RAPM factors (and more)

[Crow]

I count 16 players last season with 3 RAPM factors at or above +1. Some names might surprise, like Isaac, Landale, Capela, Looney, Gobert, Camara and Herb Jones.

2 is still uncommon. Not all big names even have 2, including Jokic and Embid. Doncic only has 1 but is just .01 from second. J Butler has 1. Adebayo 1. Harden 1. Zion W. none.

Markannen is closest to 4 but officially only has 2.

Checking for most famous with 2 -1s or worse, Porzingis is first found. Camara almost has 2 of these to go along with his 3 +1s.
I Joe is 2 positive, amost 2 negative by that amount. Okogie is close to 3 - 2 but officially 2-1. Kessler is 2-2.

LeVert P Watson, Niang,
Drummond, Zubac,
Eubanks, Mathurin, J Harris,
Achiuwa, D Sharpe and even more have 2 -1s. Vucevic is 2-3.

There are some single factor -3s out there. Of course some +3 factors as well. Plenty to see and consider.

[TeemoTeejay]

Embiid and Post GSW KD impact on OREB is interesting since that seems fairly consistent

[Crow]

Both -1s on that.

NBA.com has Durant at zero boxouts on offensive glass in regular season and playoffs. Embid 127th on o-boxout% in regular season, 62nd in playoffs but the competition was cut in half and the rate was cut to a measly 15%.

Different strategies out there but don't appear successful here.

Durant probably rarely close enough to try. Embid, could be different explanations. Including others not trying hard when sharing court with him and him having a 40% usage. 2.5% lesser team o-reb% with him than without.

Helpful to know when RAPM factors are consistent with other checked data.

[Crow]

Jalen Williams, slightly negative last season here. +1.5 by KMedved time-decayed RAPM. Not a huge difference but it is a difference to note and compare with everything else.

Close agreement on positive defensive impact. Difference is entirely on offensive impact estimate with Darko at about +0.5 while he is almost -1 here.

Darko sees JDub as good on offense, on boxscore component and giving back some outside the boxscore. This RAPM factor data suggests the issue is with oreb impact and ts% impact (at team level). The former is believable given the 1.9% oreb%. The latter is more mysterious.

EPM loves JDub on both sides of the court, way beyond these previous metrics. LeBron estimates him between the 2 RAPMs, making EPM the huge outlier.

What do others think and do from here if anything?


Collin Sexton, EPM is again the outlier but more moderately. This RAPM is a huge outlier on defense but in the middle on offensive estimate. The simple consensus average would be about +1 player, while the range is from more than +2 to slightly negative. Either expression seems reasonable to use, over a single metric.

The importance of blends.

Crafted weighted blend is +0.6.
Average the blends and he comes out as +0.8. I'll go with that.

For Adebayo it is a closer set. +2.1 by simple sum of 4, +2.8 by Crafted blend. EPM was almost the high outlier of 4 again missing by 0.1. The average of the 2 blends is +2.4.

I didn't include BPM on these but might in future.


A machine learning optimized metric blend would be what I'd want. For a team, for the few anywhere who would study it. Including at least me.

[TeemoTeejay]

Jalen Williams, slightly negative last season here. +1.5 by KMedved time-decayed RAPM. Not a huge difference but it is a difference to note and compare with everything else.

Close agreement on positive defensive impact. Difference is entirely on offensive impact estimate with Darko at about +0.5 while he is almost -1 here.

Darko sees JDub as good on offense, on boxscore component and giving back some outside the boxscore. This RAPM factor data suggests the issue is with oreb impact and ts% impact (at team level). The former is believable given the 1.9% oreb%. The latter is more mysterious.

EPM loves JDub on both sides of the court, way beyond these previous metrics. LeBron estimates him between the 2 RAPMs, making EPM the huge outlier.

What do others think and do from here if anything?


Collin Sexton, EPM is again the outlier but more moderately. This RAPM is a huge outlier on defense but in the middle on offensive estimate. The simple consensus average would be about +1 player, while the range is from more than +2 to slightly negative. Either expression seems reasonable to use, over a single metric.

The importance of blends.

Crafted weighted blend is +0.6.
Average the blends and he comes out as +0.8. I'll go with that.

For Adebayo it is a closer set. +2.1 by simple sum of 4, +2.8 by Crafted blend. EPM was almost the high outlier of 4 again missing by 0.1. The average of the 2 blends is +2.4.

I didn't include BPM on these but might in future.


A machine learning optimized metric blend would be what I'd want. For a team, for the few anywhere who would study it. Including at least me.

This is all 2 year samples, the fact that season 1 is 2 years before on the columns in hindsight was a bit of a dumb choice on my end (Since 2022 to 2024 would imply its 2021-22 to 2023-24, but I just thought of 2023-24 and overthought it lol)

[Crow]

Dating and explanation of it could be clearer.

I assumed "2022 2024" was '22-23 and '23-24. Correct or no?

[Crow]

MAMBA has Jalen Williams at +3.5, close to EPM.

Sexton at almost +1.9.

Adebayo at +2.8.



BPM has Jalen Williams at +2.3.

Sexton at +1.9.

Adebayo at +2.4.


Overall the fit is closest on Adebayo, moderately broad on Sexton and wide on Williams.

A full comparison table is needed imo. Crafted NBA breaks out its component measurements. Another comparison may become known in near future. But some newer / lesser known measures would take lobbying for inclusion or a new table.

[Crow]

Compare some but not all 4 factor RAPM estimates here to craftedNBA at this page, https://craftednba.com/player-stats
advanced stat filter selection.

For Jalen Williams the factor estimates are mostly kind of close. rts% at Crafted goes completely different direction than here. The scale is probably different.

A quick comparison of 6 more (DeRozan, Sabonis, Banchero, Durant, Edwards and Jokic) shows kinda close on 3 and way way different on 3.

This needs more general comparison and attention to differences if it is to be used.
Any chance author will speak to author?



Compare some but not all overall metrics on the plus minus filter selection.

Looking just at Jokic, Embid and Doncic, CraftedPM blend, MAMBA, BPM and Darko all put in same order. BPM showed them as more extreme outliers than the others, around 50% more.

[TeemoTeejay]

Dating and explanation of ot could be clearer.

I assumed "2022 2024" was '22-23 and '23-24. Correct or no?

Yeah ur right

[TeemoTeejay]

Going to try to do it with a few more years in the data, have to start from scratch but shouldnt take too long either way.

[Crow]

SGA, strong on many things but 10th worst defense rebounding factor in league. I Joe 4th worst. Giddey probably affected. Will want to see how it goes now. Caruso probably not helping. Will SGA & Joe take much more load or will Thunder rebounding stay very low?


I Jackson, 5th worst. Naz Reid 14th worst.

Westbrook 44th lowest. In 2017 he was barely top 100. In 2020 barely negative. Looks so far like a slide. Yep. 2022 fits that pattern too.

Hundreds or thousand of tidbits to possibly check. How many team personnel have done how much or ever will?

Tales from the deep.

Brandon Boston, 7th worst offensive TS% factor estimate. Spurs know or not? Care or not? Make team or not?

Pokusevski 9th worst. Bye. Fournier, 11th worst. Bye. Jabari Smith 12th worst. Ooh so much potential, future star... 19th worst overall here.

Wiseman 2nd worst overall estimate. Dick 5th worst. Z Williams 7th worst. S Henderson 8th worst. Reddish 15th worst. J Poole 35th worst. Tell their GM or no? Depends on your position and posture.

Under-recognized gems by this measure? Dean Wade, Landale. Others if you want to look or want to ask me to look.

K Ellis about 110th, one of the better new guys. All from defense, almost all from scoring defense. Not a surprise but confirming.

[Crow]

Giddey was only +0.25 on defensive rebounding RAPM factor. So he may not be missed much. But he also wasn't holding a lot of value possibly at the expense of others.

[Crow]

A bit over 40 players were -2 or worse on estimated offensive impact here and same on defense. The only three that low on both were Nwora, Dick and Pokusevski. Raptors had 2 of them. Might not have grasped this by eyesight or surface review of discrete stats. Came out as this low on this metric and about half the others. The other half were low but not the lowest.


Only 9 over +2 both ways. Celtics have 2 (White and Holiday). Clippers had 2, now the Sixers do. Nuggets, Lakers, Pelicans (H Jones) and Cavs have the others. None for Knicks, Bucks, Thunder, Mavs, TWolves or anybody else. May have a superstar but not this level of 2 way balanced impact. Fwiw.

[TeemoTeejay]

Giddey was only +0.25 on defensive rebounding RAPM factor. So he may not be missed much. But he also wasn't holding a lot of value possibly at the expense of others.

The rebounding RAPM for this is gonna be off compared to the new one I posted