RealGM 2020 Top 100 Project: #26 (John Stockton)

[DQuinn1575]

[quote="Doctor MJ"][quote="trex_8063"]Replying to a couple posters, though mostly to Doctor MJ.

Thoughts on ranking, which maybe should go elsewhere, but seems applicable here:

I guess my problem with the CORP is that it is above replacement, and being above replacement level doesn't help a team win a title. In fact, as we see in the case of Stockton and his Utah days, his championship team hopes are hurt by this. Fundamentally in order to win a championship you have to be somewhere around a 6.5 SRS team, if replacement value is -10 then you need to be 16.5 better than SRS to win. With 19,680 minutes a season, that is about 1.65 points per 2,000 minutes. So if you are 1.0 better than replacement you force the team to be 15.5 points better in 18,000, up to 1.75 points in the remaining minutes. So having a player being slightly above replacement forces the team to get better players, and in the long run hurts the teams chances since you are filling limited minutes with players that have less value then the rest of the league (average).

If you look at VORP per B-Ref for Utah's best team, 1997, and compare it to the Bulls, the top 4 players by VORP total 20.1 for the Bulls and 20.4 for the Jazz. If you like win shares instead, it is 47.9 for the Jazz, 45.8 for the Bulls. Utah's problem, as documented in earlier posts, is that other than Hornacek (and here RUssell), the Jazz get no support after that.
Ostertag is 0.8 VORP playing 1,818 minutes, 4th most on the team. No one else on the team with less minutes was better, so it's the best they had. The Bulls meanwhile had Harper, Kukoc, Rodman, Kerr playing around that level of minutes but all contributing way more.

There are 3 different levels we are talking about - level to make the league, which is what baseball uses and the origin of replacement level theory, a game level, because in basketball if Karl Malone went down his 38 minutes arent replaced by one guy, but instead the minutes are spread, so the 3rd forward plays more, the 4th forward plays more, etc. But long term it really is average player, because long term the team will "on average" have an average player. So for the following year, if the team is trying to win a championship, they are going to try and upgrade from Ostertag. I luckily picked an example that shows that, his minutes declined the next year.

So to summarize, CORP seems to be a flawed concept, as you don't win championships by exceeding replacement level. Or you don't get to values to win a championship if your hurdle is replacement; to get a championship you have to zet values greater than the rest of the league - not just above replacement.

[Doctor MJ]

Replying to a couple posters, though mostly to Doctor MJ.

Thoughts on ranking, which maybe should go elsewhere, but seems applicable here:

I guess my problem with the CORP is that it is above replacement, and being above replacement level doesn't help a team win a title. In fact, as we see in the case of Stockton and his Utah days, his championship team hopes are hurt by this. Fundamentally in order to win a championship you have to be somewhere around a 6.5 SRS team, if replacement value is -10 then you need to be 16.5 better than SRS to win. With 19,680 minutes a season, that is about 1.65 points per 2,000 minutes. So if you are 1.0 better than replacement you force the team to be 15.5 points better in 18,000, up to 1.75 points in the remaining minutes. So having a player being slightly above replacement forces the team to get better players, and in the long run hurts the teams chances since you are filling limited minutes with players that have less value then the rest of the league (average).

If you look at VORP per B-Ref for Utah's best team, 1997, and compare it to the Bulls, the top 4 players by VORP total 20.1 for the Bulls and 20.4 for the Jazz. If you like win shares instead, it is 47.9 for the Jazz, 45.8 for the Bulls. Utah's problem, as documented in earlier posts, is that other than Hornacek (and here RUssell), the Jazz get no support after that.
Ostertag is 0.8 VORP playing 1,818 minutes, 4th most on the team. No one else on the team with less minutes was better, so it's the best they had. The Bulls meanwhile had Harper, Kukoc, Rodman, Kerr playing around that level of minutes but all contributing way more.

There are 3 different levels we are talking about - level to make the league, which is what baseball uses and the origin of replacement level theory, a game level, because in basketball if Karl Malone went down his 38 minutes arent replaced by one guy, but instead the minutes are spread, so the 3rd forward plays more, the 4th forward plays more, etc. But long term it really is average player, because long term the team will "on average" have an average player. So for the following year, if the team is trying to win a championship, they are going to try and upgrade from Ostertag. I luckily picked an example that shows that, his minutes declined the next year.

So to summarize, CORP seems to be a flawed concept, as you don't win championships by exceeding replacement level. Or you don't get to values to win a championship if your hurdle is replacement; to get a championship you have to zet values greater than the rest of the league - not just above replacement.

So one thing to be clear:

ElGee's CORP is not a linear WAR metric:



That superlinearity he has in there is specifically designed to map on to championship odds. Make sense? If not, feel free to ask questions.

But to be clear, any mathematical here is quite coarse. I would not suggest that any use his superlinear function as a holy grail, and I'm sure neither would ElGee. Rather, he's just trying to address concerns like this with as vanilla of a mathematical approach as possible so that we can easily understand how his metric came to say what it did, and we have a good sense of how our own methods deviate from his so that we can calibrate from his algorithm in first pass analysis. If I know higher on Attribute X than he is, then I can expect that all other things being equal, more often than not my in-depth assessment will deviate from his on this front with a predictable vector.

Incidentally, this is how I use stats like PER, WS, BPM, etc, and why I often take my time before I incorporate a new stat into my analytical toolset. It's also why I've been so frustrated with Jeremias Englemann on his RAPM/XRAPM/RPM journey. Once you put a stat out, I don't want to see a different stat given the same name from the same source because it makes everything confusing. It's part of why I always said that statisticians like him are confused about their purpose, which is to help analysts come to conclusions about all facets of a player not to make the most predictive all-in-one metric.

[penbeast0]

Go Bird Go!

[Cavsfansince84]

1. John Stockton
-very good prime where he is consistently running offenses that rate very highly in ORtg(top 11 every year from 1990-2003 and top 7 every year from 92-01)
-top 6 in vorp for 10 straight years
-obviously extremely durable and still a very good player until he retired at the age of 40
-career ts add of 2466 and consistently in the 140-200 range for 15 straight years
-led league in assists 9 straight years
-made 5 all defensive 2nd teams
-top 15 in mvp voting 13 times despite being sort of overshadowed by the Mailman
-I think somewhat underrated in terms of how dominant the Jazz were in those playoff runs in the 97/98 seasons until the finals. The Jazz were 22-6 vs the west in those years while going against 4 teams that won 56+ games including a sweep of a 61 win Laker team.

2. Steve Nash
-12 year prime where his teams were top 6 in ORtg 10x and top 4 8x
-led league in apg 5x and ast% 5x
-top 10 in mvp voting 6x, 2x winner
-very durable on top of having a long prime, averaging 76g per year during 12 year prime
-fair amount of playoff success(4x wcf) with Nash playing very well in those runs. Efficiency tended to hold up well.
-very efficient scorer with career ts add of 2127 and 3 straight years over 200 during peak years

3. John Havlicek
-11x all nba 1st or 2nd team(4 1st)
-7x all nba defense(5 1st) and played 6 years before those were created
-7x top 10 in mvp voting
-8x champ and fmvp
-3x led title teams in playoff scoring, assists and win shares(68, 69 and 74)
-great all around skills(led Celtics in scoring 7x, apg 7x, win shares 6x)
-great chemistry/leadership guy

[LA Bird]

Vote: John Stockton

Essentially I think this comes down to the philosophical question of how one much one values peak vs longevity. Wade and Nash had clearly better peaks but Stockton had clearly better longevity. Some people only use longevity as a form of tiebreaker between players on the same "tier" but the definition of prime duration to determine such tiers always seemed arbitrary to me. Stockton won't be able to lead a championship level team as the best player like Wade and Nash but he can still be a championship level #2. If we replace 97/98 Malone with 12/13 LeBron, that's likely two rings for him and Stockton was playing around or above that level for 15 years. He wasn't a top 5 player but it's not like Stockton's incredible longevity comprised of meaningless post-prime seasons like the ones from Udonis Haslem the last few years.

As a player, I think Stockton has a similar issue as Wilt when it comes to stats. Just because they can put up all those numbers doesn't mean they should in an ideal scenario. But at the same time, people often go too far the other way when they try to tear down the gaudy box scores. The sheer volume of Stockton's assists vastly overrated his value as an on-ball playmaker in the half court and pigeonholed him in a role that magnified his offensive limitations while overlooking his off-ball value. I think Stockton's shooting, off ball movement, and willingness to set screens can be quite valuable in addition to his passing next to a better #1 option than Malone on a team with more secondary playmakers. There are some point guards who peaked higher than Stockton on either ends of the floor during his career (eg. KJ on offense, Blaylock on defense) but none were able to have the overall two-way impact and play as long as Stockton did. I think his overwhelming longevity advantage is enough to give him the edge over Nash and Wade even if he doesn't have a peak close to either of them.

[trex_8063]

OK, to recap the original vote count was as follows:

John Stockton - 5 (Baski, Clyde Frazier, Joao Saraiva, penbeast0, trex_8063)
Dwyane Wade - 4 (Dr Positivity, Dutchball97, iggymcfrack, Odinn21)
Steve Nash - 4 (Doctor MJ, eminence, Jordan Syndrome, Whopper_Sr)
James Harden - 2 (DQuinn1575, Magic Is Magic)
Elgin Baylor - 1 (Hal14)


Doing our usual method of eliminating the low vote-getters and transferring those votes to alternate selections left us with a 3-way tie between Stockton, Nash, and Wade (5 votes each), with one ghosted vote.

From the people I have heard back from, if I did the Condorcet method [which we have precedent of using this, if necessary, before], the results would be as follows:
Stockton beats Nash 8-7
Stockton ties Wade 7-7
Wade beats Nash 8-7

So Nash would be the loser in this method, but still have no clear victor between Stockton and Nash. At this point, we're a full 24 hours past the original deadline, and I have not heard back from two of the original 16 voters to [potentially, anyway] settle this deadlock.

fwiw, if we were doing a ballot system, Wade would win by the narrowest of margins....
A 5/3/1-point ballot system, the scores would be:
Wade - 40
Stockton - 38
Nash - 24

If it were a 3/2/1-point ballot system, the scores would be:
Wade - 26
Stockton - 24
Nash - 16


However, since I asked the question we've had TWO additional voters [both whom have been active in this project AND who are in good standing on the forum] cast their votes. Additionally, there have been no fewer than FOUR other consistent participants [by way of direct statement or "And1"'s] saying they're in favour of just counting the votes of those two [even though they're late], rather than compromise the selected voting system we've been working with.

I must admit, I'd rather "stay the course" on our voting system as much as is reasonably possible, too.
As they both voted in favour of John Stockton, I'm going to call this one with Stockton as the winner.
We'll say this is setting another precedent in this project [which, fwiw, is consistent with prior top 100 projects] wherein if using our ranked vote system AND Condorcet method as a back-up, we still find ourselves in a deadlock: we'll go into some manner of runoff allowing new/late votes to decide it.

HOWEVER, please try to get your votes in on time in the future.
Putting me in a position where I have to make some manner of discretionary call on how to proceed and decide a winner is precisely the circumstance that frequently ends with me enduring criticism and accusations of bias/agenda/manipulation (which ain't fun).

Spoiler:

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

[quote=”sansterre”].[/quote]

.

.

.

.

.

.

.

.

.

.

.

.

[DQuinn1575]

So one thing to be clear:

ElGee's CORP is not a linear WAR metric:



That superlinearity he has in there is specifically designed to map on to championship odds. Make sense? If not, feel free to ask questions.

But to be clear, any mathematical here is quite coarse. I would not suggest that any use his superlinear function as a holy grail, and I'm sure neither would ElGee. Rather, he's just trying to address concerns like this with as vanilla of a mathematical approach as possible so that we can easily understand how his metric came to say what it did, and we have a good sense of how our own methods deviate from his so that we can calibrate from his algorithm in first pass analysis. If I know higher on Attribute X than he is, then I can expect that all other things being equal, more often than not my in-depth assessment will deviate from his on this front with a predictable vector.

Incidentally, this is how I use stats like PER, WS, BPM, etc, and why I often take my time before I incorporate a new stat into my analytical toolset. It's also why I've been so frustrated with Jeremias Englemann on his RAPM/XRAPM/RPM journey. Once you put a stat out, I don't want to see a different stat given the same name from the same source because it makes everything confusing. It's part of why I always said that statisticians like him are confused about their purpose, which is to help analysts come to conclusions about all facets of a player not to make the most predictive all-in-one metric.

Thanks, the graph is helpful, I'm good with the odds, and realize that a +6 year is more than twice as good as a +3 year. I guess my thought is that the point where I think we go positive is the average starter line, and any season below that doesnt help you win the championship. Fundamentally x years from now your competitors are going to have on average that level of starter, so getting someone that good, although required, doesnt really advance you to a championship as on average everyone else is going to have that. Salary cap limits, talent, draft, etc. disperses the talent in the long term, so that the Top 150 players isn't contributing hardly any if any value to a championship. On average every team has 5 - once you get better than that you are (1) adding value needed to get to a 6.5 win team, and (2) at the point where demand/supply gets the talent to be a scarce quantity in the long run. I havent done the math, but the top 150 is more like 0%

If I do the math off the graph, I get

Approx
Rank Number Odds Total

0.875 0.88 21.0% 0.18
5 4.12 12.5% 0.52
10 5 9.0% 0.45
25 15 6.0% 0.90
40 15 5.0% 0.75
150 110 3.0% 3.30 Eyeball between 40 and 150
300 150 1.0% 1.50

7.42 Use 7.5 for ease/rounding

Top 40 2.80
37.7%

So about 7.5 championships added, not sure what the number should be - maybe 5 for 5 players on the court. But I get about 37.7% of the value for championships out of the Top 40 players in the league, with 62% from players outside that. So I think the floor is too high, and if you are evaluating championships it all gets lowered 2.5% or so, which would move it to 5.00 total, and also give little but some value to a top 150 season. It also only gives the Top 5 players a 71.0% chance of winning the title (assuming no cases of top 5 players on same team), and say the Top 3 around .43 (.18 + .125 + .125) when 538 did this they got around 70% of Top 3
https://fivethirtyeight.com/features/do-the-celtics-have-enough-star-power-to-win-a-title-not-yet/

So my point really is that I dont think when you are the 200th best player you are contributing to championships, and even when you are 100 you are contibuting so little it's almost meaningless when evaluating Top 25 ever.

[DQuinn1575]

OK, to recap the original vote count was as follows:

John Stockton - 5 (Baski, Clyde Frazier, Joao Saraiva, penbeast0, trex_8063)
Dwyane Wade - 4 (Dr Positivity, Dutchball97, iggymcfrack, Odinn21)
Steve Nash - 4 (Doctor MJ, eminence, Jordan Syndrome, Whopper_Sr)
James Harden - 2 (DQuinn1575, Magic Is Magic)
Elgin Baylor - 1 (Hal14)


Doing our usual method of eliminating the low vote-getters and transferring those votes to alternate selections left us with a 3-way tie between Stockton, Nash, and Wade (5 votes each), with one ghosted vote.

From the people I have heard back from, if I did the Condorcet method [which we have precedent of using this, if necessary, before], the results would be as follows:
Stockton beats Nash 8-7
Stockton ties Wade 7-7
Wade beats Nash 8-7

So Nash would be the loser in this method, but still have no clear victor between Stockton and Nash. At this point, we're a full 24 hours past the original deadline, and I have not heard back from two of the original 16 voters to [potentially, anyway] settle this deadlock.

fwiw, if we were doing a ballot system, Wade would win by the narrowest of margins....
A 5/3/1-point ballot system, the scores would be:
Wade - 40
Stockton - 38
Nash - 24

If it were a 3/2/1-point ballot system, the scores would be:
Wade - 26
Stockton - 24
Nash - 16


However, since I asked the question we've had TWO additional voters [both whom have been active in this project AND who are in good standing on the forum] cast their votes. Additionally, there have been no fewer than FOUR other consistent participants [by way of direct statement or "And1"'s] saying they're in favour of just counting the votes of those two [even though they're late], rather than compromise the selecting voting system we've been working with.

I must admit, I'd rather "stay the course" on our voting system as much as is reasonably possible, too.
As they both voted in favour of John Stockton, I'm going to call this one with Stockton as the winner. We'll say this is again setting a precedent [which, fwiw, is consistent with prior top 100 projects] wherein even in using our ranked vote system AND Condorcet method as a back-up, if we still find ourselves in a deadlock: we'll go into some manner of runoff, allowing new/late votes to decide it.

HOWEVER, please try to get your votes in on time in the future.
Putting me in a position where I have to make some manner of discretionary call on how to proceed and decide a winner is precisely the circumstance that frequently ends with me enduring criticism and accusations of bias/agenda/manipulation (which ain't fun).

Spoiler:

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

[quote=”sansterre”].

.

.

.

.

.

.

.

.

.

.

.

.

[Odinn21]

OK, to recap the original vote count was as follows:

John Stockton - 5 (Baski, Clyde Frazier, Joao Saraiva, penbeast0, trex_8063)
Dwyane Wade - 4 (Dr Positivity, Dutchball97, iggymcfrack, Odinn21)
Steve Nash - 4 (Doctor MJ, eminence, Jordan Syndrome, Whopper_Sr)
James Harden - 2 (DQuinn1575, Magic Is Magic)
Elgin Baylor - 1 (Hal14)

TBH, I'm not comfortable with this result because in one of the close results before, when we couldn't achieve a majority, you applied points system and declared a winner.

LA Bird's vote doesn't even follow our procedure. He basically voted for Stockton vs. Nash, all other voted for #26. LA Bird could still put Stockton as his vote. He could vote Stockton/Baylor/Zeke or, Harden/Stockton/Nash. His effective vote still could be Stockton, but it didn't happen that way.

I'd like to see a run-off or usage of the point system as you did before. This is the 1st time we had a different winner when we stopped the clock.

Edit; I'm particularly uncomfortable on this occasion before even though there were times I couldn't vote, whenever I vote, I'm usually one of the first ones to submit selections and with a result like this, I feel like it was down to tactical voting* of who weren't on time. It doesn't feel nice.
* The situation was tactical by nature with a result so close. I'm not putting it on LA Bird or Cavsfansince84.

We had similar situations in #16 and #19.
In #16, even though we didn't get a majority, Malone was the winner by points and Condorcet method, and you declared him winner.
In #19, Mikan was leading in votes and points, you asked 3 ghost votes as their preferences between Mikan and Moses. 2 of them replied Mikan and you declared Mikan as the winner in 40 minutes.

This round was a 3-way tie in our usual process. It was a tie between Wade and Stockton in Condorcet method. Wade had the most points.
And with votes which weren't on time, Stockton got this place. Not cause of ghost votes. And was it a full day?..

[trex_8063]

HOWEVER, please try to get your votes in on time in the future.
Putting me in a position where I have to make some manner of discretionary call on how to proceed and decide a winner is precisely the circumstance that frequently ends with me enduring criticism and accusations of bias/agenda/manipulation (which ain't fun).

Spoiler:

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

[quote=”sansterre”].

.

.

.

.

.

.

.

.

.

.

.

.

[Odinn21]

First, I think I should make it clear that I’m not criticising you on any personal level and I’m not suggesting that you called this for Stockton when you saw it’s Stockton.

Second, I tried to keep it short about the previous result processes.

Third, what makes me uncomfortable that my on time vote was an actual factor since it was such a close result and it became a non-factor after an overtime we never had before with late votes. I think it’s natural that I’m upset with it. I know that those two votes came from very respectable posters on here. It doesn’t mean that I should overlook the fact that results were not dictated by ghost / on time votes and were dictated by (very) late votes.

Edit: This page has massive visual issues right now.

[LA Bird]

So we have the 3-way tie as indicated at the tally yesterday; I've been trying to reach involved parties both via PM and by quoting them here, but have not heard back. In about 3 hours will be a full [24-hour] day past the deadline. I can fall back on some alternate methods (ballot scoring) to decide a winner at that point, which I am seriously considering doing if I haven't heard from these posters in the next 3 hours......UNLESS ANYONE OBJECTS.

The way I interpreted this was that ballot scoring would be used to break the tie at the 3 hour deadline if no one objected. Because our votes were counted in this interim period and the tie was broken before the deadline, Stockton was declared the winner by ranked choice vote.

Both Cavsfansince84 and I would have voted Stockton #1 based on our previous voting history anyway so the nature of the 3 way tie didn't change anything.

Cavsfansince84 vote in #25
1. Pettit
2. Stockton
3. Nash

My most recent vote in #23
1. Paul
2. Stockton
3. Nash

Edit: Yeah, the nested spoiler in post #108 crashed the page. Best if we just move on

[Odinn21]

So, it's OK to be this late in a project / thread that's been going on for days and I didn't see "if any objections for 3 hours". I'm still not those votes level late.
I object.

[Magic Is Magic]

I am a little embarrassed to see that John Stockton was ranked 26th to be honest.

0 MVPs
0 Rings
0 All Defensive 1st team
2 All NBA 1st team
0 RS WS/48 > .250
0 PO WS/48 > .250
2 Finals

He had amazing longevity, no doubt about it, but that's about it. He does not stack up to everyone else on the top 30 list and he has no reason to be there. But anyway, I digress...

Voting for the #27 SPOT is up next.

[colts18]

I am a little embarrassed to see that John Stockton was ranked 26th to be honest.

0 MVPs
0 Rings
0 All Defensive 1st team
2 All NBA 1st team
0 RS WS/48 > .250
0 PO WS/48 > .250
2 Finals

He had amazing longevity, no doubt about it, but that's about it. He does not stack up to everyone else on the top 30 list and he has no reason to be there. But anyway, I digress...

Voting for the #27 SPOT is up next.

Stockton is 6th all-time in Win Shares. 17th all-time in WS/48. Last I checked #6 and #17 is lower than 26th all-time.

Stockton is tied for 12th all-time in All-NBA team selections with 11. The players he is tied to include Michael Jordan and Bill Russell.

How about 538's new Raptor stat. Stockton ranks #3 all-time in per 100 possessions. His career value is 1st all-time. His 7 year peak is ranked #3 all-time.

Every single stat has Stockton ranked higher than #26 all-time. Most of them have his career in the top 10.

[Baski]

I am a little embarrassed to see that John Stockton was ranked 26th to be honest.

0 MVPs
0 Rings
0 All Defensive 1st team
2 All NBA 1st team
0 RS WS/48 > .250
0 PO WS/48 > .250
2 Finals

He had amazing longevity, no doubt about it, but that's about it. He does not stack up to everyone else on the top 30 list and he has no reason to be there. But anyway, I digress...

Voting for the #27 SPOT is up next.

Right off the bat your selected criteria and cutoff points seem tailored to make him look as unimpressive as possible.