2025-2026 College RAPM Estimates

[CptCrunch]

Decided to crunch out a proper RAPM for this season tonight. RAPM is what BPM tries to estimate from box-scores alone for the NBA.

Challenges in College RAPM Estimation

1) Sparse Data Sets : College-level possession data is significantly limited; even in the best-case scenarios, we are often working with only ~2,500 possessions.
2) Most other RAPM with Opinionated Priors are wrong: Many existing college RAPM estimates are fundamentally flawed because they use non-diffuse priors. These priors tend to use BPM derived metrics as as opposed to allowing the data to speak for itself.
3) High Statistical Noise: Due to the shorter season and limited lineup rotations, college RAPM is inherently noisier than its NBA counterparts.
4) Non-Standard Data: RAPM requires scraping of possession by possession level data. I scraped data from 860,625 possessions, over 6,135 Div games with 4,894 college players to calculate this for the very good people of RealGM.
5) Lack of Standardized Estimators: Because of the technical and conceptual hurdles mentioned above, "proper" college RAPM is rarely produced, making it a difficult metric to calculate accurately.


Here are the top 50 players; some class labels may be wrong in case that player played Div II/JUCO. You can blame ESPN for this as their API doesn't correctly label class. I manually overwrote Yexel fyi.

Code: Select all

 #    Player                       Yr      RAPM   Poss
  --------------------------------------------------
  1    MJ Collins Jr.               Sr     +5.95  2,497
  2    Keaton Wagler                Fr     +5.39  2,558
  3    Yaxel Lendeborg              Gr     +5.39  2,422
  4    Cameron Boozer               Fr     +4.64  2,663
  5    Ja'Kobi Gillespie            Sr     +4.56  2,653
  6    Henri Veesaar                Sr     +4.55  2,325
  7    Bennett Stirtz               So     +4.54  2,848
  8    Greedy Williams              Sr     +4.51  2,166
  9    Rafael Castro                Sr     +4.43  1,772
  10   Joshua Jefferson             Sr     +4.43  2,724
  11   Devin McGlockton             Sr     +4.39  2,093
  12   Johann Grunloh               Fr     +4.36  1,513
  13   Motiejus Krivas              Jr     +4.31  2,118
  14   C.J. Cox                     So     +4.29  1,984
  15   Mark Mitchell                Sr     +4.27  2,615
  16   Cameron Carr                 Jr     +4.18  2,549
  17   Rueben Chinyelu              Jr     +4.17  2,257
  18   Alex Karaban                 Sr     +4.17  2,465
  19   Armani Mighty                Sr     +4.13  2,415
  20   Joshua Dent                  So     +4.11  2,644
  21   Aday Mara                    Jr     +4.10  1,909
  22   Paulius Murauskas            Jr     +4.10  2,440
  23   Houston Mallette             Gr     +4.08  1,805
  24   Oscar Cluff                  Jr     +4.06  1,966
  25   Nolan Winter                 Jr     +4.02  2,121
  26   Duke Brennan                 Sr     +3.97  2,126
  27   Chris Cenac Jr.              Fr     +3.96  2,015
  28   Jaxon Kohler                 Sr     +3.90  2,081
  29   Mario Saint-Supery           Fr     +3.87  1,804
  30   Baba Miller                  Sr     +3.85  2,440
  31   Boogie Fland                 So     +3.81  2,616
  32   David Green                  Jr     +3.81  1,979
  33   Caleb Wilson                 Fr     +3.78  1,871
  34   Evan Mahaffey                Sr     +3.78  2,188
  35   Brayden Burries              Fr     +3.77  2,582
  36   Quentin Jones                Jr     +3.73  2,092
  37   Themus Fulks                 Gr     +3.72  2,589
  38   Somtochukwu Cyril            So     +3.70  1,832
  39   Pablo Tamba                  Gr     +3.66  2,111
  40   Tyon Grant-Foster            Gr     +3.63  1,759
  41   Braden Smith                 Sr     +3.62  2,918
  42   Trey McKenney                Fr     +3.53  1,706
  43   Killyan Toure                Fr     +3.52  2,001
  44   Dillon Mitchell              Sr     +3.52  2,198
  45   Jaden Bradley                Sr     +3.52  2,662
  46   Trevor Mullin                Jr     +3.52  1,728
  47   Jeremy Fears Jr.             Jr     +3.51  2,331
  48   Koa Peat                     Fr     +3.51  2,243
  49   Michael Osei-Bonsu           So     +3.51  1,973
  50   Flory Bidunga                So     +3.49  2,538


Here are the Tankathon top 30 mock

Code: Select all

 Pick  Player                       Yr      RAPM   Poss
  ------------------------------------------------------
  1     AJ Dybantsa                  Fr     +0.49  2,871
  2     Darryn Peterson              Fr     -0.16  1,581
  3     Cameron Boozer               Fr     +4.64  2,663
  4     Caleb Wilson                 Fr     +3.78  1,871
  5     Kingston Flemings            Fr     +1.74  2,621
  6     Keaton Wagler                Fr     +5.39  2,558
  7     Nate Ament                   Fr     +0.18  2,190
  8     Darius Acuff Jr.             Fr     +3.45  3,090
  9     Mikel Brown Jr.              Fr     +2.05  1,552
  10    Labaron Philon Jr.           So     +2.23  2,578
  11    Hannes Steinbach             Fr     +1.73  2,540
  12    Braylon Mullins              Fr     -0.43  1,784
  13    Patrick Ngongba II           So     +2.01  1,509
  14    Yaxel Lendeborg              Gr     +5.39  2,422
  15    Brayden Burries              Fr     +3.77  2,582
  16    Jayden Quaintance            So     -0.51    161
  17    Thomas Haugh                 Jr     +3.27  2,853
  18    (Karim López — intl)
  19    Bennett Stirtz               So     +4.54  2,848
  20    Christian Anderson           So     +2.54  2,915
  21    Tounde Yessoufou             Fr     +0.22  2,438
  22    Koa Peat                     Fr     +3.51  2,243
  23    Chris Cenac Jr.              Fr     +3.96  2,015
  24    Joshua Jefferson             Sr     +4.43  2,724
  25    Aday Mara                    Jr     +4.10  1,909
  26    Cameron Carr                 Jr     +4.18  2,549
  27    Morez Johnson Jr.            So     +2.59  2,117
  28    Tyler Tanner                 So     +3.20  2,741
  29    Amari Allen                  Fr     +1.64  2,119
  30    Ebuka Okorie                 Fr     +2.74  2,474


Previous list, sorted

Code: Select all

Pick  Player                       Yr      RAPM   Poss
  ------------------------------------------------------
  6     Keaton Wagler                Fr     +5.39  2,558
  14    Yaxel Lendeborg              Gr     +5.39  2,422
  3     Cameron Boozer               Fr     +4.64  2,663
  19    Bennett Stirtz               So     +4.54  2,848
  24    Joshua Jefferson             Sr     +4.43  2,724
  26    Cameron Carr                 Jr     +4.18  2,549
  25    Aday Mara                    Jr     +4.10  1,909
  23    Chris Cenac Jr.              Fr     +3.96  2,015
  8     Darius Acuff Jr.             Fr     +3.45  3,090
  17    Thomas Haugh                 Jr     +3.27  2,853
  22    Koa Peat                     Fr     +3.51  2,243
  15    Brayden Burries              Fr     +3.77  2,582
  4     Caleb Wilson                 Fr     +3.78  1,871
  28    Tyler Tanner                 So     +3.20  2,741
  27    Morez Johnson Jr.            So     +2.59  2,117
  20    Christian Anderson           So     +2.54  2,915
  30    Ebuka Okorie                 Fr     +2.74  2,474
  10    Labaron Philon Jr.           So     +2.23  2,578
  9     Mikel Brown Jr.              Fr     +2.05  1,552
  13    Patrick Ngongba II           So     +2.01  1,509
  29    Amari Allen                  Fr     +1.64  2,119
  5     Kingston Flemings            Fr     +1.74  2,621
  11    Hannes Steinbach             Fr     +1.73  2,540
  1     AJ Dybantsa                  Fr     +0.49  2,871
  7     Nate Ament                   Fr     +0.18  2,190
  21    Tounde Yessoufou             Fr     +0.22  2,438
  2     Darryn Peterson              Fr     -0.16  1,581
  12    Braylon Mullins              Fr     -0.43  1,784
  16    Jayden Quaintance            So     -0.51    161
  18    (Karim López — intl)


For context, here are the percentiles, a 4.16 puts a player in the top 1% in college in impact. This estimator has a lot more signal than I thought.

Code: Select all

  p1  : -3.22
  p10 : -1.62
  p25 : -0.71
  p50 : +0.30
  p75 : +1.38
  p90 : +2.43
  p99 : +4.16


[CptCrunch]

No hindsight bias, but I did put Keaton #2 on my big board weeks ago.

viewtopic.php?f=38&t=2499680

[ReggiesKnicks]

RAPM already struggles with sample size issues in the NBA where you need 3-5 year samples.

Good work, but rather useless.

[CptCrunch]

RAPM already struggles with sample size issues in the NBA where you need 3-5 year samples.

Good work, but rather useless.

Nice parrot speech, do tell us how one calculates multiple year RAPM for college freshmen.

Fact: there is nothing fundamentally wrong with one year RAPM. The noise you are referring to here is a disagreement of results with accepted perceptions.

Because of all possession based on off metrics in college is essentially a mixture of RAPM and some BPM-like metric.

Just pointing out that ~2/3 of the first round right now are players with 90 percentile plus impact in college. There is clear signal even in one year RAPM. I'm not saying this is a golden source of truth, neither is BPM or any other box score based metric out there, these are signals. You read all the signals including reports from scouts who don't even know middle school algebra to come to a final ranking.

[ReggiesKnicks]

RAPM already struggles with sample size issues in the NBA where you need 3-5 year samples.

Good work, but rather useless.

Nice parrot speech, do tell us how one calculates multiple year RAPM for college freshmen.

You don't. That's the point

Fact: there is nothing fundamentally wrong with one year RAPM. The noise you are referring to here is a disagreement of results with accepted perceptions.

Because of all possession based on off metrics in college is essentially a mixture of RAPM and some BPM-like metric.

Just point out that 2/3 of the first round right now are players with 90 percentile plus impact in college.

There is nothing wrong with one year, 30.game RAPM samples. I agree. There is nothing wrong with looking at points scored or rebounds or steals or blocks or 3P%. I agree.

The problem arises when we fixate too much on a singular data point.

I appreciate your work. I appreciate your time. This is a singular data point no different than 100s we already have.

[Caneman786]

Decided to crunch out a proper RAPM for this season tonight. RAPM is what BPM tries to estimate from box-scores alone for the NBA.

Challenges in College RAPM Estimation

1) Sparse Data Sets : College-level possession data is significantly limited; even in the best-case scenarios, we are often working with only ~2,500 possessions.
2) Most other RAPM with Opinionated Priors are wrong: Many existing college RAPM estimates are fundamentally flawed because they use non-diffuse priors. These priors tend to use BPM derived metrics as as opposed to allowing the data to speak for itself.
3) High Statistical Noise: Due to the shorter season and limited lineup rotations, college RAPM is inherently noisier than its NBA counterparts.
4) Non-Standard Data: RAPM requires scraping of possession by possession level data. I scraped data from 860,625 possessions, over 6,135 Div games with 4,894 college players to calculate this for the very good people of RealGM.
5) Lack of Standardized Estimators: Because of the technical and conceptual hurdles mentioned above, "proper" college RAPM is rarely produced, making it a difficult metric to calculate accurately.

Great work CptCrunch, I enjoyed reading it a lot and poring over the data. Although it is one singular datapoint, I think it's pretty useful and has a place on this board. And it seems pretty accurate, too for what you tried to measure (as in there's no serious flaws in your formula that you made).

Keaton Wagler (as you mentioned!) stands out here, and he has an incredibly high Luck-Adjusted RAPM on Hoop-Explorer (and in fact he ranks third in the class with +13.0) compared to his BPM of +10.8 (he ranks 17th), so that makes it look accurate.

Pretty we can kind of calculate the whole list if we compare raw Bart-Torvik BPM to Luck-Adjusted RAPM and seeing where the largest deltas are and which way they go. Darryn Peterson really stands out here being negative (!) while he has the 7th highest BPM (at +12.3) of any D1 college player in Bart-Torvik's database, and that's reflected in his #216 rank in Luck-Adjusted RAPM. The most striking example of course is the leader MJ Collins Jr. posting almost +6 in nearly 2,500 possessions. He's +5.5 in Bart-Torvik BPM (ranked 217 of 2,275 qualifiers). Meanwhile in Luck-Adjusted RAPM MJ Collins Jr. is #18 at +9.8.

For some draft prospects I'm interested in who are more fringe, I'd expect Allen Graves to be slightly positive but close to 0, Dailyn Swain maybe a little bit above him, while Milan Momcilovic and Adrian Wooley would be quite high (maybe even right outside the top 50). There's a good chance these aren't true and I'd love to know their actual values if you're fine with it.

Something to mention is that raw impact has a relation to the player's role (especially in these small sample sizes and when they're just getting started playing in the higher levels), and I think that's one of the reasons we've seen Kon Knueppel be so successful so far (his 1-year RAPM so far in fact is 91st percentile). I think Darryn Peterson has a bunch of valid excuses for coming up negative (or at least not very positive like you'd expect), being his injury and his lack of time to mesh with the team as a whole. Kansas has done some pretty amazing things, and even beat Arizona (when they were the 1 ranked team) without Peterson, so I imagine the chemistry was pretty good being a huge part of it. Sometimes they flow way worse with him (and I don't think it's completely his fault, but probably something to do with how much time he's spent apart from the team, but it is a red flag of course).

Another thing is that we don't have the real confidence intervals, which would be very helpful in trying to interpret the data. Really, I expect most of the confidence intervals, except a few players (such as probably Wagler, Lendeborg, and Boozer), to covering the range of neutral (like an interval of -3 to +5 or something). I don't think we can learn anything at all from Quaintance's RAPM as well, the samples play a role in it.

If anyone else is interested in this type of work (maybe even you CptCrunch don't know about it yourself?), there's one other guy who calculated pure college RAPM was former Mavericks staff and one of the primary authorities on basketball analytics today, Jeremias Engelmann, who did it for last season (at the end of the season, so after the tournament). He may even have been the first to do it like that.

Here's the link to his work: https://www.roycewebb.com/p/introducing-adjusted-plus-minus-for (luckily a free article!)

I don't think the actual values are public (although he clearly calculated both the offensive and defensive values, which might be difficult for you to do). However, he put some rankings, these notable ones:

2nd nationally (1st in the draft class). Kam Jones (a senior at Marquette)
4th nationally (2nd in the draft class). Cooper Flagg
6th nationally (3rd in the draft class). Ace Bailey
7th nationally (4th in the draft class). Nique Clifford
8th nationally (5th in the draft class). Kon Knueppel
10th nationally (6th in the draft class). Tyrese Proctor
...
19th nationally (10th in the draft class). Will Riley
...
42nd nationally (14th in the draft class). Derik Queen
50th nationally (15th in the draft class) (1st in defense among all players nationally!). Thomas Sorber

Some players that stood out negatively were Asa Newell (only negative player who was drafted in the first round!), Tre Johnson, and VJ Edgecombe.

I thought it was a good indicator showing Ace Bailey being very underrated, and Dylan Harper being overrated and getting carried by Ace. The box numbers were definitely missing something with those two (of course this is just my opinion, and right now in the rookie season Harper seems to be outperforming Ace in impact, I suspect it will flip soon).

[CptCrunch]

Decided to crunch out a proper RAPM for this season tonight. RAPM is what BPM tries to estimate from box-scores alone for the NBA.

Challenges in College RAPM Estimation

1) Sparse Data Sets : College-level possession data is significantly limited; even in the best-case scenarios, we are often working with only ~2,500 possessions.
2) Most other RAPM with Opinionated Priors are wrong: Many existing college RAPM estimates are fundamentally flawed because they use non-diffuse priors. These priors tend to use BPM derived metrics as as opposed to allowing the data to speak for itself.
3) High Statistical Noise: Due to the shorter season and limited lineup rotations, college RAPM is inherently noisier than its NBA counterparts.
4) Non-Standard Data: RAPM requires scraping of possession by possession level data. I scraped data from 860,625 possessions, over 6,135 Div games with 4,894 college players to calculate this for the very good people of RealGM.
5) Lack of Standardized Estimators: Because of the technical and conceptual hurdles mentioned above, "proper" college RAPM is rarely produced, making it a difficult metric to calculate accurately.

Great work CptCrunch, I enjoyed reading it a lot and poring over the data. Although it is one singular datapoint, I think it's pretty useful and has a place on this board. And it seems pretty accurate, too for what you tried to measure (as in there's no serious flaws in your formula that you made).

Keaton Wagler (as you mentioned!) stands out here, and he has an incredibly high Luck-Adjusted RAPM on Hoop-Explorer (and in fact he ranks third in the class with +13.0) compared to his BPM of +10.8 (he ranks 17th), so that makes it look accurate.

Pretty we can kind of calculate the whole list if we compare raw Bart-Torvik BPM to Luck-Adjusted RAPM and seeing where the largest deltas are and which way they go. Darryn Peterson really stands out here being negative (!) while he has the 7th highest BPM (at +12.3) of any D1 college player in Bart-Torvik's database, and that's reflected in his #216 rank in Luck-Adjusted RAPM. The most striking example of course is the leader MJ Collins Jr. posting almost +6 in nearly 2,500 possessions. He's +5.5 in Bart-Torvik BPM (ranked 217 of 2,275 qualifiers). Meanwhile in Luck-Adjusted RAPM MJ Collins Jr. is #18 at +9.8.

For some draft prospects I'm interested in who are more fringe, I'd expect Allen Graves to be slightly positive but close to 0, Dailyn Swain maybe a little bit above him, while Milan Momcilovic and Adrian Wooley to be quite high (probably right outside the top 50). There's a good chance these aren't true and I'd love to know their actual values if you're fine with it.

Something to mention is that raw impact has a relation to the player's role (especially in these small sample sizes and when they're just getting started playing in the higher levels), and I think that's one of the reasons we've seen Kon Knueppel be so successful so far (his 1-year RAPM so far in fact is 91st percentile). I think Darryn Peterson has a bunch of valid excuses for coming up negative (or at least not very positive like you'd expect), being his injury and his lack of time to mesh with the team as a whole. Kansas has done some pretty amazing things, and even beat Arizona (when they were the 1 ranked team) without Peterson, so I imagine the chemistry was pretty good being a huge part of it. Sometimes they flow way worse with him (and I don't think it's completely his fault, but probably something to do with how much time he's spent apart from the team, but it is a red flag of course).

Another thing is that we don't have the real confidence intervals, which would be very helpful in trying to interpret the data. Really, I expect most of the confidence intervals, except a few players (such as probably Wagler, Lendeborg, and Boozer), to covering the range of neutral (like an interval of -3 to +5 or something). I don't think we can learn anything at all from Quaintance's RAPM as well, the samples play a role in it.

If anyone else is interested in this type of work (maybe even you CptCrunch don't know about it yourself?), there's one other guy who calculated pure college RAPM was former Mavericks staff and one of the primary authorities on basketball analytics today, Jeremias Engelmann, who did it for last season (at the end of the season, so after the tournament). He may even have been the first to do it like that.

Here's the link to his work: https://www.roycewebb.com/p/introducing-adjusted-plus-minus-for (luckily a free article!)

I don't think the actual values are public, but he put some rankings, these notable ones:

2nd nationally (1st in the draft class). Kam Jones (a senior at Marquette)
4th nationally (2nd in the draft class). Cooper Flagg
6th nationally (3rd in the draft class). Ace Bailey
7th nationally (4th in the draft class). Nique Clifford
8th nationally (5th in the draft class). Kon Knueppel
10th nationally (6th in the draft class). Tyrese Proctor
...
19th nationally (10th in the draft class). Will Riley
...
42nd nationally (14th in the draft class). Derik Queen
50th nationally (15th in the draft class) (1st in defense among all players nationally!). Thomas Sorber

Some players that stood out negatively were Asa Newell (only negative player who was drafted in the first round!), Tre Johnson, and VJ Edgecombe.

I thought it was a good indicator showing Ace Bailey being very underrated, and Dylan Harper being overrated and getting carried by Ace. The box numbers were definitely missing something with those two (of course this is just my opinion).

I'm not sure if Hoop-Explorer is using real RAPM or is using a BPM-ish estimator, but with him having MJ Collins that high is likely an indicator that it's using RAPM instead of BPM type of stats. His RAPM is likely multiple year hence dragging MJ down bit.

I only ran these last night for the current season. Will produce multiple season RAPM soon and put on my site.

One note is that RAPM requires a player to win while on court. Counting stats don't matter. This may result in a large divergence between players with great stats like Peterson who didn't manage to win much in their limited minutes. Peterson's limited possessions in college, and I would argue shot jacking behavior for NBA audition hurt his team (and depressed his passing stats).

A note on why RAPM estimates don't produce confidence intervals.

1) in general advanced analytics don't show CIs (my site has BPM variance estimates called CPM, which you can use with CPM/sqrt(n) for standard errors).

2) RAPM uses ridge regression, this breaks statistical 'coverage' by intentionally introducing bias as to lower variance. All standard systems for estimating ridge regressions do not produce CI estimates as a result as these CI would not have a theoretical statistical interpretation. But this was on my radar. I have bootstrapped estimates for this season.

These are pseudo CI estimating the low and high bounds, do not interpret like a regular CI.

Code: Select all

 Pick  Player                       RAPM      95% CI         SE   Poss
  --------------------------------------------------------------------
  6     Keaton Wagler               +5.39  [+3.03, +7.44]   1.13  2,558
  14    Yaxel Lendeborg             +5.39  [+2.67, +8.20]   1.40  2,422
  3     Cameron Boozer              +4.64  [+2.81, +6.33]   1.01  2,663
  19    Bennett Stirtz              +4.54  [+2.39, +6.47]   1.13  2,848
  24    Joshua Jefferson            +4.43  [+1.97, +6.56]   1.14  2,724
  26    Cameron Carr                +4.18  [+1.92, +6.46]   1.11  2,549
  25    Aday Mara                   +4.10  [+2.09, +6.70]   1.19  1,909
  23    Chris Cenac Jr.             +3.96  [+1.50, +6.67]   1.39  2,015
  4     Caleb Wilson                +3.78  [+1.55, +6.37]   1.22  1,871
  15    Brayden Burries             +3.77  [+1.50, +6.17]   1.23  2,582
  22    Koa Peat                    +3.51  [+0.88, +5.88]   1.31  2,243
  8     Darius Acuff Jr.            +3.45  [+1.44, +5.22]   1.08  3,090
  17    Thomas Haugh                +3.27  [+0.94, +4.89]   0.98  2,853
  28    Tyler Tanner                +3.20  [+0.86, +5.24]   1.14  2,741
  30    Ebuka Okorie                +2.74  [+0.88, +5.20]   1.17  2,474
  27    Morez Johnson Jr.           +2.59  [-0.01, +5.21]   1.37  2,117
  20    Christian Anderson          +2.54  [+1.27, +3.73]   0.61  2,915
  10    Labaron Philon Jr.          +2.23  [-0.25, +4.84]   1.36  2,578
  9     Mikel Brown Jr.             +2.05  [-0.29, +4.72]   1.32  1,552
  13    Patrick Ngongba II          +2.01  [-0.09, +4.08]   1.16  1,509
  5     Kingston Flemings           +1.74  [+0.06, +3.77]   1.00  2,621
  11    Hannes Steinbach            +1.73  [-0.23, +3.08]   0.88  2,540
  29    Amari Allen                 +1.64  [+0.04, +3.32]   0.93  2,119
  1     AJ Dybantsa                 +0.49  [-1.23, +2.07]   0.85  2,871
  21    Tounde Yessoufou            +0.22  [-2.09, +2.45]   1.24  2,438
  7     Nate Ament                  +0.18  [-1.74, +2.18]   0.93  2,190
  2     Darryn Peterson             -0.16  [-2.64, +2.81]   1.42  1,581
  12    Braylon Mullins             -0.43  [-2.33, +1.52]   1.03  1,784
  16    Jayden Quaintance           -0.51  [-1.17, +0.50]   0.41    161



Format might be wonky. Posting from phone via ssh shell.

And yes Engelmann is the original RAPM creator for the NBA. When he was first creating RAPM. Everyone was effectively eating crayons.

[CptCrunch]

One note to mention is that raw RAPM isn't strength adjusted so schedule strength doesn't matter. Mj's positive contribution at USU is likely not as 'worthy' as Keaton's.

This is important in college because there is such talent disparity. This is a lot less problematic in the NBA.

[CptCrunch]

Full RAPM for 2025-2026 released on site, click link for shared view - http://vstatball.com?s=730a3f

Make sure to try on your phone. I have spent much effort optimizing for mobile.

Edit: RAPM released for all years back to 2009. For non freshman, RAPM is based on up to 3 years of data.

[King Ken]

One note to mention is that raw RAPM isn't strength adjusted so schedule strength doesn't matter. Mj's positive contribution at USU is likely not as 'worthy' as Keaton's.

This is important in college because there is such talent disparity. This is a lot less problematic in the NBA.

Eye test matters a lot. When you watched CJ, Dame, and Steph, they looked like lottery picks going into their last year and in their last year. Some guys dominate at levels just because they are better at that level but not at the next