harden as better or similar level players.
When we add defensive end to the mix, Kobe stays in the top 10 but instead of a good argument for top 5, he probably is in 6-10 range as LeBron and Shaq were more impactful defenders than Kobe and there's also defensive savants as Timmy and KG.

The most important thing about all these, it's that this is only since 1996-97. Kobe's stature as an icon is top 5 material but as a player, he doesn't come close enough to be top 5 ever. He's barely top 5 in the last 25 years of ball.

Why are Doug Christies numbers so different in the first link you posted compared to other sources that show his PI RAPM in (i think) 02 and 03 as godly?

[Morb]

Wow, they fixed Andrew Declerq DRAPM))
T-Mac '98-'99 — –2.71 (???)
T-Mac '99-'00 — –2.14 (lol what?))
T-Mac '00-'01 — +0.3
T-Mac '01-'02 — +1.45
T-Mac '02-'03 — +0.5
Darrell Armstrong '02-'03 — –0.58
Bruh...(( I think it's trash stat.(
Sent from my Nokia 5.1 Plus using RealGM mobile app

[No-more-rings]

He's been pretty consistently good sometimes great by RAPM, but it does sort of give guys like KG and Dirk ammunition over him since those 2 seemed to fair better on average.

Edit: Actually looking again, I'm not sure Dirk actually has a decisive advantage outside of when he was 1st in 2011. They seem comparable.

[DirtyDez]

RAPM is only available for 1996-97 and onwards because it's the mark for league wide +/- tracking.

There are two types RAPM;
- PI, prior informed; this helps with overall regression / reliability of results. It basically gives a head start to coefficients, instead of starting from scratch. But it also causes a carrying effect.
- NPI, non-prior informed; lower reliability ratings but no carrying effect.
In general, PI-RAPM is the better option between the two.

Some reliable RAPM sources for Bryant's career;
https://docs.google.com/spreadsheets/d/1ZW5UBk5BwcrEhh4mEsWl_H1Y9QGTaMPaT4v6dSEIBIo/edit#gid=0 (create a copy from the top left side under File section to browse through seasons and players with filters / also these numbers have rs + ps play under consideration, not just rs)
https://public.tableau.com/profile/dsmok1#!/vizhome/BPMvs_RAPM/BoxPlusMinusvs_14YearRAPM (I think Meyers used only rs numbers of Engelmann in these)

http://nbashotcharts.com/rapm?id=-2146555570 (just rs consideration)
This is for numbers after 2009.

All these sources have PI-RAPM.

In short;
Kobe was arguably a top 5 offensive player in the league since 1996-97. There's LeBron, Nash, Dirk, Shaq, Steph, Wade, Paul, Durant, Harden as better or similar level players.
When we add defensive end to the mix, Kobe stays in the top 10 but instead of a good argument for top 5, he probably is in 6-10 range as LeBron and Shaq were more impactful defenders than Kobe and there's also defensive savants as Timmy and KG.

The most important thing about all these, it's that this is only since 1996-97. Kobe's stature as an icon is top 5 material but as a player, he doesn't come close enough to be top 5 ever. He's barely top 5 in the last 25 years of ball.

On a random note where can I see single season RAPM for 97-00’? All those links seem to start at 01’.. thx.

[IdolW0rm]

Caveats about noise, contextual value rather than necessarily pure player goodness etc would have to prefix any conclusions ... Also there isn't one definitive RAPM, others knowing more than me can dive deeper if you/they desire.

https://sites.google.com/site/rapmstats/97-14-rapm-2
https://sites.google.com/site/rapmstats/97-14-rapm

The above covers most of Kobe's prime and all of Kobe's useful years. It covers his same generation guys (Duncan, Garnett, Nowitzki) and adjacent with substantial crossover guys (Shaq, LeBron). I believe it to be reliable (but am open to being wrong on this).

If you were to heavily buy into it (this version of RAPM) as your sole or primary (with a substantial weight) player evaluation tool, then personally I think it would kill the idea of Kobe top 5 all time (and substantially hurt top 10 all time), placing, as it does, Kobe outside the top 5 (6th, below the guys mentioned above) in value added above average in this span, and rate wise significantly lower (so for instance if you think Stockton kept up or increased his impact in earlier years, his career would come out ahead). You can eke him ahead of Shaq (within that span, note that this is ignoring Shaq's first 3 years, Kobe's last two, not necessarily a fair swap) in value above replacement (people may differ on the value being "above replacement"). The above idea doesn't lock him out of top 5 in era, but I say, if heavily weighted it would be very tricky to get him top 5 all time (and as I say hard for top 10 as, with these numbers as your baseline, it might be bullish to see him as even with Nowitzki, or put another way there are a lot of good players with the bulk or entirity of their career outside this window).

People's interpretations may differ, few will put that level of stock into one number and people's criteria for GOAT listings differ (longevity of quality/championship probability added versus peak; playoff weighting; interpreting eras; level of play versus narrative significance and ring; trust in accolades as a measure of level of play etc). Other calculations of RAPM may look different. (I wouldn't know what to do with it, but fwiw, his raw playoff on-off numbers are better than RS and you'd think the impact family in general is kinder in the playoffs - though this version of playoff RAPM - can't speak to reputational accuracy - isn't too bullish
https://docs.google.com/spreadsheets/d/e/2PACX-1vQdG8Zv84zqKEzETDjd8KPsClcw9bPETX9v_x_KEAxjv9NrFaWikOoiSaciy1jbMiygg2D-V8DUQn0O/pubhtml?gid=112475182&single=true
cf: viewtopic.php?f=64&p=76043070, https://www.reddit.com/r/nba/comments/db5hpt/19982019_playoffs_rapm_1_lebron_james_1_draymond/)

RAPM offers little on LeBron versus Jordan as RAPM only exists for the last 2 years of the second 3-peat and Wizards Jordan.

That's a pretty superficial, surface level take. I imagine others who understand it all better could tell you more.

Why do enourmous variations happen between the first link you provided and the spreadsheet provided by Odinn (https://docs.google.com/spreadsheets/d/1ZW5UBk5BwcrEhh4mEsWl_H1Y9QGTaMPaT4v6dSEIBIo/edit#gid=0)?
For example 2003 McGrady being a +1.1 on your link and a +4.43 on the second link. It'a gigantic difference.

[Owly]

Caveats about noise, contextual value rather than necessarily pure player goodness etc would have to prefix any conclusions ... Also there isn't one definitive RAPM, others knowing more than me can dive deeper if you/they desire.

https://sites.google.com/site/rapmstats/97-14-rapm-2
https://sites.google.com/site/rapmstats/97-14-rapm

The above covers most of Kobe's prime and all of Kobe's useful years. It covers his same generation guys (Duncan, Garnett, Nowitzki) and adjacent with substantial crossover guys (Shaq, LeBron). I believe it to be reliable (but am open to being wrong on this).

If you were to heavily buy into it (this version of RAPM) as your sole or primary (with a substantial weight) player evaluation tool, then personally I think it would kill the idea of Kobe top 5 all time (and substantially hurt top 10 all time), placing, as it does, Kobe outside the top 5 (6th, below the guys mentioned above) in value added above average in this span, and rate wise significantly lower (so for instance if you think Stockton kept up or increased his impact in earlier years, his career would come out ahead). You can eke him ahead of Shaq (within that span, note that this is ignoring Shaq's first 3 years, Kobe's last two, not necessarily a fair swap) in value above replacement (people may differ on the value being "above replacement"). The above idea doesn't lock him out of top 5 in era, but I say, if heavily weighted it would be very tricky to get him top 5 all time (and as I say hard for top 10 as, with these numbers as your baseline, it might be bullish to see him as even with Nowitzki, or put another way there are a lot of good players with the bulk or entirity of their career outside this window).

People's interpretations may differ, few will put that level of stock into one number and people's criteria for GOAT listings differ (longevity of quality/championship probability added versus peak; playoff weighting; interpreting eras; level of play versus narrative significance and ring; trust in accolades as a measure of level of play etc). Other calculations of RAPM may look different. (I wouldn't know what to do with it, but fwiw, his raw playoff on-off numbers are better than RS and you'd think the impact family in general is kinder in the playoffs - though this version of playoff RAPM - can't speak to reputational accuracy - isn't too bullish
https://docs.google.com/spreadsheets/d/e/2PACX-1vQdG8Zv84zqKEzETDjd8KPsClcw9bPETX9v_x_KEAxjv9NrFaWikOoiSaciy1jbMiygg2D-V8DUQn0O/pubhtml?gid=112475182&single=true
cf: viewtopic.php?f=64&p=76043070, https://www.reddit.com/r/nba/comments/db5hpt/19982019_playoffs_rapm_1_lebron_james_1_draymond/)

RAPM offers little on LeBron versus Jordan as RAPM only exists for the last 2 years of the second 3-peat and Wizards Jordan.

That's a pretty superficial, surface level take. I imagine others who understand it all better could tell you more.

Why do enourmous variations happen between the first link you provided and the spreadsheet provided by Odinn (https://docs.google.com/spreadsheets/d/1ZW5UBk5BwcrEhh4mEsWl_H1Y9QGTaMPaT4v6dSEIBIo/edit#gid=0)?
For example 2003 McGrady being a +1.1 on your link and a +4.43 on the second link. It'a gigantic difference.

well first see above in that
1) Others are far better placed in their understanding on this than me.
and
2) There isn't one defenitive RAPM, it's more, as I understand it a process.

My best guess is that RAPM output (and process) is kind of an interlocking puzzle. It's adjusting for how good it thinks your teammates are. I looked at an older JE RAPM and it was much more bullish on McGrady and much less so on DeClerq than either version from the source I noted (indeed clearly lower on DeClerq than the link you provide which I have saved also as a JE RAPM).

Different versions often seem to scale differently which might account for some of the difference "my" leader in '03 is Duncan at 6.1, the second link (JE via Odinn) has Duncan too, but at 9.11.

The curious thing is that in my source McGrady come out significantly worse in the PI version (both raw number but also rank) than NPI which is odd in that he'd had an on-off north of +10 for the previous two years so it seems odd that his prior should drag him down.

Final notes:
- you have to figure out which sources you trust and how you'd test credibility;
- you have to figure out how large a sample you need to trust on-off type measures;
- this hasn't always been helped by unclear labeling;
- the NBA poaching/hiring and gagging what are presumably most of the better minds in the metrics field probably doesn't help either;
- fwiw, looking at this makes me a little less confident in the source I cited.

edit cf: viewtopic.php?f=64&t=2152308&start=80#p96017970 for a better informed, very recent post touching on RAPM issues and its assumptions.

[IdolW0rm]

Caveats about noise, contextual value rather than necessarily pure player goodness etc would have to prefix any conclusions ... Also there isn't one definitive RAPM, others knowing more than me can dive deeper if you/they desire.

https://sites.google.com/site/rapmstats/97-14-rapm-2
https://sites.google.com/site/rapmstats/97-14-rapm

The above covers most of Kobe's prime and all of Kobe's useful years. It covers his same generation guys (Duncan, Garnett, Nowitzki) and adjacent with substantial crossover guys (Shaq, LeBron). I believe it to be reliable (but am open to being wrong on this).

If you were to heavily buy into it (this version of RAPM) as your sole or primary (with a substantial weight) player evaluation tool, then personally I think it would kill the idea of Kobe top 5 all time (and substantially hurt top 10 all time), placing, as it does, Kobe outside the top 5 (6th, below the guys mentioned above) in value added above average in this span, and rate wise significantly lower (so for instance if you think Stockton kept up or increased his impact in earlier years, his career would come out ahead). You can eke him ahead of Shaq (within that span, note that this is ignoring Shaq's first 3 years, Kobe's last two, not necessarily a fair swap) in value above replacement (people may differ on the value being "above replacement"). The above idea doesn't lock him out of top 5 in era, but I say, if heavily weighted it would be very tricky to get him top 5 all time (and as I say hard for top 10 as, with these numbers as your baseline, it might be bullish to see him as even with Nowitzki, or put another way there are a lot of good players with the bulk or entirity of their career outside this window).

People's interpretations may differ, few will put that level of stock into one number and people's criteria for GOAT listings differ (longevity of quality/championship probability added versus peak; playoff weighting; interpreting eras; level of play versus narrative significance and ring; trust in accolades as a measure of level of play etc). Other calculations of RAPM may look different. (I wouldn't know what to do with it, but fwiw, his raw playoff on-off numbers are better than RS and you'd think the impact family in general is kinder in the playoffs - though this version of playoff RAPM - can't speak to reputational accuracy - isn't too bullish
https://docs.google.com/spreadsheets/d/e/2PACX-1vQdG8Zv84zqKEzETDjd8KPsClcw9bPETX9v_x_KEAxjv9NrFaWikOoiSaciy1jbMiygg2D-V8DUQn0O/pubhtml?gid=112475182&single=true
cf: viewtopic.php?f=64&p=76043070, https://www.reddit.com/r/nba/comments/db5hpt/19982019_playoffs_rapm_1_lebron_james_1_draymond/)

RAPM offers little on LeBron versus Jordan as RAPM only exists for the last 2 years of the second 3-peat and Wizards Jordan.

That's a pretty superficial, surface level take. I imagine others who understand it all better could tell you more.

Why do enourmous variations happen between the first link you provided and the spreadsheet provided by Odinn (https://docs.google.com/spreadsheets/d/1ZW5UBk5BwcrEhh4mEsWl_H1Y9QGTaMPaT4v6dSEIBIo/edit#gid=0)?
For example 2003 McGrady being a +1.1 on your link and a +4.43 on the second link. It'a gigantic difference.

well first see above in that
1) Others are far better placed in their understanding on this than me.
and
2) There isn't one defenitive RAPM, it's more, as I understand it a process.

My best guess is that RAPM output (and process) is kind of an interlocking puzzle. It's adjusting for how good it thinks your teammates are. I looked at an older JE RAPM and it was much more bullish on McGrady and much less so on DeClerq than either version from the source I noted (indeed clearly lower on DeClerq than the link you provide which I have saved also as a JE RAPM).

Different versions often seem to scale differently which might account for some of the difference "my" leader in '03 is Duncan at 6.1, the second link (JE via Odinn) has Duncan too, but at 9.11.

The curious thing is that in my source McGrady come out significantly worse in the PI version (both raw number but also rank) than NPI which is odd in that he'd had an on-off north of +10 for the previous two years so it seems odd that his prior should drag him down.

Final notes:
- you have to figure out which sources you trust and how you'd test credibility;
- you have to figure out how large a sample you need to trust on-off type measures;
- this hasn't always been helped by unclear labeling;
- the NBA poaching/hiring and gagging what are presumably most of the better minds in the metrics field probably doesn't help either;
- fwiw, looking at this makes me a little less confident in the source I cited.

Thanks for the input.
From what I've understood, PI RAPM isn't seen as a good single season evaluation tool and I wonder why that is, since the vast majority of the values (season ranks), seem to match the eye-test pretty good..

[ceoofkobefans]

In 5 year AuPM/RAPM from 1994-2017 Kobe Bryant ranked t10th at a +6.6 with Michael Jordan (who technically didn’t qualify with only 3 years). And was only .1 points behind Dwyane wade who was 9th (if you include guys like Br magic Wilt and bird who would likely be as good or better than Kobe he gets closer to 15). Ben Taylor’s 1yr Scaled APM has Kobe as the t11th best peak with Christian Laettner at +5.8 (If you can’t tell this can be kinda wonky in a single year sample). I have been working on a Multi year peaks database and currently have ≈ 70 players tracked. I collected multi year scaled APM and Prior Informed RAPM (averaged each single year While weighting for games played). In this Database (which is only all time peaks) 06-09 Kobe ranked 9th of 25 players in Scaled APM at a +5.3 just below 01-05 KG at a +5.4 and 13th/25 at a +5.34. If you include AuPM in the Samples he’s 10/32 And 14th/32. So The conclusion is that APM data Will likely range Kobe in the lower end of the top 15 peaks (this isn’t 100% accurate Tho and shouldn’t be taken as fact. I personally have his peak closer to 10 but I could see having him at 13 for peaks.)