RealGM 2023 Top 100 Project - General Thread

[Rishkar]

how u compare kg with mikan. i thought we didnt have anything on him outside of what ppl said n his chips

It's honestly kinda just a gut feeling and I don't have a strong opinion on it. We don't have good numbers on Mikan, but I place him in my bigs tier of 7-10 with Hakeem, Shaq, KG, and Mikan.

well he prob was far and away the best player n he won 7 chips

I evaluate him as being the best player in the world for 7 years, before having somewhat of a decline in his final seasons. I should probably have him higher, but only playing 9 seasons is just really tough to swallow.

[ShaqAttac]

It's honestly kinda just a gut feeling and I don't have a strong opinion on it. We don't have good numbers on Mikan, but I place him in my bigs tier of 7-10 with Hakeem, Shaq, KG, and Mikan.

well he prob was far and away the best player n he won 7 chips

I evaluate him as being the best player in the world for 7 years, before having somewhat of a decline in his final seasons. I should probably have him higher, but only playing 9 seasons is just really tough to swallow.

did ya see docs post on him

[Rishkar]

No, could you link it (or just tell me which thread it was in)?

[ShaqAttac]

No, could you link it (or just tell me which thread it was in)?

i put in in my post in thebeggining of the top 100 #1 thread

it was from a big post doc made ranking everyone in the 40s and 50s but idr the threaf

[sp6r=underrated]

BTW, thank you to all the active participants in these threads. As always the ranking matters way less than the conversation.

[Statlanta]

The Draymond vs. Dwight conversation is how I feel with Giannis vs. Anthony Davis via the talk of accolades underrating/overrating a player.

[DSMok1]

In one of the individual Top 100 project voting threads I mentioned this data; I have a subset of the overall data that is reasonable to share here:

I have created a visualization with selected players active within 1997-2020 range. The visualization shows the players' BPM-informed 4-year RAPM over six separate 4-year stints. I usually use this sort of data as part of my BPM development but it is relevant to the evaluation of modern-era players for this Top 100 project. Note the past 3 seasons are not included in this dataset.

This does not speak to a player's minutes or availability; this just describes their performance per 100 possessions within those 4 seasons. This could be a couple hundred minutes or many thousands of minutes.

This is created via the following routine: create a base prior from team rating and player playing time. Update that prior with information from BPM (if few minutes, this will not influence the prior much). Then run RAPM based on that updated prior.

Here is the Viz: Tableau Visualization of BPM-Informed RAPM

[eminence]

Quick question to the above, does it start with the '96-'97 season and end with the '19-'20 season?

Think so, but not 100% sure because there's been 3 seasons since then, but you said 2 at one point.

And very nice visualization, thanks for providing!

[DSMok1]

Quick question to the above, does it start with the '96-'97 season and end with the '19-'20 season?

Think so, but not 100% sure because there's been 3 seasons since then, but you said 2 at one point.

And very nice visualization, thanks for providing!

You are correct. Time is getting away from me... Edited

[eminence]

Quick question to the above, does it start with the '96-'97 season and end with the '19-'20 season?

Think so, but not 100% sure because there's been 3 seasons since then, but you said 2 at one point.

And very nice visualization, thanks for providing!

You are correct. Time is getting away from me... Edited

No worries at all, and thank you much

[eminence]

.

Got a chance to browse through, very nice stuff.

A couple guys I looked for but didn't see and was curious if you'd mind sharing their results - Deron Williams (homer) and Klay Thompson, Tony Parker, Pau Gasol, Chris Bosh, Scottie Pippen, Dennis Rodman, Kyrie Irving, Chauncey Billups (was going to look at various big 3 type teams).

Oh, and Allen Iverson/Derrick Rose for MVP completeness.

[ChartFiction]

Games played and minutes per game should be more consciously factored into a players rating.

If player A is always a +1 in every game, but plays 48 minutes per game and 82 games, he is a 82-0 player
If player B is always a +5 in every game, but plays 32 minutes a game and plays 70 games, who knows what his record looks like

Player A is the peak NBA player despite a modest +/- (ignoring other factors)

That's not to say, for example, "Jordan played x more minutes than y player. His peak season is thus that % better", especially since sports culture changes between eras will lead modern players to always have less games and less minutes. But Jordans Bulls have some of the best records in NBA history, and a prime reason is that he was almost always on the court.

[OhayoKD]

Games played and minutes per game should be more consciously factored in more into a players rating.

If player A is always a +1 in every game, but plays 48 minutes per game and 82 games, he is a 82-0 player
If player B is always a +5 in every game, but plays 32 minutes a game and plays 70 games, who knows what his record looks like

Player A is the peak NBA player despite a modest +/- (ignoring other factors)

That's not to say, for example, "Jordan played x more minutes than y player. His peak season is thus that % better", especially since sports culture changes between eras will lead modern players to always have less games and less minutes. But Jordans Bulls have some of the best records in NBA history, and a prime reason is that he was almost always on the court.

They should be, but I think it's rather important that one factors them in considering
A. the general picture as opposed to simply looking at an isolated stretch...

(Working off your chosen example):

Spoiler:

1985-98 Jordan: 35887 minutes in 13 seasons - 2760 minutes per season

2004-18 James: 44298 minutes in 15 seasons - 2953 minutes per season
...

viewtopic.php?p=108112722#p108112722
Lebron was averaging 39+ minutes in the nba over 79 games as a 19 year old.

Before he entered the nba at 21, Jordan did not hit 32 minutes in college once. Then, once he entered NBA(excluding the 2 years Jordan took off)

...

This gap expands if we consider the postseason where Lebron played at least 500 total minutes every year excepting 2010(460) and 2021(224). Jordan played less than 500 minutes 7 times. He played less than 200 each of his first 3 years in the league.

B. How much players are playing, relative to the field...

(working off the same example):

Spoiler:

Jordan’s minutes per game ranks by season: 4th, injured, 3rd, 1st, 1st, 4th, 26th (the season everyone picks as his peak…), 6th, 6th, retired, partially retired, 19th, 28th, 17th, retired again.

Lebron, on the other hand? In minutes per game, he finished: 10th, 1st, 2nd, 3rd, 3rd, 13th, 5th, 6th, 6th, 8th, 6th, 5th, 12th, 1st, 1st, injured (was roughly 7th across 55 games), 17th, injured (but would not have made top 30), 20th.

He still has three years leading the league in minutes played (same as Jordan), another second place finish (same as Jordan), six more top ten finishes (to Jordan’s three), and five more top twenty finishes (to Jordan’s four).

C. Minute distributions relative to top-teammates...

(Duncan played way more than Drob or Manu, Draymond/Pippen/Wade played a similar amount as Curry/Jordan/Lebron respectively

[Owly]

Games played and minutes per game should be more consciously factored into a players rating.

If player A is always a +1 in every game, but plays 48 minutes per game and 82 games, he is a 82-0 player
If player B is always a +5 in every game, but plays 32 minutes a game and plays 70 games, who knows what his record looks like

Player A is the peak NBA player despite a modest +/- (ignoring other factors)

That's not to say, for example, "Jordan played x more minutes than y player. His peak season is thus that % better", especially since sports culture changes between eras will lead modern players to always have less games and less minutes. But Jordans Bulls have some of the best records in NBA history, and a prime reason is that he was almost always on the court.

I assume the "+" is their "on"/net.

I think (insofar as I do/can imagine it, and overlook the ... I'll get to it) I'd rather have B. "A" is super lucky in that particular regular season. Points margin is the better long term predictor. +1 standard versus even moderate opponents means I feel like besides homecourt I'm at the whim of near 50:50 in every round.

Now if you're talking about extending it across seasons including playoffs maybe A is better. But now you're not talking about the virtue of +1 versus every opponent in a smaller sample so much as the magical win fairy who defies probability (even at the 82 game level it's absurdly unrealistic).

The value of minutes is real.
Variance even, can be a discussion, though probably with people smarter than me.
This example, with the (super healthy) win guaranteer, seems to me to fall apart on contact with reality. As such it fails to lend any support in the way an example should.

I think a discussion of always winning by one is probably a discussion about points shaving and match fixing.

[HeartBreakKid]

I think we should consider changing voting system again.

Now that we are getting to the half way mark, votes are starting to split in many different directions for nominations and there is now a huge amount of disfranchisement in the most recent threads, and it's likely going to be like this for 50 more threads.

[Doctor MJ]

I think we should consider changing voting system again.

Now that we are getting to the half way mark, votes are starting to split in many different directions for nominations and there is now a huge amount of disfranchisement in the most recent threads, and it's likely going to be like this for 50 more threads.

I'll consider ideas.

What would you propose? and How do you think it would help?

[HeartBreakKid]

I think we should consider changing voting system again.

Now that we are getting to the half way mark, votes are starting to split in many different directions for nominations and there is now a huge amount of disfranchisement in the most recent threads, and it's likely going to be like this for 50 more threads.

I'll consider ideas.

What would you propose? and How do you think it would help?

I'm here to complain not help

I actually haven't thought about it beyond the systems we already used, but they might be too much labor. I didn't really think it was much of an issue earlier in the project, but might become a trend where guys are getting nominated with only a couple votes.

Then again - it's just the nomination process that is going to get very wide. So maybe not that big of a deal.

[AEnigma]

The only real change I might suggest is to look at the top three nominees to start rather than the top two. To some extent people should be paying attention to whether their nomination ordering makes sense (e.g. if you are the only person voting for or nominating a player, you should be prepared to make them your alternate if listing them as your primary is sabotaging your intended alternates).

Another idea might be to count the top two based on primaries… and then also add in whomever had the most alternate votes as a third. Say we have a hypothetical initial nomination distribution of 1-1-3-4-2, but the voters for Players B, C, and E all have Player A as their alternate (Player D voters declined to list any). So then you go to the second round, and the count becomes 4-0-3-4-0, and then third round the count becomes 7-0-0-4-0. That is an extreme hypothetical which to some extent I would expect to solve itself (if two people switch their order, that would suffice), but the potential marginalisation of the most popular secondary votes is the only major “flaw” I see in the present system.

[homecourtloss]

In one of the individual Top 100 project voting threads I mentioned this data; I have a subset of the overall data that is reasonable to share here:

I have created a visualization with selected players active within 1997-2020 range. The visualization shows the players' BPM-informed 4-year RAPM over six separate 4-year stints. I usually use this sort of data as part of my BPM development but it is relevant to the evaluation of modern-era players for this Top 100 project. Note the past 3 seasons are not included in this dataset.

This does not speak to a player's minutes or availability; this just describes their performance per 100 possessions within those 4 seasons. This could be a couple hundred minutes or many thousands of minutes.

This is created via the following routine: create a base prior from team rating and player playing time. Update that prior with information from BPM (if few minutes, this will not influence the prior much). Then run RAPM based on that updated prior.

Here is the Viz: Tableau Visualization of BPM-Informed RAPM

This deserves its own thread.

[HeartBreakKid]

How's about for tie-breakers we just do a sudden-death type of thing? Next vote wins (like Pen voted a bit after the tie breaker for Pierce).