Calculating Championship Odds Based on SRS

DQuinn1575 · Post #21 » by **DQuinn1575** » Sat Oct 11, 2014 10:04 pm

ElGee wrote:

The actual use-cases for this kind of adjustment are as follows: Imagine a player who thinks he THE MAN and also clearly commands a max salary on the open market. The odds of this player landing on an elite team in free agency are very small, and in fact, only Wilt (69), Moses (83) and LeBron (11) exist as counterexamples. Players can be traded, but rarely are trades so lopsided (and involving a max-contract player!) that the team landing the superstar would be a title contender without him. Finally, teams can land players through the draft, and while occasionally a super-elite young player might develop on a great team (e.g. 98 Duncan), there are still forces that prevent this happening (namely the lottery)
)

Actually wilt was traded; there was no free agency. Moses was a free agent, but in those days it was like a trade and sign with the sixers giving up Caldwell jones and a number 1 pick,

You left out shaq on the Lakers.

You also left out any examples where a top player went to a weak team ( other than staying with his team) thru free agency.

You have lebron shaq and moses going to a better team. If your theory is true you should be able to come up with 5-6 examples of very good players taking the money and going to a weak team. I really can't find those.

The restrictions are the lottery and the minimum salary cap for a team.

Sent from my iPhone using RealGM Forums

ElGee · Post #22 » by **ElGee** » Sat Oct 18, 2014 5:53 pm

DQuinn1575 wrote:
ElGee wrote:The actual use-cases for this kind of adjustment are as follows: Imagine a player who thinks he THE MAN and also clearly commands a max salary on the open market. The odds of this player landing on an elite team in free agency are very small, and in fact, only Wilt (69), Moses (83) and LeBron (11) exist as counterexamples. Players can be traded, but rarely are trades so lopsided (and involving a max-contract player!) that the team landing the superstar would be a title contender without him. Finally, teams can land players through the draft, and while occasionally a super-elite young player might develop on a great team (e.g. 98 Duncan), there are still forces that prevent this happening (namely the lottery)
)

Actually wilt was traded; there was no free agency. Moses was a free agent, but in those days it was like a trade and sign with the sixers giving up Caldwell jones and a number 1 pick,

You left out shaq on the Lakers.

You also left out any examples where a top player went to a weak team ( other than staying with his team) thru free agency.

You have lebron shaq and moses going to a better team. If your theory is true you should be able to come up with 5-6 examples of very good players taking the money and going to a weak team. I really can't find those.

The restrictions are the lottery and the minimum salary cap for a team.

Sent from my iPhone using RealGM Forums

Not really sure what to say. You either missed the entire point, or you simply love being a contrarian so much that it prevents you from productive discourse.

Technically LeBron was traded too. Furthermore, the Heat weren't necessarily an elite team without him. The entire point of trying to cite counterexamples was to show how rare it was for ELITE ("the Guy") players to move to ELITE (already title contending) teams in any transactional form. Can't really happen via draft. Injuries, FA and trades are really the only ways.

Most don't think Archie Clark makes the 68 Lakers elite,Same with Caldwell and the 82 Sixers elite. I didn't include Shaq because LA traded Divac and Magic retired off a 4 SRS/53-win season before Shaq signed...it would be bizarre in the context of this theory to include that as a player joining an elite team.

I can look at all the players I have ranked over +4 at their peaks...this misses a few players who put up huge stats and would have commanded a max contract (to say nothing of ego) but if we just look at these 50 guys, most stayed with their teams during their prime. We would very rarely expect to see examples of players signing on terrible teams because...why would they sign on terrible teams unless there were special circumstances?? (Like LeBron going back to Cleveland, a 33-win team, with young talent.)

Of the elite players who were traded or moved, they joined a team of:

Wilt 68 (traded) ~55W

Oscar (traded) ~55W (including expansion)

Moses (traded) ~55W

As I've said many times, it could be done properly by looking at salaries. You'll welcome to do this analysis.

Finally, the flip side would really be to look at the top teams by SRS and simply see how many elite players have ever been added to these teams in the following seasons (assuming lineup continuity) and then compare that to the number of all-star level or mid-level players added. There have been 185 55-win teams in NBA history. Again, you're welcome to go through and check...the 70 Bucks and 82 76ers look like the only ones who did it. But these teams add lesser players constantly, so I think the premise is correct (harder for better players to be on good teams), the question is simply to what degree...

colts18 · Post #23 » by **colts18** » Sat Oct 18, 2014 6:19 pm

Elgee, if you factor in salary, shouldn't high peak guys get rated even higher? LeBron gets the max and he is probably a +8 player. But Chris Bosh/Joe Johnson/Carmelo get that same max too despite having half the value of LeBron. That would make elite players more important because they alongside with rookie contract players are the only players in the NBA with surplus value relative to their contract.

ElGee · Post #24 » by **ElGee** » Sun Oct 19, 2014 12:09 am

colts18 wrote:Elgee, if you factor in salary, shouldn't high peak guys get rated even higher? LeBron gets the max and he is probably a +8 player. But Chris Bosh/Joe Johnson/Carmelo get that same max too despite having half the value of LeBron. That would make elite players more important because they alongside with rookie contract players are the only players in the NBA with surplus value relative to their contract.

Up front -- I'm not 100% certain on what the data says with salary, and that's the first place I'd look.

With that said, the way I'm looking at this (since the economic environment in the league is always changing) is something like this Teams don't have an unlimited amount of money to spend. Players who eat up huge salaries (these days, a Max-Contract player) occupy resources that make it less likely for the supporting cast to be elite. This is independent of other players getting overpaid.

In general, you're point is correct and reflected in a model that is incorporating this variable. Here is a "salary" adjustment that falls between "no adjustment" and the "35%" model I presented earlier. My guess is that this is closer to real life, and it's the current version I'm using to help with things like GOAT lists:

Which yields odds over replacement like this for "normal" portability and 100% health:

So yes, what happens is that once you hit "max contract" land (+3.5, +4, +4.5 for most players) you see some serious growth in value because it's no longer really less likely to be on a good -- players can only occupy so much salary.

With that said, here's my latest top-50 you were asking about based on this model:

Rank - Player - Titles above Replacement -- Peak Season above Replacement
1 Jordan 3.21 36.0%
2 Russell 3.00 30.1%
3 Kareem 3.00 25.8%
4 Olajuwon 2.86 32.3%
5 Duncan 2.66 29.6%
6 Shaq 2.65 37.8%
7 KG 2.63 25.6%
8 LBJ 2.39 32.4%
9 Bird 2.33 30.8%
10 Chamberlain 2.28 30.7%
11 Malone 2.17 18.9%
12 Magic 2.16 27.1%
13 Erving 2.07 26.0%
14 Kobe 2.05 20.0%
15 Robinson 1.95 26.9%
16 Dirk 1.87 18.7%
17 Oscar 1.85 20.1%
18 West 1.67 22.1%
19 Barkley 1.47 20.2%
20 Nash 1.39 18.1%
21 Ewing 1.34 19.5%
22 Pippen 1.31 15.2%
23 Wade 1.29 22.4%
24 Stockton 1.26 10.1%
25 Miller 1.24 10.4%
26 Moses 1.22 17.9%
27 Paul 1.15 20.1%
28 Pierce 1.14 12.8%
29 Barry 1.12 16.2%
30 Payton 1.07 12.2%
31 Mourning 1.06 16.3%
32 Mutombo 1.06 12.4%
33 McHale 1.05 14.6%
34 Havlicek 1.05 12.0%
35 Gilmore 1.05 14.4%
36 Drexler 1.02 11.9%
37 Kidd 1.00 9.6%
38 Pettit 0.98 12.4%
39 Frazier 0.98 15.7%
40 Allen 0.95 8.0%
41 McGrady 0.94 18.0%
42 Thomas 0.91 11.5%
43 Gasol 0.91 11.1%
44 Cowens 0.90 14.1%
45 Baylor 0.89 13.6%
46 Hayes 0.89 8.8%
47 Gervin 0.86 10.9%
48 Parish 0.83 9.5%
49 Durant 0.82 20.0%
50 Parker 0.80 9.4%

colts18 · Post #25 » by **colts18** » Sun Oct 19, 2014 12:42 am

Great post, Elgee. A few things jumped out to me on your list.

-Tony Parker in the top 50. Never saw that coming, especially ahead of Manu.
-No Dwight Howard
-You are pretty low on Magic and Kareem's peak

Top 10 from your list based on peak:
Shaq 37.80%
Jordan 36.00%
LBJ 32.40%
Olajuwon 32.30%
Bird 30.80%
Chamberlain 30.70%
Russell 30.10%
Duncan 29.60%
Magic 27.10%
Robinson 26.90%

Interesting that you have Shaq as the top peak. I remember that your stats were low on him a few years ago so I was surprised to see him #1 in the new version.

DQuinn1575 · Post #26 » by **DQuinn1575** » Sun Oct 19, 2014 2:50 am

ElGee wrote:so I think the premise is correct (harder for better players to be on good teams),
the question is simply to what degree...

I agree with your premise - what I misunderstood, or took too much impact on was your points that:

ElGee wrote:.(Although ego plays a part.)

and

ElGee wrote:.Imagine a player who thinks he THE MAN

THe point I was trying to make is that it doesn't occur because of Ego - if that were true we would see guys moving to
lesser teams to be a star, and that doesn't happen often.

Not to get lost in the weeds, but Archie Clark led the Lakers in minutes played, and had TS% much better than Elgin, and Win Shares
about the same as him. But I really really don't want to argue that on this thread.

So we agree on the premise that it is harder for great players to get added to great teams.

SideshowBob · Post #27 » by **SideshowBob** » Sun Oct 19, 2014 3:06 am

ElGee wrote:
Spoiler:
colts18 wrote:Elgee, if you factor in salary, shouldn't high peak guys get rated even higher? LeBron gets the max and he is probably a +8 player. But Chris Bosh/Joe Johnson/Carmelo get that same max too despite having half the value of LeBron. That would make elite players more important because they alongside with rookie contract players are the only players in the NBA with surplus value relative to their contract.

Up front -- I'm not 100% certain on what the data says with salary, and that's the first place I'd look.

With that said, the way I'm looking at this (since the economic environment in the league is always changing) is something like this Teams don't have an unlimited amount of money to spend. Players who eat up huge salaries (these days, a Max-Contract player) occupy resources that make it less likely for the supporting cast to be elite. This is independent of other players getting overpaid.

In general, you're point is correct and reflected in a model that is incorporating this variable. Here is a "salary" adjustment that falls between "no adjustment" and the "35%" model I presented earlier. My guess is that this is closer to real life, and it's the current version I'm using to help with things like GOAT lists:

For +2 players and better: 0.0063*Impact^2-0.0118*Impact+0.0744
For +1.5 players or worse: 0.0025*impact^2 + 0.0148*Impact + 0.0358

Which yields odds over replacement like this for "normal" portability and 100% health:
9 46.6%
8.5 41.6%
8 37.0%
7.5 32.7%
7 28.8%
6.5 25.1%
6 21.7%
5.5 18.7%
5 16.0%
4.5 13.6%
4 11.5%
3.5 9.7%
3 8.3%
2.5 7.1%
2 6.3%
1.5 5.1%
1 4.0%
0.5 3.1%
0 2.3%
-0.5 1.6%
-1 1.1%
-1.5 0.6%
-2 0.3%
-2.5 0.1%
-3 0.1%

So yes, what happens is that once you hit "max contract" land (+3.5, +4, +4.5 for most players) you see some serious growth in value because it's no longer really less likely to be on a good -- players can only occupy so much salary.

With that said, here's my latest top-50 you were asking about based on this model:

Rank - Player - Titles above Replacement -- Peak Season above Replacement
1 Jordan 3.21 36.0%
2 Russell 3.00 30.1%
3 Kareem 3.00 25.8%
4 Olajuwon 2.86 32.3%
5 Duncan 2.66 29.6%
6 Shaq 2.65 37.8%
7 KG 2.63 25.6%
8 LBJ 2.39 32.4%
9 Bird 2.33 30.8%
10 Chamberlain 2.28 30.7%
11 Malone 2.17 18.9%
12 Magic 2.16 27.1%
13 Erving 2.07 26.0%
14 Kobe 2.05 20.0%
15 Robinson 1.95 26.9%
16 Dirk 1.87 18.7%
17 Oscar 1.85 20.1%
18 West 1.67 22.1%
19 Barkley 1.47 20.2%
20 Nash 1.39 18.1%
21 Ewing 1.34 19.5%
22 Pippen 1.31 15.2%
23 Wade 1.29 22.4%
24 Stockton 1.26 10.1%
25 Miller 1.24 10.4%
26 Moses 1.22 17.9%
27 Paul 1.15 20.1%
28 Pierce 1.14 12.8%
29 Barry 1.12 16.2%
30 Payton 1.07 12.2%
31 Mourning 1.06 16.3%
32 Mutombo 1.06 12.4%
33 McHale 1.05 14.6%
34 Havlicek 1.05 12.0%
35 Gilmore 1.05 14.4%
36 Drexler 1.02 11.9%
37 Kidd 1.00 9.6%
38 Pettit 0.98 12.4%
39 Frazier 0.98 15.7%
40 Allen 0.95 8.0%
41 McGrady 0.94 18.0%
42 Thomas 0.91 11.5%
43 Gasol 0.91 11.1%
44 Cowens 0.90 14.1%
45 Baylor 0.89 13.6%
46 Hayes 0.89 8.8%
47 Gervin 0.86 10.9%
48 Parish 0.83 9.5%
49 Durant 0.82 20.0%
50 Parker 0.80 9.4%

So, I recall you mentioning that you felt that perhaps you'd consider tightening your spread due to underrating how much sub-optimal situations dragging down average player impact. Judging by your updated list vs. the one you posted last spring on the PC board (as well as the peaks project), it seems like you've made that adjustment now, no?

If you don't mind, could you get into a brief summary on the new historical top ends for offense/defense? I understand that qualitatively your views are the same, but just want to grasp how you go about fitting that on to your new scale.

ElGee · Post #28 » by **ElGee** » Sun Oct 19, 2014 3:32 am

colts18 wrote:Great post, Elgee. A few things jumped out to me on your list.

-Tony Parker in the top 50. Never saw that coming, especially ahead of Manu.
-No Dwight Howard
-You are pretty low on Magic and Kareem's peak

Top 10 from your list based on peak:
Shaq 37.80%
Jordan 36.00%
LBJ 32.40%
Olajuwon 32.30%
Bird 30.80%
Chamberlain 30.70%
Russell 30.10%
Duncan 29.60%
Magic 27.10%
Robinson 26.90%

Interesting that you have Shaq as the top peak. I remember that your stats were low on him a few years ago so I was surprised to see him #1 in the new version.

-Manu is right behind at 0.75 -- has a better peak, but loses the "race" to Parker because of injuries that wore him down in 2008 and then caused him to miss 2009, along with the series in 2011 vs. Memphis. Parker has a nice little peak and has piled up some good longevity -- a number of good years from 2007-2014.

-With Dwight, it's not that I'm necessarily low on his peak (14.4%). It's that he has major longevity issues in a list like this. I'm not saying he can't be top-50, but he's in the 50's for me because he's got 2008-2011, a half-speed 2013 and 2014 at a lower level than his 2009-2011 years. It's like a 3-year peak with 2 or 3 all-star years around it.

-Kareem's peak I'm pretty comfortable with, I think if you go higher on Kareem's peak you really should consider him GOAT. I've spoken about it a lot in the past. Magic I'm a touch down on lately and I'll explain why -- it primarily has to do with portability/team-building.

I think Magic's game makes him the GOAT at taking just about any offense below about +2 and making them an elite offense, GOAT-level passer. Fast breaks. Post game. PnR game. Pressures the absolute crap out of defenses. But he needs the ball, and I think this creates some diminishing returns, The natural question is "but they had like, the GOAT offense -- isn't that proof of scalability?" Sort of. One, they did a lot with offensive rebounding, Second, it's not that I don't think Magic's still great with great talent around him, just more like that I think his Lakers teams were kind of awesome offensively without him. Forget Cap for a second -- I'm really high on Worthy's game (shooting, passing, cutting), Scott is a great shooter to space the floor, Cooper could pass and shoot, and then there's the rebounding. I actually think Bird would have hit greater heights with an offense like that. Then, I don't really like Mqgic's defense relative to all these other elite guys.

-Shaq as the top peak is a new thing for me. I'm not sure how comfortable I am with it, but I'm not uncomfortable. It's possible I make a minor adjustment and he drops back -- I do find his offense hard to gauge because he's so unique. At those levels, everything is magnified so a minor change in my evaluation will move the needle a few percentage points. Frankly, I'd love people to do some in depth analysis on Shaq's offense.

ElGee · Post #29 » by **ElGee** » Sun Oct 19, 2014 4:07 am

SideshowBob wrote:
ElGee wrote:
Spoiler:
colts18 wrote:Elgee, if you factor in salary, shouldn't high peak guys get rated even higher? LeBron gets the max and he is probably a +8 player. But Chris Bosh/Joe Johnson/Carmelo get that same max too despite having half the value of LeBron. That would make elite players more important because they alongside with rookie contract players are the only players in the NBA with surplus value relative to their contract.

Up front -- I'm not 100% certain on what the data says with salary, and that's the first place I'd look.

With that said, the way I'm looking at this (since the economic environment in the league is always changing) is something like this Teams don't have an unlimited amount of money to spend. Players who eat up huge salaries (these days, a Max-Contract player) occupy resources that make it less likely for the supporting cast to be elite. This is independent of other players getting overpaid.

In general, you're point is correct and reflected in a model that is incorporating this variable. Here is a "salary" adjustment that falls between "no adjustment" and the "35%" model I presented earlier. My guess is that this is closer to real life, and it's the current version I'm using to help with things like GOAT lists:

For +2 players and better: 0.0063*Impact^2-0.0118*Impact+0.0744
For +1.5 players or worse: 0.0025*impact^2 + 0.0148*Impact + 0.0358

Which yields odds over replacement like this for "normal" portability and 100% health:
9 46.6%
8.5 41.6%
8 37.0%
7.5 32.7%
7 28.8%
6.5 25.1%
6 21.7%
5.5 18.7%
5 16.0%
4.5 13.6%
4 11.5%
3.5 9.7%
3 8.3%
2.5 7.1%
2 6.3%
1.5 5.1%
1 4.0%
0.5 3.1%
0 2.3%
-0.5 1.6%
-1 1.1%
-1.5 0.6%
-2 0.3%
-2.5 0.1%
-3 0.1%

So yes, what happens is that once you hit "max contract" land (+3.5, +4, +4.5 for most players) you see some serious growth in value because it's no longer really less likely to be on a good -- players can only occupy so much salary.

With that said, here's my latest top-50 you were asking about based on this model:

Rank - Player - Titles above Replacement -- Peak Season above Replacement
1 Jordan 3.21 36.0%
2 Russell 3.00 30.1%
3 Kareem 3.00 25.8%
4 Olajuwon 2.86 32.3%
5 Duncan 2.66 29.6%
6 Shaq 2.65 37.8%
7 KG 2.63 25.6%
8 LBJ 2.39 32.4%
9 Bird 2.33 30.8%
10 Chamberlain 2.28 30.7%
11 Malone 2.17 18.9%
12 Magic 2.16 27.1%
13 Erving 2.07 26.0%
14 Kobe 2.05 20.0%
15 Robinson 1.95 26.9%
16 Dirk 1.87 18.7%
17 Oscar 1.85 20.1%
18 West 1.67 22.1%
19 Barkley 1.47 20.2%
20 Nash 1.39 18.1%
21 Ewing 1.34 19.5%
22 Pippen 1.31 15.2%
23 Wade 1.29 22.4%
24 Stockton 1.26 10.1%
25 Miller 1.24 10.4%
26 Moses 1.22 17.9%
27 Paul 1.15 20.1%
28 Pierce 1.14 12.8%
29 Barry 1.12 16.2%
30 Payton 1.07 12.2%
31 Mourning 1.06 16.3%
32 Mutombo 1.06 12.4%
33 McHale 1.05 14.6%
34 Havlicek 1.05 12.0%
35 Gilmore 1.05 14.4%
36 Drexler 1.02 11.9%
37 Kidd 1.00 9.6%
38 Pettit 0.98 12.4%
39 Frazier 0.98 15.7%
40 Allen 0.95 8.0%
41 McGrady 0.94 18.0%
42 Thomas 0.91 11.5%
43 Gasol 0.91 11.1%
44 Cowens 0.90 14.1%
45 Baylor 0.89 13.6%
46 Hayes 0.89 8.8%
47 Gervin 0.86 10.9%
48 Parish 0.83 9.5%
49 Durant 0.82 20.0%
50 Parker 0.80 9.4%

So, I recall you mentioning that you felt that perhaps you'd consider tightening your spread due to underrating how much sub-optimal situations dragging down average player impact. Judging by your updated list vs. the one you posted last spring on the PC board (as well as the peaks project), it seems like you've made that adjustment now, no?

If you don't mind, could you get into a brief summary on the new historical top ends for offense/defense? I understand that qualitatively your views are the same, but just want to grasp how you go about fitting that on to your new scale.

So yeah, in general, I see things as more compacted. I think it's a bit of an illusion that MVP-level players are that much better than the top-8, top-12 guys on a point-per-game basis.

I'm still using "how good is the player on a 0 SRS team" - I'll post those numbers in a second -- but I have a 5-point portability scale on offense (from -2 to +2) which shapes the curve for players as they would impact better and better teams. As for the heights of players by SRS: Old-time defense (Russell) I have at +7. Modern bigs peaking at +4.5 and wings at +2.5.

it's a little trickier because of the portability curve. For offense, I do have Magic at +7 but with weaker portability to carry that value (see previous post) and Bird at +6.75 holding his value more on better teams. This is a critical evolution for me in using this scale to try and guide me -- it's not that Magic has "bad portability," it's just that his kick-ass value degrades more steeply IMO than someone like Bird's. In other words, I'm less interested in where a player crosses 0 (SRS on a neutral team) and more interested in modeling the whole shape of their curves as they impact better and better teams than I have ever been before.

Overall, I have Jordan and Shaq combining to +8, and it's possible I'll move off of anyone being +8 in the future. (+8, regardless of portability, means if I put that guy on a random team 0 SRS team, they would average 62.5 wins. This is right in the ballpark of where it needs to be for me guys like Jordan and Shaq, but it still might be a touch high.)

SideshowBob · Post #30 » by **SideshowBob** » Sun Oct 19, 2014 4:46 am

ElGee wrote:So yeah, in general, I see things as more compacted. I think it's a bit of an illusion that MVP-level players are that much better than the top-8, top-12 guys on a point-per-game basis.

I'm still using "how good is the player on a 0 SRS team" - I'll post those numbers in a second -- but I have a 5-point portability scale on offense (from -2 to +2) which shapes the curve for players as they would impact better and better teams.

Cool. This is where I figured you were going after checking out your updated odds. I think I had a brief discussion with fpliii on the same matter, introducing a slightly more nuanced portability scale (going beyond the simple Low/Avg/High). Makes sense IMO.

As for the heights of players by SRS: Old-time defense (Russell) I have at +7. Modern bigs peaking at +4.5 and wings at +2.5.

It's a little trickier because of the portability curve. For offense, I do have Magic at +7 but with weaker portability to carry that value (see previous post) and Bird at +6.75 holding his value more on better teams. This is a critical evolution for me in using this scale to try and guide me -- it's not that Magic has "bad portability," it's just that his kick-ass value degrades more steeply IMO than someone like Bird's. In other words, I'm less interested in where a player crosses 0 (SRS on a neutral team) and more interested in modeling the whole shape of their curves as they impact better and better teams than I have ever been before.

Again, this makes a lot of sense. I know you've spent a lot more time developing this line of thinking. I think a good step forward for the rest of us on the PC board (now that we see folks really getting comfortable with the idea of portability) would be to have an extended dialogue beginning to hash out the dynamics that create distinction between these different "levels" (in terms of particular skillsets, quality of those skillsets, etc.).

It's a good move forward IMO.

Overall, I have Jordan and Shaq combining to +8, and it's possible I'll move off of anyone being +8 in the future. (+8, regardless of portability, means if I put that guy on a random team 0 SRS team, they would average 62.5 wins. This is right in the ballpark of where it needs to be for me guys like Jordan and Shaq, but it still might be a touch high.)

Got it. And Shaq comes out on top due to a larger % of that coming from defense I believe.

colts18 · Post #31 » by **colts18** » Tue Oct 21, 2014 2:15 pm

Elgee, a few questions on your methodology and portability:

1. You mention that superstars are more likely to be on a bad team than great team. Shouldn't portability be a sliding scale in that case? The higher your SRS impact is, the less portability matters. Typically worse teams need a ball dominant guy and high SRS impact guys are usually ball dominant. That would mean that portability matters less for a guy like Kobe because he is more than likely to be on a team that needs ball dominance whereas Reggie Miller is a lesser player but would be on a better team therefore needing more portability and less ball dominance.

2. Do you add in era portability in your calculation? Someone like Ray Allen would have less value because for half of NBA history there was no 3 point line. Does Bill Russell's impact become lower because his value would decline for half of NBA history.

3. Do you account for defensive portability? There is 2 types of defensive portability to me: Versatility and skill sets. LeBron gets a lot of points for being versatile. His team can play him 1-4 occasional 5 so his defensive portability is off the scale. You can add him to any team and they will be better defensively because he can guard whatever weakspot 1-4 that the team has. Larry Bird while being a solid defender gets dinged because his team needs someone to cover the SF for him. If he is on a team like San Antonio in place of Kawhi Leonard, the Spurs D suffers because Bird would have to cover quick SF's due to the fact that Duncan/Splitter can't do that.

There is also defensive skill set portability. Being a rim protector is valuable because every team could use a rim protector. Even if your team has rim protection, adding another one won't hurt them. Look at San Antonio, adding Duncan didn't result in diminishing returns. He added a lot of value next to David Robinson. A good defensive rebounder has less portability because adding rebounding for a lot of teams doesn't add too much value. A good post man defender has less value because its useless against half the teams in the league while a good perimeter man defender has value because he can guard up to 3 positions.

ElGee · Post #32 » by **ElGee** » Tue Oct 21, 2014 4:34 pm

colts18 wrote:Elgee, a few questions on your methodology and portability:

1. You mention that superstars are more likely to be on a bad team than great team. Shouldn't portability be a sliding scale in that case? The higher your SRS impact is, the less portability matters. Typically worse teams need a ball dominant guy and high SRS impact guys are usually ball dominant. That would mean that portability matters less for a guy like Kobe because he is more than likely to be on a team that needs ball dominance whereas Reggie Miller is a lesser player but would be on a better team therefore needing more portability and less ball dominance.

2. Do you add in era portability in your calculation? Someone like Ray Allen would have less value because for half of NBA history there was no 3 point line. Does Bill Russell's impact become lower because his value would decline for half of NBA history.

3. Do you account for defensive portability? There is 2 types of defensive portability to me: Versatility and skill sets. LeBron gets a lot of points for being versatile. His team can play him 1-4 occasional 5 so his defensive portability is off the scale. You can add him to any team and they will be better defensively because he can guard whatever weakspot 1-4 that the team has. Larry Bird while being a solid defender gets dinged because his team needs someone to cover the SF for him. If he is on a team like San Antonio in place of Kawhi Leonard, the Spurs D suffers because Bird would have to cover quick SF's due to the fact that Duncan/Splitter can't do that.

There is also defensive skill set portability. Being a rim protector is valuable because every team could use a rim protector. Even if your team has rim protection, adding another one won't hurt them. Look at San Antonio, adding Duncan didn't result in diminishing returns. He added a lot of value next to David Robinson. A good defensive rebounder has less portability because adding rebounding for a lot of teams doesn't add too much value. A good post man defender has less value because its useless against half the teams in the league while a good perimeter man defender has value because he can guard up to 3 positions.

Good questions.

1. Hmm -- that's a bit circular. If you are ball-dominant (e.g. Magic, Nash, Oscar) you are ball-dominant. If you can impact the game without being ball dominant (e.g. Bird) at a high-level, then you can blend better with a ball-dominant player.

2. No -- For things like top 100, I'm looking at players in their own era.

3. I now treat offensive and defensive portability separately in my title odds calculation model.

With that said, I don't agree with you -- I think defensive portability is a relative constant. Relative is the key word here -- it's true that LeBron adds some extra value in rare circumstances where he isn't playing his "normal" defensive role, but I think those are rare circumstances and thus have minimal impact on title odds. In general, great defensive players do their thing defensively unless they are asked to play out of position...which they are rarely asked to do. The question becomes how much extra value can a guy carry (LeBron? McHale? Duncan? Kidd?) being asked to play out of position? I see it as inconsequential to the evaluation.

DQuinn1575 · Post #33 » by **DQuinn1575** » Tue Oct 21, 2014 10:24 pm

A couple of questions for you:

1 is it fair to penalize the rating of a great player for being more likely on a bad team? I realize it is reality, but a) he has little control over that, and b) it may affect the odds but doesn't change his talent. Maybe the inverse of that is true- you are solving for the odds and not the talent. If so I'm good with that.

2, I struggle with portability because it seems very judgmental and not very quantifiable. I'm not saying it doesn't exist, my concern is it appears to be very open to interpretation

Thanks

Sent from my iPhone using RealGM Forums

ElGee · Post #34 » by **ElGee** » Thu Oct 23, 2014 1:17 am

DQuinn1575 wrote:A couple of questions for you:

1 is it fair to penalize the rating of a great player for being more likely on a bad team? I realize it is reality, but a) he has little control over that, and b) it may affect the odds but doesn't change his talent. Maybe the inverse of that is true- you are solving for the odds and not the talent. If so I'm good with that.

2, I struggle with portability because it seems very judgmental and not very quantifiable. I'm not saying it doesn't exist, my concern is it appears to be very open to interpretation

Thanks

Sent from my iPhone using RealGM Forums

1. I have the same question. "Solving for the odds" is not necessarily the same solving for talent, and thus penalizes some. I'm not wild about that concept. What I am good with -- and want to do -- is to model factors that matter. It might seem strange, but think through the entire thing:

The last point is essentially a "rules of the game" thing as well as an inescapable result of the distribution of talent in basketball. So while odds != talent, the question the "odds" part of that equation addresses still perfectly correlates to talent (in theory).

2. I don't think portability is any more subjective than just ranking a player. It is, indeed, part of the evaluation process. A critical part. Because if you are interested in winning in basketball, you need to build elite teams, and those cannot consist of players who are heavily redundant with one another. It needs to consist of players who fill up the key dimensions of success for a team. (Portability really is more like "scalability," but when the concept was first broached it was about the revelation that if you're skills only fit in a single environment, you aren't as good as if you're skills fit in any environment. Oh well.)

Iso scorers are the poster boys for anti-portability. Without delving into deep theory (and I will publish something shortly on this theory in detail) think of it like this:
-holding the ball for a long time in a possession prevents others from pressuring the D
-creating mediocre scoring opportunities from nothing will make horrible offenses better, but it will make great offenses worse.

Thus, as a simplistic thought experiment, imagine a player who can score 100 pts/g on 50% TS with his isolation offense. If you surround him with the worst team ever, and have them run the offense through him, they will have a 100 ORTg. If you then surround him with the Dream Team, but still run every possession through him, the team will have a 100 ORtg. So if the player can't do other things (like pass/create well for others) that help the Global Offense of his improved teammates (as in, literally take other actions than his 100 ppg/50% TS offense) he's not actually "very good" on offense.

Spoiler:

DQuinn1575 · Post #35 » by **DQuinn1575** » Thu Oct 23, 2014 4:01 am

Thanks for 1.

2. I don't think portability is any more subjective than just ranking a player. [/quote]

As long as we understand that it is a rather subjective measure, then I'm okay with it.
I'm not disagreeing with anything you are saying about it.

colts18 · Post #36 » by **colts18** » Thu Oct 23, 2014 8:25 pm

ElGee wrote:Good questions.

1. Hmm -- that's a bit circular. If you are ball-dominant (e.g. Magic, Nash, Oscar) you are ball-dominant. If you can impact the game without being ball dominant (e.g. Bird) at a high-level, then you can blend better with a ball-dominant player.

2. No -- For things like top 100, I'm looking at players in their own era.

3. I now treat offensive and defensive portability separately in my title odds calculation model.

With that said, I don't agree with you -- I think defensive portability is a relative constant. Relative is the key word here -- it's true that LeBron adds some extra value in rare circumstances where he isn't playing his "normal" defensive role, but I think those are rare circumstances and thus have minimal impact on title odds. In general, great defensive players do their thing defensively unless they are asked to play out of position...which they are rarely asked to do. The question becomes how much extra value can a guy carry (LeBron? McHale? Duncan? Kidd?) being asked to play out of position? I see it as inconsequential to the evaluation.

1. Is it really fair to dock ball dominant guys points from portability when almost every team with a ball dominant guy tried to build their team around him rather focus on portability? Its rare to have a situation like 2011 Miami with 2 ball dominant guys on the same time. They almost always play on different teams. Im not sure its right to dock Michael Jordan for being ball dominant when every single team would build around his ball dominance and would never play him off ball

2. If you aren't looking at era portability, why even use portability at all? I see little difference between era portability and in season portability. Both of them are hypotheticals. If you want to look at how a player does in different situations (portability) then it also makes sense to include situations from different eras.

3. Defenders do guard other positions quite often. For example LeBron guarded Tiago Splitter, Tony Parker, David West, and Paul George all in the same playoff. There is some value there. But I think the versatility provides value for 2 reasons:

1. You can hide your bad defenders due to versatile defenders. The Heat could play Mike Miller because LeBron and Wade would take on the top 2 perimeter threats while Miller could guard the weakest. LeBron is the reason they are able to get away with that.

2. An indirect benefit to the defensive versatility is offensive value provided by it. You can play offensive lineups as a result of having a LeBron or Hakeem on the court. LeBron was able to guard David West on one end but West couldn't guard LeBron which created matchup issues for the Pacers. As a result David West was guarding Ray Allen in crunch time. Thats a huge win for the Heat in terms of offense. The Heat weren't mismatched defensively because Ray Allen could guard his natural position (SG) and LeBron could guard David West. In the 2013 finals those matchup issues presented themselves again. LeBron guarded Splitter on one end but Splitter was forced to guard either LeBron or Wade. Thats a huge mismatch that adds offensive value to the Heat.

Another question:
What skills are you using in portability calculations? Rank the skills that add the most to least portability? What defensive skills are you looking at in your defensive portability calculations?

ElGee · Post #37 » by **ElGee** » Fri Oct 24, 2014 5:49 am

colts18 wrote:
ElGee wrote:Good questions.

1. Hmm -- that's a bit circular. If you are ball-dominant (e.g. Magic, Nash, Oscar) you are ball-dominant. If you can impact the game without being ball dominant (e.g. Bird) at a high-level, then you can blend better with a ball-dominant player.

2. No -- For things like top 100, I'm looking at players in their own era.

3. I now treat offensive and defensive portability separately in my title odds calculation model.

With that said, I don't agree with you -- I think defensive portability is a relative constant. Relative is the key word here -- it's true that LeBron adds some extra value in rare circumstances where he isn't playing his "normal" defensive role, but I think those are rare circumstances and thus have minimal impact on title odds. In general, great defensive players do their thing defensively unless they are asked to play out of position...which they are rarely asked to do. The question becomes how much extra value can a guy carry (LeBron? McHale? Duncan? Kidd?) being asked to play out of position? I see it as inconsequential to the evaluation.

1. Is it really fair to dock ball dominant guys points from portability when almost every team with a ball dominant guy tried to build their team around him rather focus on portability? Its rare to have a situation like 2011 Miami with 2 ball dominant guys on the same time. They almost always play on different teams. Im not sure its right to dock Michael Jordan for being ball dominant when every single team would build around his ball dominance and would never play him off ball

2. If you aren't looking at era portability, why even use portability at all? I see little difference between era portability and in season portability. Both of them are hypotheticals. If you want to look at how a player does in different situations (portability) then it also makes sense to include situations from different eras.

3. Defenders do guard other positions quite often. For example LeBron guarded Tiago Splitter, Tony Parker, David West, and Paul George all in the same playoff. There is some value there. But I think the versatility provides value for 2 reasons:

1. You can hide your bad defenders due to versatile defenders. The Heat could play Mike Miller because LeBron and Wade would take on the top 2 perimeter threats while Miller could guard the weakest. LeBron is the reason they are able to get away with that.

2. An indirect benefit to the defensive versatility is offensive value provided by it. You can play offensive lineups as a result of having a LeBron or Hakeem on the court. LeBron was able to guard David West on one end but West couldn't guard LeBron which created matchup issues for the Pacers. As a result David West was guarding Ray Allen in crunch time. Thats a huge win for the Heat in terms of offense. The Heat weren't mismatched defensively because Ray Allen could guard his natural position (SG) and LeBron could guard David West. In the 2013 finals those matchup issues presented themselves again. LeBron guarded Splitter on one end but Splitter was forced to guard either LeBron or Wade. Thats a huge mismatch that adds offensive value to the Heat.

Another question:
What skills are you using in portability calculations? Rank the skills that add the most to least portability? What defensive skills are you looking at in your defensive portability calculations?

1. It's not necessarily "docking" them -- it's a reflection of reality. As you build offenses and fill up the dimensions that make teams successful, certain skills are redundant or carry greater diminishing returns. There's only one ball. It's one of the reasons why the Hoosiers, unselfish, 70's Knicks, 77 Blazers, 80's Celtics all the way to Pop's Spurs last year have romanticized ball movement and the ball not sticking...it's simply the nature of the game.

2. I answered cross-era comparisons in the latest top 100 project in some detail. The gist is this: in-season portability is an abstraction of skills you watch, using current exemplars (similar player-types/skills) and a variety of lineups within a given team that affords insight into different roles and teammates. It's hypothetical, it's abstraction, but it's in the ballpark of what most people (fans, GM's, coaches, etc.) do anyway. Conversely, cross-era requires an exponentially larger bit of abstraction because of all the additional combined variables. You have "what-ifs" about the basic development of skills (e.g. 3-point shot) that involve guessing the basic skills. Those two thinks are night and day to me.

3. You're talking about versatility, not portability. LeBron can do that (assuming he's the best equipped on his team) whether he is surrounded by 4 defensive scrubs or 4 good defenders. And the reason I think defensive portability is almost a constant is It becomes less likely that he guards those guys on a great defensive team, because his teammates typically take the assignments.

Your other points, I'm not seeing how they related to portability.

Other question -- it's not as simple as making a list because the way the player uses his skills matters a lot. I think Kobe has the skills to be hyper-portable, but he doesn't play like that. If I had to list skills, shooting and passing are by far the most dominant for melding with elite offenses. Offensive rebounding is a huge one as well. And in general, the ability to have value off the ball (drawing defenders or with some spacing). What all these things have in common is they are synergistic with the other 4 guys (Global impact) and not just the player himself. The least portable skills and offensive games follow the inverse pattern.

colts18 · Post #38 » by **colts18** » Mon Nov 17, 2014 5:40 pm

Elgee, I think Westbrook/Durant is going to be a great test for your system. How much does stars missing RS games really matter. Will the lack of HCA kill them, assuming they even make it? I would love to see them as a #8 seed because thats the best chance we have of seeing a #8 win a title or watch a #1 seed tank to avoid a #8 team.

SideshowBob · Post #39 » by **SideshowBob** » Tue Nov 18, 2014 8:03 pm

Is it though? This is an extreme scenario, not the norm.

PCProductions · Post #40 » by **PCProductions** » Wed Sep 14, 2016 9:39 pm

ElGee wrote:
Spoiler:
colts18 wrote:Elgee, if you factor in salary, shouldn't high peak guys get rated even higher? LeBron gets the max and he is probably a +8 player. But Chris Bosh/Joe Johnson/Carmelo get that same max too despite having half the value of LeBron. That would make elite players more important because they alongside with rookie contract players are the only players in the NBA with surplus value relative to their contract.

Up front -- I'm not 100% certain on what the data says with salary, and that's the first place I'd look.

With that said, the way I'm looking at this (since the economic environment in the league is always changing) is something like this Teams don't have an unlimited amount of money to spend. Players who eat up huge salaries (these days, a Max-Contract player) occupy resources that make it less likely for the supporting cast to be elite. This is independent of other players getting overpaid.

In general, you're point is correct and reflected in a model that is incorporating this variable. Here is a "salary" adjustment that falls between "no adjustment" and the "35%" model I presented earlier. My guess is that this is closer to real life, and it's the current version I'm using to help with things like GOAT lists:

For +2 players and better: 0.0063*Impact^2-0.0118*Impact+0.0744
For +1.5 players or worse: 0.0025*impact^2 + 0.0148*Impact + 0.0358

Which yields odds over replacement like this for "normal" portability and 100% health:
9 46.6%
8.5 41.6%
8 37.0%
7.5 32.7%
7 28.8%
6.5 25.1%
6 21.7%
5.5 18.7%
5 16.0%
4.5 13.6%
4 11.5%
3.5 9.7%
3 8.3%
2.5 7.1%
2 6.3%
1.5 5.1%
1 4.0%
0.5 3.1%
0 2.3%
-0.5 1.6%
-1 1.1%
-1.5 0.6%
-2 0.3%
-2.5 0.1%
-3 0.1%

So yes, what happens is that once you hit "max contract" land (+3.5, +4, +4.5 for most players) you see some serious growth in value because it's no longer really less likely to be on a good -- players can only occupy so much salary.

With that said, here's my latest top-50 you were asking about based on this model:

Rank - Player - Titles above Replacement -- Peak Season above Replacement
1 Jordan 3.21 36.0%
2 Russell 3.00 30.1%
3 Kareem 3.00 25.8%
4 Olajuwon 2.86 32.3%
5 Duncan 2.66 29.6%
6 Shaq 2.65 37.8%
7 KG 2.63 25.6%
8 LBJ 2.39 32.4%
9 Bird 2.33 30.8%
10 Chamberlain 2.28 30.7%
11 Malone 2.17 18.9%
12 Magic 2.16 27.1%
13 Erving 2.07 26.0%
14 Kobe 2.05 20.0%
15 Robinson 1.95 26.9%
16 Dirk 1.87 18.7%
17 Oscar 1.85 20.1%
18 West 1.67 22.1%
19 Barkley 1.47 20.2%
20 Nash 1.39 18.1%
21 Ewing 1.34 19.5%
22 Pippen 1.31 15.2%
23 Wade 1.29 22.4%
24 Stockton 1.26 10.1%
25 Miller 1.24 10.4%
26 Moses 1.22 17.9%
27 Paul 1.15 20.1%
28 Pierce 1.14 12.8%
29 Barry 1.12 16.2%
30 Payton 1.07 12.2%
31 Mourning 1.06 16.3%
32 Mutombo 1.06 12.4%
33 McHale 1.05 14.6%
34 Havlicek 1.05 12.0%
35 Gilmore 1.05 14.4%
36 Drexler 1.02 11.9%
37 Kidd 1.00 9.6%
38 Pettit 0.98 12.4%
39 Frazier 0.98 15.7%
40 Allen 0.95 8.0%
41 McGrady 0.94 18.0%
42 Thomas 0.91 11.5%
43 Gasol 0.91 11.1%
44 Cowens 0.90 14.1%
45 Baylor 0.89 13.6%
46 Hayes 0.89 8.8%
47 Gervin 0.86 10.9%
48 Parish 0.83 9.5%
49 Durant 0.82 20.0%
50 Parker 0.80 9.4%

I would understand if you don't, but do you have publicly available year-to-year SRS impact data for these players?