2024-25 RealGM All-Season Awards Discussion Thread

[BusywithBball]

[quote=“Lessthanjake”]If OKC wins the Finals, then it seems extremely straightforward to me that SGA should be POY. Even if we think Jokic played a bit better overall this year (which I do actually agree with), being on a 68-win team that wins the title is a massive achievement and should easily allow him to leapfrog Jokic. It actually gets more complicated if the Thunder get upset in the Finals. At that point, I think Jokic has a shout for it, since IMO he’d have played a bit better and SGA wouldn’t have converted on having the best team. It’d still be close, though, because even just the 68-wins and SRS record was really impressive. But I very much doubt the Thunder will lose the Finals, so we very likely won’t have to really confront that question.

After SGA and Jokic, I do think Giannis was clearly the 3rd best player. If the Pacers win the title, then I think Haliburton gets #3 (assuming he’s good in the Finals, but it’s hard to imagine them winning without that being the case), because winning the title is a big deal. But otherwise, Giannis is #3 IMO. I think Haliburton has already done enough to be #4, so he will stay there unless he’s bad in a Finals loss. And I’d say #5 is Tatum. There might be an argument for Donovan Mitchell at #5 too. Both Tatum and Mitchell led great regular season teams, both of them were generally good overall in the playoffs but had a couple crucial bad games in the round their team lost. Tatum getting injured maybe weighs against him (though it does also mean his team didn’t quite lose a series with him), but overall I just think Tatum is a better player than Donovan, so I put him ahead. If Haliburton is bad in the Finals, then Tatum and Haliburton would probably be flipped.[/quote]
I agree alot and it reminds me of MVP. Jokic better yes but you have to applaud sixty-eight wins. I think even if OKC lose if Shai plays great like first 4 I can’t pick against.

I think Jokic might won too with this great team but what you do matters too.

[Doctor MJ]

Siakim has more impact?

Some indicators say so yes.

[Doctor MJ]

I think it was here we had a short discussion on replacement level.

<cool graph>

Folks will set it at the very worst handful of players in the league, when it should be after the sharp decrease at -1. It doesn't make any logical sense for the talent distribution to crater after -1 in between the 'just in the league' and 'just out of the league' level. And the play of the -1 guys in other leagues certainly doesn't suggest it. Much more likely that there are more -2 players in the world than -1 guys.

*Numbers are specific to this study, though pretty consistent to others I've seen/done. Notes on distribution hold across other numbers.

I don't remember if it was this thread or not, but I think I was involved and I've been thinking about it a lot lately.

I get your thinking that wherever the mode of the league is, that should be the replacement level. I'm not going to say it's wrong, but the higher you set the replacement level, the more guys who are going to rate as below replacement level, and the more awkward things can become.

So in my recent RAPM VORP study, I used a replacement level of -3.35. Here are guys I have playing more than 20K minutes with a negative cume VORP, along with their career earnings.

Jeff Green $102 mill
Ricky Davis $42 mill
Drew Gooden $67 mill
Kevin Martin $83 mill
Tristan Thompson $122 mill

Why include the money? Not to do a bashing of these guys, but just as a reminder:

Actual "replacement players" in the NBA get paid peanuts compared to these numbers. So this is the awkward thing:

How messed up is it that one guy makes $100 mill while apparently being worse at basketball than a horde of other guys - most of whom won't ever get rich playing ball? What's going on there?

I do think there's good conversation to be had about each of these guys, but I suppose my real concern comes when you use a much higher level. If I do the same query with a -1 replacement and I get way more guys, will we really believe it logical that all these guys spent their careers actively damaging their team on average compared to what a guy they could pick up from the G-league at any point?

Alright, here's that list then of guys I haveplaying 20K minutes who'd have a negative VORP with the -1 standard:

Spoiler:

Jeff Green
Ricky Davis
Drew Gooden
Kevin Martin
Tristan Thompson
Jarrett Jack
Juwan Howard
Luke Ridnour
Jason Williams
Samuel Dalembert
John Salmons
Larry Hughes
DJ Augustin
Jordan Clarkson
Antoine Walker
Nick Van Exel
Donyell Marshall
Shareef Abdur-Rahim
Erick Dampier
Jerami Grant
Mo Williams
Zach LaVine
Michael Finley
Al Jefferson
Damon Stoudamire
Jamal Crawford
Andrew Wiggins
Arron Afflalo
Jalen Rose
Clifford Robinson
Jim Jackson
Rajon Rondo
Carlos Boozer
Jonas Valanciunas
Joe Smith
Robin Lopez

I'll leave it without further comment right now.

Last note:

In theory, a team full of replacement level players should yield an SRS 5 times the player APM. So if replacement level is -1, then we should expect a team like that to be able to perform at -5 SRS. Level -2? -10 SRS. Level -3? -15 SRS.

For me, this is something pushing toward more extreme values of replacement level. I'm just really skeptical that you could create a new team out of leftovers, and expect that team to have a chance being better than a roster of some of the guys listed in the spoiler.

I can understand the example where an NBA team is emphasizing youth, coupled with injuries and a front office that would like a high draft pick, and how that could lead to a negative VORP team, so I'm not saying it's literally impossible for a real NBA team to be that bad, but I wouldn't say that's what was going on with the guys listed.

[Doctor MJ]

So a few kids have been telling me about impact statistic and RAPM but I haven’t seen it in 2025. Is Siakam better?

I don’t want to only look at stats but if Haliburton is not the #1 on floor I have to reconsider his greatness a little. I haven’t really thought about Tyrese vs Siakam much but Haliburton is assisting more and scoring almost same on better ts% and playing better on Thunder too.

Well, worth getting into some data, though I always caution that there's no one definitive measure.

By Englemann's RAPM for this year, here's how the main 5 Pacers look look:

Player Offense Defense Overall
Pascal Siakam 0.3 -2.1 2.4
Aaron Nesmith 2.6 0.2 2.4
Tyrese Haliburton 3.1 1.1 2.0
Andrew Nembhard -0.0 -1.7 1.6
Myles Turner 0.1 0.1 0.0

So, Pascal comes out the top defender and overall (apparently slightly ahead of Nesmith), while Haliburton comes out on top for offense. Makes sense. Also paints a picture that isn't super excited about these guys at all really. I mean, their finals opponent Shai tops the measure with +8.3, so clearly the way they're playing right now, the Pacers have found some new gears as the season has progressed.

Now to use a hybrid RAPM metric, here's the same for Season EPM Estimated Wins:

Tyrese Haliburton 4.2
Pascal Siakam 3.3
Aaron Nesmith 1.6
Myles Turner 0.7
Andrew Nembhard 0.6

So, this one favors Hali, fwiw.

[eminence]

I think it was here we had a short discussion on replacement level.

<cool graph>

Folks will set it at the very worst handful of players in the league, when it should be after the sharp decrease at -1. It doesn't make any logical sense for the talent distribution to crater after -1 in between the 'just in the league' and 'just out of the league' level. And the play of the -1 guys in other leagues certainly doesn't suggest it. Much more likely that there are more -2 players in the world than -1 guys.

*Numbers are specific to this study, though pretty consistent to others I've seen/done. Notes on distribution hold across other numbers.

I don't remember if it was this thread or not, but I think I was involved and I've been thinking about it a lot lately.

I get your thinking that wherever the mode of the league is, that should be the replacement level. I'm not going to say it's wrong, but the higher you set the replacement level, the more guys who are going to rate as below replacement level, and the more awkward things can become.

So in my recent RAPM VORP study, I used a replacement level of -3.35. Here are guys I have playing more than 20K minutes with a negative cume VORP, along with their career earnings.

Jeff Green $102 mill
Ricky Davis $42 mill
Drew Gooden $67 mill
Kevin Martin $83 mill
Tristan Thompson $122 mill

Why include the money? Not to do a bashing of these guys, but just as a reminder:

Actual "replacement players" in the NBA get paid peanuts compared to these numbers. So this is the awkward thing:

How messed up is it that one guy makes $100 mill while apparently being worse at basketball than a horde of other guys - most of whom won't ever get rich playing ball? What's going on there?

I do think there's good conversation to be had about each of these guys, but I suppose my real concern comes when you use a much higher level. If I do the same query with a -1 replacement and I get way more guys, will we really believe it logical that all these guys spent their careers actively damaging their team on average compared to what a guy they could pick up from the G-league at any point?

Alright, here's that list then of guys I haveplaying 20K minutes who'd have a negative VORP with the -1 standard:

Spoiler:

Jeff Green
Ricky Davis
Drew Gooden
Kevin Martin
Tristan Thompson
Jarrett Jack
Juwan Howard
Luke Ridnour
Jason Williams
Samuel Dalembert
John Salmons
Larry Hughes
DJ Augustin
Jordan Clarkson
Antoine Walker
Nick Van Exel
Donyell Marshall
Shareef Abdur-Rahim
Erick Dampier
Jerami Grant
Mo Williams
Zach LaVine
Michael Finley
Al Jefferson
Damon Stoudamire
Jamal Crawford
Andrew Wiggins
Arron Afflalo
Jalen Rose
Clifford Robinson
Jim Jackson
Rajon Rondo
Carlos Boozer
Jonas Valanciunas
Joe Smith
Robin Lopez

I'll leave it without further comment right now.

Last note:

In theory, a team full of replacement level players should yield an SRS 5 times the player APM. So if replacement level is -1, then we should expect a team like that to be able to perform at -5 SRS. Level -2? -10 SRS. Level -3? -15 SRS.

For me, this is something pushing toward more extreme values of replacement level. I'm just really skeptical that you could create a new team out of leftovers, and expect that team to have a chance being better than a roster of some of the guys listed in the spoiler.

I can understand the example where an NBA team is emphasizing youth, coupled with injuries and a front office that would like a high draft pick, and how that could lead to a negative VORP team, so I'm not saying it's literally impossible for a real NBA team to be that bad, but I wouldn't say that's what was going on with the guys listed.

We got something mixed up.

I wouldn’t advocate for the mode, but for the slot below the mode (-2 in that example). Assuming a somewhat similar distribution (numbers may vary, but that shape should remain). I very very strongly suspect the ‘true’ talent distribution of players who make the NBA resembles that curve.

I am advocating for setting the replacement level at ~10% instead of at the ~1% we see on some stats. BPM is an example that is actually too far in the other direction imo.

Jeff Green is the only player I consistently see below career replacement level at those minutes levels. Sorry Jeff Green.

Edit: Team level is what initially led me in the direction of -2 over -3. I believe we've seen teams that are at least very close to averaging replacement player. No team ever has won at a -3 rate. 6 teams under -2.5: '48 Steamrollers, '73 Sixers, '93 Mavs, '98 Nuggets, '12 Bobcats, '16 Sixers.

Generally I was trying to focus where on the curve to set as replacement and not the particular numbers though. Some folks have very different spreads on their curves and that's a bit of a different discussion (see JEs career curve that goes all the way from -9 to +10).

[Doctor MJ]

I think it was here we had a short discussion on replacement level.

<cool graph>

Folks will set it at the very worst handful of players in the league, when it should be after the sharp decrease at -1. It doesn't make any logical sense for the talent distribution to crater after -1 in between the 'just in the league' and 'just out of the league' level. And the play of the -1 guys in other leagues certainly doesn't suggest it. Much more likely that there are more -2 players in the world than -1 guys.

*Numbers are specific to this study, though pretty consistent to others I've seen/done. Notes on distribution hold across other numbers.

I don't remember if it was this thread or not, but I think I was involved and I've been thinking about it a lot lately.

I get your thinking that wherever the mode of the league is, that should be the replacement level. I'm not going to say it's wrong, but the higher you set the replacement level, the more guys who are going to rate as below replacement level, and the more awkward things can become.

So in my recent RAPM VORP study, I used a replacement level of -3.35. Here are guys I have playing more than 20K minutes with a negative cume VORP, along with their career earnings.

Jeff Green $102 mill
Ricky Davis $42 mill
Drew Gooden $67 mill
Kevin Martin $83 mill
Tristan Thompson $122 mill

Why include the money? Not to do a bashing of these guys, but just as a reminder:

Actual "replacement players" in the NBA get paid peanuts compared to these numbers. So this is the awkward thing:

How messed up is it that one guy makes $100 mill while apparently being worse at basketball than a horde of other guys - most of whom won't ever get rich playing ball? What's going on there?

I do think there's good conversation to be had about each of these guys, but I suppose my real concern comes when you use a much higher level. If I do the same query with a -1 replacement and I get way more guys, will we really believe it logical that all these guys spent their careers actively damaging their team on average compared to what a guy they could pick up from the G-league at any point?

Alright, here's that list then of guys I haveplaying 20K minutes who'd have a negative VORP with the -1 standard:

Spoiler:

Jeff Green
Ricky Davis
Drew Gooden
Kevin Martin
Tristan Thompson
Jarrett Jack
Juwan Howard
Luke Ridnour
Jason Williams
Samuel Dalembert
John Salmons
Larry Hughes
DJ Augustin
Jordan Clarkson
Antoine Walker
Nick Van Exel
Donyell Marshall
Shareef Abdur-Rahim
Erick Dampier
Jerami Grant
Mo Williams
Zach LaVine
Michael Finley
Al Jefferson
Damon Stoudamire
Jamal Crawford
Andrew Wiggins
Arron Afflalo
Jalen Rose
Clifford Robinson
Jim Jackson
Rajon Rondo
Carlos Boozer
Jonas Valanciunas
Joe Smith
Robin Lopez

I'll leave it without further comment right now.

Last note:

In theory, a team full of replacement level players should yield an SRS 5 times the player APM. So if replacement level is -1, then we should expect a team like that to be able to perform at -5 SRS. Level -2? -10 SRS. Level -3? -15 SRS.

For me, this is something pushing toward more extreme values of replacement level. I'm just really skeptical that you could create a new team out of leftovers, and expect that team to have a chance being better than a roster of some of the guys listed in the spoiler.

I can understand the example where an NBA team is emphasizing youth, coupled with injuries and a front office that would like a high draft pick, and how that could lead to a negative VORP team, so I'm not saying it's literally impossible for a real NBA team to be that bad, but I wouldn't say that's what was going on with the guys listed.

We got something mixed up.

I wouldn’t advocate for the mode, but for the slot below the mode (-2 in that example). Assuming a somewhat similar distribution (numbers may vary, but that shape should remain). I very very strongly suspect the ‘true’ talent distribution of players who make the NBA resembles that curve.

I am advocating for setting the replacement level at ~10% instead of at the ~1% we see on some stats. BPM is an example that is actually too far in the other direction imo.

Jeff Green is the only player I consistently see below career replacement level at those minutes levels. Sorry Jeff Green.

Edit: Team level is what initially led me in the direction of -2 over -3. I believe we've seen teams that are at least very close to averaging replacement player. No team ever has won at a -3 rate. 6 teams under -2.5: '48 Steamrollers, '73 Sixers, '93 Mavs, '98 Nuggets, '12 Bobcats, '16 Sixers.

Generally I was trying to focus where on the curve to set as replacement and not the particular numbers though. Some folks have very different spreads on their curves and that's a bit of a different discussion (see JEs career curve that goes all the way from -9 to +10).

Oh interesting.

Okay, I do get the categorical appeal of "the slot below the mode", it's just that I'm looking to operate in a real number space and so don't really think in terms of slots so much as I think about methods of getting a number that feels like it's not coming from arbitrary decisions.

I don't know how everyone who has calculated replacement player before has done it, but I'll put forward two ways, the first I know has been done and the second I'd expect has:

1. Make a graph of all the players in a given sample based on RAPM vs MP, generate a line of best fit, and whatever the RAPM is when the line crosses 0 MP is your replacement level.

2. Replace all players below a certain minutes threshold in your sample with one "Replacement Player", and then proceed with the regression as normal. Whatever your Replacement Player's RAPM ends up as, is your Replacement Level.

Definitely curious if others have looked into these methods more than me as it's something I haven't thought about in a good while.

Re: only Jeff Green below replacement. Okay, but was I clear about the fact that even with my extremely negative replacement level, you're getting more than just Green as a long-career veteran below replacement? Like I say, this is the thing that's key for me.

Let me focus in on Tristan Thompson for a second here. This dude wasn't just on a championship team, he was an absolutely critical part of that team. Do I expect that's possible with a typical replacement player? No, I really don't. So this sticks in my craw a bit.

Now, just because you peaked well above replacement level doesn't mean you spent most your career there, and so that's part of why seeing Thompson on that list doesn't lead me to see a red flag, If I go on nbarapm and look at his 3-year RAPM samples with thresholds at -1, -2, and -3.35 here's how his career would split:

-1: At or Above replacement 3 samples (16, 17, 18), below replacement 11 samples.
-2: At or Above replacement 6 samples (13, 15, 16, 17, 18, 25), below replacement 8 samples.
-3.35 At or Above replacement 11 samples, below replacement 3 samples (12, 20, 21).

Okay so first, I gotta say, that last number is a problem for me I think in terms of me mixing and matching Englemann's career RAPM and nbarapm's RAPM studies because it's awfully hard for me to believe that a player with 11 out of his 14 years being above replacement would end up with a career below replacement.

But I think the basketball question here is just: What seems most right based on what we know of this player's career? I'll leave it there as a stem if folks want to examine that one more with me.

Re: believe we've seen teams that would average a replacement player. I think that's a great discussion for us to have. I'll start the conversation for 3 PBP era teams:

'97-98 Denver Nuggets
11-71
SRS -11.74
rORtg -6.0
rDRtg +7.1

Main 5 players by minutes with age:
Dean Garrett (31)
LaPhonzo Ellis (27)
Anthony Goldwire (26)
Johnny Newman (34)
Bobby Jackson (24)

How many of those guys were Main 5 the prior year? 1 (Ellis)
How many of those guys were Main 5 the next year? 0, and none even on the roster.

Worst Season PM with On-Off:

Dean Garrett -756 (-6.2)
LaPhonso Ellis -604 (-0.8)
Anthony Goldwire -522 (+0.9)
Eric Washington -519 (-7.1)
Bobby Jackson -491 (+1.7)

This group actually feels like a strong case for being what a replacement team would expect to look like in the sense that these guys were only in town for a hot minute, and then from there with Dean Garrett you've got perhaps the ultimate not-good +/- player we've ever seen.

The weird, and thus interesting, thing though, is the fact that this wasn't a young team, and nor was this a big tank year the way '96-97 was for Duncan. Worth doing more research into what happened at that time, as while I was around back then, I wasn't so invested to know what all the bad teams were trying to do.

Before I leave them though it's worth noting that the next year would see Denver hire Mike D'Antoni for his first NBA coaching gig, and the offense immediately spike to decency only for the GM (Dan Issel) to fire him, and hire himself. While at the time the subsequent results didn't look terrible, I think it's worth marveling at having the gall of firing the guy whose vision (already a known thing at the time) would remake the sport in the 21st century...because you thought you could do it better. It looks bad ass when you do this sort of thing and are a massive success (like Pop in SA), but man, I'd sure as hell rather have stuck with D'Antoni.

'11-12 Charlotte Bobcats
7-59
SRS -13.96
rORtg -9.4
rDRtg +5.8

Main 5 players by minutes with age:
Gerald Henderson (24)
Kemba Walker (21)
Byron Mullens (22)
Bismack Biyombo (19)
Derrick Brown (24)

How many of those guys were Main 5 the prior year? 0.
How many of those guys were Main 5 the next year? 3 (Walker, Biyombo, Henderson).

Worst Season PM with On-Off:

Kemba Walker -505 (+0.7)
Gerald Henderson -492 (+2.5)
Bismack Biyombo -462 (-2.0)
DJ Augustin -404 (+0.5)
Derrick Brown -379 (+1.7)

So this is a different case I'd say. This is a clear youth movement team who was very much hoping to win the lottery (2012 was a tank year for AD), and ended up with the #2 pick where they selected Michael Kidd-Gilchrist.

Further, when we see Kemba Walker in particular here, we know he would become a great player and that's what the team was trying to get him to become.

This then to say: This roster may or may not be better/worse than a replacement level team, but I don't think it should be a standard candle we look at here the way the '97-98 Nuggets could be.

Let me also note before I leave that in this shortened season, the quality of play might also be lower than typical for the NBA, and this might in particular hurt teams that were looking to build something new but lacked a typical season ramp-up get that going. This is another reason to look at these Bobcats as potentially even worse relative to replacement team level.

'15-16 Philadelphia 76ers
10-72
SRS -9.92
rORtg -7.6
rDRtg +2.8

Main 5 players by minutes with age:
Hollis Thompson (24)
Jerami Grant (21)
Isaiah Canaan (24)
Nerlens Noel (21)
Robert Covington (25)

How many of those guys were Main 5 the prior year? 3 (Noel, Covington, Thompson).
How many of those guys were Main 5 the next year? 1 (Covington).

Worst Season PM with On-Off:

Jahlil Okafor -495 (-9.4)
Jerami Grant -492 (+1.3)
Isaiah Canaan -484 (-4.7)
Nerlens Noel -459 (-1.4)
Nik Stauskas -405 (-2.3)

Okay so we know this is legendarily the most tanky team in the PBP era so I don't think we can use it for a standard candle either. We might say though that it's logically a candidate for a worse than replacement team so it's noteworthy that they didn't even have an SRS below -10.

The elephant in this particular room is the disaster that was Okafor. Okafor's OnCourt was a -16.4, so obviously that's well under not just -10 but -15 while playing with a bunch of other players well above -15. And of course Okafor tops the team for the season despite playing only 53 games - during which he was the team's primary scoring option which led him to make All-Rookie while playing at a level that really seems about as clearly sub-replacement level as you're ever going to see.

It's interesting. This was the season that made the outrage go crazy about Philly's tanking, but while they clearly qualify as a tanking team, I would say that the reason for the outrage was the team getting so, so bad this year, and so much of that was just a natural product of giving their top draft pick the primacy he'd expect to have if he can do his thing in the NBA. Okafor was just one of the most extreme cases of "nope, turns out you can't just do your thing against NBA players" from players drafted as potential stars that we've ever seen.

I'll right. I'll leave it there and just note again how much the '97-98 Nuggets seem like about as close as we're ever going to come to a true replacement team, and their SRS of -11.74 would yield a replacement player level of -2.348. I am still wary of using an actual team as the standard here, but I do think it's worth having a team in mind that feels in spirit as close to a replacement team as possible, and those Nuggets seem like it.