RAPM: 5 year sample

[GeorgeMarcus]

Here are the top 15 in RAPM over that time period:

1. Steph Curry
2. Giannis Antetokounmpo
3. Chris Paul
4. Joel Embiid
5. Jayson Tatum
6. Paul George
7. Kawhi Leonard
8. Rudy Gobert
9. Kevin Durant
10. Nikola Jokic
11. Jrue Holiday
12. LeBron James
13. Jimmy Butler
14. Pascal Siakam
15. Damian Lillard

Every one of those players have been All Stars, and all but Jrue All-NBA. I don't think people appreciate how cool it is that pure impact metrics (with proper sample size) can produce results like this without incorporating a single box stat. No, it's not a "ranking" so FFS leave the Straw Man alone. No stat in existence can ever or will ever translate to a player ranking.

For the benefit of basketball enthusiasts everywhere, please limit your analysis to:
1) standard box stats (I lump eFG% and TS% under this umbrella because there's nothing advanced about measuring efficiency in such obvious ways)
2) impact metrics (+/-, Net Rtg, on/off, RAPM)
3) "tracking" data which NBA.com has recently expanded on big time, giving us very specific insights like efficiency from a specific area on the floor, screening effectiveness, rim protection, you name it; depending on what you're trying to accomplish, I caution people to realize there is a lot of important context that gets removed when isolating variables this way

"Awfully presumptuous of you to tell people what stats they should analyze, George." Yeah... I suppose I'm a bit of a ****. But I feel physical pain when I see arguments centered around BPM/VORP, WS, PER, and whatever flavor of new umbrella metrics they invent by the day. I don't believe any of those stats provide even a drop of analytical value. If you can't explain to someone what a stat is actually measuring then it loses all value. There aren't so many box stats that we have to fuse them together with arbitrary weighting systems. When those umbrella stats produce lists like the one above, it's only because people tweaked their convoluted equations until the results "look right" unlike RAPM which cares not for our subjective opinions.

[Duffman100]

Pascal Siakam eh?

[eminence]

I see that top 15 cut off to leave off Carushow!

[clippertown]

With the exception of Jrue and Saikam (though over 5 years maybe) it looks pretty good. Interesting to see that Paul George is ahead of Kawhi. To be honest, that is what my eyes tell me, even if my brain screams otherwise.

[GeorgeMarcus]

I see that top 15 cut off to leave of Carushow!

Shhhh they aren't suppose to know about him

[GeorgeMarcus]

Pascal Siakam eh?

Lowry passed Pascal the impact torch which has kept the Raps aflame

[GeorgeMarcus]

With the exception of Jrue and Saikam (though over 5 years maybe) it looks pretty good. Interesting to see that Paul George is ahead of Kawhi. To be honest, that is what my eyes tell me, even if my brain screams otherwise.

I think there is truth to Kawhi's value increasing in the playoffs relative to PG, mostly because his polished iso game gets featured more (and perhaps also being more "clutch"/mentally resilient). But I would also agree that regular season PG is at least as impactful as Kawhi, not even counting games missed.

[Chanel Bomber]

Here are the top 15 in RAPM over that time period:

1. Steph Curry
2. Giannis Antetokounmpo
3. Chris Paul
4. Joel Embiid
5. Jayson Tatum
6. Paul George
7. Kawhi Leonard
8. Rudy Gobert
9. Kevin Durant
10. Nikola Jokic
11. Jrue Holiday
12. LeBron James
13. Jimmy Butler
14. Pascal Siakam
15. Damian Lillard

Every one of those players have been All Stars, and all but Jrue All-NBA. I don't think people appreciate how cool it is that pure impact metrics (with proper sample size) can produce results like this without incorporating a single box stat. No, it's not a "ranking" so FFS leave the Straw Man alone. No stat in existence can ever or will ever translate to a player ranking.

For the benefit of basketball enthusiasts everywhere, please limit your analysis to:
1) standard box stats (I lump eFG% and TS% under this umbrella because there's nothing advanced about measuring efficiency in such obvious ways)
2) impact metrics (+/-, Net Rtg, on/off, RAPM)
3) "tracking" data which NBA.com has recently expanded on big time, giving us very specific insights like efficiency from a specific area on the floor, screening effectiveness, rim protection, you name it; depending on what you're trying to accomplish, I caution people to realize there is a lot of important context that gets removed when isolating variables this way

"Awfully presumptuous of you to tell people what stats they should analyze, George." Yeah... I suppose I'm a bit of a ****. But I feel physical pain when I see arguments centered around BPM/VORP, WS, PER, and whatever flavor of new umbrella metrics they invent by the day. I don't believe any of those stats provide even a drop of analytical value. If you can't explain to someone what a stat is actually measuring then it loses all value. There aren't so many box stats that we have to fuse them together with arbitrary weighting systems. When those umbrella stats produce lists like the one above, it's only because people tweaked their convoluted equations until the results "look right" unlike RAPM which cares not for our subjective opinions.

I genuinely don't understand what point - if any - you are making by sharing these RAPM rankings.

Are you simply marveling at how it captures impact based on it corroborating what you have seen in the last 5 years? Or do you think it is merely an object of curiosity? What is the value of RAPM in your mind? Do you mind elaborating?

I am more and more interested in this metric, thanks to dhilv (whose name I'm probably butchering, sorry for that). Although, as I mentioned in that debate, RAPM still seems context-dependent, and as we know context can change dramatically in a playoff series (from a regular season-long sample size), where smart coaches make adjustments that can affect the impact of a great player who lacks versatility (re: Gobert, to a lesser degree Giannis when he was used as the lead ball-handler and not the screener).

[NBA4Lyfe]

no real nba evaluator uses rapm to rank players.. the data is too hit or miss

[Doctor MJ]

Here are the top 15 in RAPM over that time period:

1. Steph Curry
2. Giannis Antetokounmpo
3. Chris Paul
4. Joel Embiid
5. Jayson Tatum
6. Paul George
7. Kawhi Leonard
8. Rudy Gobert
9. Kevin Durant
10. Nikola Jokic
11. Jrue Holiday
12. LeBron James
13. Jimmy Butler
14. Pascal Siakam
15. Damian Lillard

Every one of those players have been All Stars, and all but Jrue All-NBA. I don't think people appreciate how cool it is that pure impact metrics (with proper sample size) can produce results like this without incorporating a single box stat. No, it's not a "ranking" so FFS leave the Straw Man alone. No stat in existence can ever or will ever translate to a player ranking.

For the benefit of basketball enthusiasts everywhere, please limit your analysis to:
1) standard box stats (I lump eFG% and TS% under this umbrella because there's nothing advanced about measuring efficiency in such obvious ways)
2) impact metrics (+/-, Net Rtg, on/off, RAPM)
3) "tracking" data which NBA.com has recently expanded on big time, giving us very specific insights like efficiency from a specific area on the floor, screening effectiveness, rim protection, you name it; depending on what you're trying to accomplish, I caution people to realize there is a lot of important context that gets removed when isolating variables this way

"Awfully presumptuous of you to tell people what stats they should analyze, George." Yeah... I suppose I'm a bit of a ****. But I feel physical pain when I see arguments centered around BPM/VORP, WS, PER, and whatever flavor of new umbrella metrics they invent by the day. I don't believe any of those stats provide even a drop of analytical value. If you can't explain to someone what a stat is actually measuring then it loses all value. There aren't so many box stats that we have to fuse them together with arbitrary weighting systems. When those umbrella stats produce lists like the one above, it's only because people tweaked their convoluted equations until the results "look right" unlike RAPM which cares not for our subjective opinions.

So I quite like RAPM but I'm a bit confused here. Did you grab this from a site? If so, please share the link.

Regardless, are we talking about the RAPM from 2017-18 to 2021-22? Does it include the playoffs? How are the playoffs weighted? Is there a prior?

[Doctor MJ]

no real nba evaluator uses rapm to rank players.. the data is too hit or miss

That sounds like criticisms people use for small sample size, not for a 5 year model.

A 5 year model is going to have it's own issues - nothing is perfect - I generally find them helpful to see.

[uraverage]

Can we get a list of players that have played at least 60% of their games over the last 5 years?

[zimpy27]

Here are the top 15 in RAPM over that time period:

1. Steph Curry
2. Giannis Antetokounmpo
3. Chris Paul
4. Joel Embiid
5. Jayson Tatum
6. Paul George
7. Kawhi Leonard
8. Rudy Gobert
9. Kevin Durant
10. Nikola Jokic
11. Jrue Holiday
12. LeBron James
13. Jimmy Butler
14. Pascal Siakam
15. Damian Lillard

Every one of those players have been All Stars, and all but Jrue All-NBA. I don't think people appreciate how cool it is that pure impact metrics (with proper sample size) can produce results like this without incorporating a single box stat. No, it's not a "ranking" so FFS leave the Straw Man alone. No stat in existence can ever or will ever translate to a player ranking.

For the benefit of basketball enthusiasts everywhere, please limit your analysis to:
1) standard box stats (I lump eFG% and TS% under this umbrella because there's nothing advanced about measuring efficiency in such obvious ways)
2) impact metrics (+/-, Net Rtg, on/off, RAPM)
3) "tracking" data which NBA.com has recently expanded on big time, giving us very specific insights like efficiency from a specific area on the floor, screening effectiveness, rim protection, you name it; depending on what you're trying to accomplish, I caution people to realize there is a lot of important context that gets removed when isolating variables this way

"Awfully presumptuous of you to tell people what stats they should analyze, George." Yeah... I suppose I'm a bit of a ****. But I feel physical pain when I see arguments centered around BPM/VORP, WS, PER, and whatever flavor of new umbrella metrics they invent by the day. I don't believe any of those stats provide even a drop of analytical value. If you can't explain to someone what a stat is actually measuring then it loses all value. There aren't so many box stats that we have to fuse them together with arbitrary weighting systems. When those umbrella stats produce lists like the one above, it's only because people tweaked their convoluted equations until the results "look right" unlike RAPM which cares not for our subjective opinions.

So I quite like RAPM but I'm a bit confused here. Did you grab this from a site? If so, please share the link.

Regardless, are we talking about the RAPM from 2017-18 to 2021-22? Does it include the playoffs? How are the playoffs weighted? Is there a prior?

http://nbashotcharts.com/rapm5?id=-1129977058

[GeorgeMarcus]

Here are the top 15 in RAPM over that time period:

1. Steph Curry
2. Giannis Antetokounmpo
3. Chris Paul
4. Joel Embiid
5. Jayson Tatum
6. Paul George
7. Kawhi Leonard
8. Rudy Gobert
9. Kevin Durant
10. Nikola Jokic
11. Jrue Holiday
12. LeBron James
13. Jimmy Butler
14. Pascal Siakam
15. Damian Lillard

Every one of those players have been All Stars, and all but Jrue All-NBA. I don't think people appreciate how cool it is that pure impact metrics (with proper sample size) can produce results like this without incorporating a single box stat. No, it's not a "ranking" so FFS leave the Straw Man alone. No stat in existence can ever or will ever translate to a player ranking.

For the benefit of basketball enthusiasts everywhere, please limit your analysis to:
1) standard box stats (I lump eFG% and TS% under this umbrella because there's nothing advanced about measuring efficiency in such obvious ways)
2) impact metrics (+/-, Net Rtg, on/off, RAPM)
3) "tracking" data which NBA.com has recently expanded on big time, giving us very specific insights like efficiency from a specific area on the floor, screening effectiveness, rim protection, you name it; depending on what you're trying to accomplish, I caution people to realize there is a lot of important context that gets removed when isolating variables this way

"Awfully presumptuous of you to tell people what stats they should analyze, George." Yeah... I suppose I'm a bit of a ****. But I feel physical pain when I see arguments centered around BPM/VORP, WS, PER, and whatever flavor of new umbrella metrics they invent by the day. I don't believe any of those stats provide even a drop of analytical value. If you can't explain to someone what a stat is actually measuring then it loses all value. There aren't so many box stats that we have to fuse them together with arbitrary weighting systems. When those umbrella stats produce lists like the one above, it's only because people tweaked their convoluted equations until the results "look right" unlike RAPM which cares not for our subjective opinions.

I genuinely don't understand what point - if any - you are making by sharing these RAPM rankings.

Are you simply marveling at how it captures impact based on it corroborating what you have seen in the last 5 years? Or do you think it is merely an object of curiosity? What is the value of RAPM in your mind? Do you mind elaborating?

I am more and more interested in this metric, thanks to dhilv (whose name I'm probably butchering, sorry for that). Although, as I mentioned in that debate, RAPM still seems context-dependent, and as we know context can change dramatically in a playoff series (from a regular season-long sample size), where smart coaches make adjustments that can affect the impact of a great player who lacks versatility (re: Gobert, to a lesser degree Giannis when he was used as the lead ball-handler and not the screener).

The goal of basketball, above all, is to win. Possessions are the most granular unit in which we can evaluate winning, and RAPM expands on that framework by neutralizing the effect of teammates/opponents to isolate individual impact as well as one could hope to.

IMO context is no more or less important when evaluating impact data than it is box data. Somebody who judged a player only by his box contributions might have trouble with a guy like Ron Harper whose numbers plummeted on the Bulls despite being no worse of a player. Whereas the impact enthusiastic wouldn't bat an eye because his positive impact was comparable in both situations.

Part of my reason for posting this thread is in response to those who take small samples and say "see these stats are meaningless", when clearly the larger trends are very meaningful. In that sense I figured the 5-year maturation would be especially poignant.

My boy Embiid depicts another way this stat is useful. His impact numbers are staggering over his career, but in one outlier season (20-21) they were "good not great". Why was that the case you ask? This is where context becomes very important. In 20-21, the Sixers trotted out a lineup of Simmons/Richardson/Tobias/Horford/Embiid. Not only was the spacing on that team horrible, but defensively Embiid could no longer anchor the post while relying on 4 wing defenders to press the perimeter. Joel wasn't any worse of a player, but his impact on both ends of the court declined because his contributions were no longer as valuable as they were when the team complimented/accentuated his skillset.

I'm meandering a bit and not sure if I'm answering your question, but I'm happy to expand on specifics as much as you'd like me to.

[zimpy27]

Luck-adjusted 5 year RAPM for same period (2017-2022):

1. Giannis Antetokounmpo
2. Stephen Curry
3. Nikola Jokic
4. Rudy Gobert
5. Jayson Tatum
6. Chris Paul
7. Joel Embiid
8. Jimmy Butler
9. Paul George
10. LeBron James
11. Kawhi Leonard
12. Jrue Holiday
13. Damian Lillard
14. James Harden
15. Steven Adams

[JN61]

Seems pretty poor statistic to convey what it tries to.

[GeorgeMarcus]

Seems pretty poor statistic to convey what it tries to.

Based on what?

[JimmyPlopper]

Can we get a list of players that have played at least 60% of their games over the last 5 years?

it's the exact same list

[Doranku]

CP3 top 3, Gobert top 10, PG > Kawhi, Jokic at 10???, no Luka in sight... yeah I'm good.

[benhillboy]

Any metric that realizes Steph is a great team defender (when he wants to be) is cool with me since so many actual humans don’t. Jimmy should be second but obviously there’s a glitch somewhere lol.