|
|
Post by Logan on Dec 4, 2020 20:00:52 GMT
ok here are 15 drafts from a top 5 only pool dating back to 1950. its a little weird and makes me think a top 5 ws pool would definitely be better than a top 5 draft slot pool
1. xavier mcdaniel, bob mcadoo, bob boozer, thomas robinson, kwame brown
2. isiah thomas, tristan thompson, raef lafrentz, rodney mccray, kendall gill
3. paul hogue, derrick coleman, darrall imhoff, tyson chandler, emeka okafor
4. walter davis, dikembe mutombo, lamelo ball, len bias, steve smith
5. zeke sinicola, charles barkley, danny vranes, marvin bagley, dominique wilkins
6. rasheed wallace, sidney moncrief, jaylen brown, gene shue, cody zeller
7. jason richardson, disck schnittker, andrea bargnani, mike miller, richard washington
8. shareef abdur-rahim, archie dees, gary payton, marvin williams, lou hudson
9. magic johnson, guy rodgers, andrew wiggins, eddy curry ,elgin baylor
10. jim krebs, scott may, mike conley, gary bradds, joe holup
11. bill hosket, johnny green, billy owens, brandon ingram, kenny walker
12. deron williams, jeff mullins, greg oden, austin carr, pau gasol
13. joel embiid, kevin mchale, aaron gordon, wally walker, art heyman
14. kris dunn, antonio mcdyess, dave bing, glenn robinson, baron davis
15. hakeem olajuwon, gerry ward, scottie pippen, vince carter, grant hill
|
|
|
|
Post by skrouse on Dec 4, 2020 20:01:10 GMT
Also mentioned in shout, if we don't want to go full RNG we could bring back declarations to add some randomness to when guys like LBJ come out
|
|
|
|
Post by Logan on Dec 4, 2020 20:05:13 GMT
Also mentioned in shout, if we don't want to go full RNG we could bring back declarations to add some randomness to when guys like LBJ come out i always hated them. its nice to be able to plan what youre going to do and theres too much randomness late in the process with declarations. you get 1.1 in the lebron draft but suddenly he isnt in there. with rng we would still know the class 5 years before giving plenty of time to plan if you see something you like. |
|
|
|
Post by Druce on Dec 4, 2020 20:13:04 GMT
With the win share thing you’re basically guaranteeing the all time greats get dispersed equally
|
|
|
|
Post by Druce on Dec 4, 2020 20:14:13 GMT
Hell you could even remove the randomness and snake draft the top 250 win share guys into the top 5 of 50 classes, so 1 and 100 are together while 50 and 51 are together etc
|
|
|
|
Post by eric on Dec 4, 2020 22:34:46 GMT
ok here are 15 drafts from a top 5 only pool dating back to 1950. its a little weird and makes me think a top 5 ws pool would definitely be better than a top 5 draft slot pool 1. xavier mcdaniel, bob mcadoo, bob boozer, thomas robinson, kwame brown 2. isiah thomas, tristan thompson, raef lafrentz, rodney mccray, kendall gill 3. paul hogue, derrick coleman, darrall imhoff, tyson chandler, emeka okafor 4. walter davis, dikembe mutombo, lamelo ball, len bias, steve smith 5. zeke sinicola, charles barkley, danny vranes, marvin bagley, dominique wilkins 6. rasheed wallace, sidney moncrief, jaylen brown, gene shue, cody zeller 7. jason richardson, disck schnittker, andrea bargnani, mike miller, richard washington 8. shareef abdur-rahim, archie dees, gary payton, marvin williams, lou hudson 9. magic johnson, guy rodgers, andrew wiggins, eddy curry ,elgin baylor 10. jim krebs, scott may, mike conley, gary bradds, joe holup 11. bill hosket, johnny green, billy owens, brandon ingram, kenny walker 12. deron williams, jeff mullins, greg oden, austin carr, pau gasol 13. joel embiid, kevin mchale, aaron gordon, wally walker, art heyman 14. kris dunn, antonio mcdyess, dave bing, glenn robinson, baron davis 15. hakeem olajuwon, gerry ward, scottie pippen, vince carter, grant hill this is great work i am strongly in agreement that we need to do something besides draft slot, that 2 class alone  i'm a little concerned that even the mighty ws won't be enough looking at those 13 and 15 classes, but i could be wrong |
|
|
|
Post by Logan on Dec 4, 2020 22:44:34 GMT
ok here are 15 drafts from a top 5 only pool dating back to 1950. its a little weird and makes me think a top 5 ws pool would definitely be better than a top 5 draft slot pool 1. xavier mcdaniel, bob mcadoo, bob boozer, thomas robinson, kwame brown 2. isiah thomas, tristan thompson, raef lafrentz, rodney mccray, kendall gill 3. paul hogue, derrick coleman, darrall imhoff, tyson chandler, emeka okafor 4. walter davis, dikembe mutombo, lamelo ball, len bias, steve smith 5. zeke sinicola, charles barkley, danny vranes, marvin bagley, dominique wilkins 6. rasheed wallace, sidney moncrief, jaylen brown, gene shue, cody zeller 7. jason richardson, disck schnittker, andrea bargnani, mike miller, richard washington 8. shareef abdur-rahim, archie dees, gary payton, marvin williams, lou hudson 9. magic johnson, guy rodgers, andrew wiggins, eddy curry ,elgin baylor 10. jim krebs, scott may, mike conley, gary bradds, joe holup 11. bill hosket, johnny green, billy owens, brandon ingram, kenny walker 12. deron williams, jeff mullins, greg oden, austin carr, pau gasol 13. joel embiid, kevin mchale, aaron gordon, wally walker, art heyman 14. kris dunn, antonio mcdyess, dave bing, glenn robinson, baron davis 15. hakeem olajuwon, gerry ward, scottie pippen, vince carter, grant hill this is great work i am strongly in agreement that we need to do something besides draft slot, that 2 class alone i'm a little concerned that even the mighty ws won't be enough looking at those 13 and 15 classes, but i could be wrong i think this is the best way to handle the "class imbalance" issue, we just need to figure out which metric to use to break up the pools. not breaking up until pools would be a huge disaster. |
|
|
|
Post by eric on Dec 4, 2020 23:16:21 GMT
this is great work i am strongly in agreement that we need to do something besides draft slot, that 2 class alone i'm a little concerned that even the mighty ws won't be enough looking at those 13 and 15 classes, but i could be wrong i think this is the best way to handle the "class imbalance" issue, we just need to figure out which metric to use to break up the pools. not breaking up until pools would be a huge disaster. yeah i think we're in agreement here. i'm not saying that the historical classes are perfect, i just wanna see how much of an improvement WS based classes are relative to how much work it takes to do them and the loss of verisimilitude, plus the historical issues of if there's only one historical three point shooter in a class they'll get a huge bump because 3 > 2 but we don't really want steve novak to be a 1.1 candidate i would have guessed draft slot randomization would have done a decent job of improving the situation but thanks to your work we've seen that's definitely, definitely not true. f***ing kendall gill and jim krebs i mean come on i've used this comp a lot before and i don't mean to belabor it but i still think it's true: it's very easy to see the actual flaws in the thing we're using and very easy to imagine the thing we're not using will be better, so it's a good idea to look at what the thing we're not using will actually do before we jump in and i think that's especially true with something like this that would be a lot more difficult to go away from after. like if we happen to take kevin love and daj from the 2008 class and then decide after a few seasons of 6.0 that this is a bust, it's not like we can put them back in the 2008 class, so then it just has no top tier players whatsoever, we don't want that |
|
|
|
Post by Druce on Dec 5, 2020 7:43:29 GMT
Here is the snake draft for the top 250 in win shares. Excel file attached 1 Michael Jordan* Tracy McGrady Paul Millsap A.C. Green John Beasley 2 David Robinson* Dikembe Mutombo Terry Dischinger Bob Dandridge Lou Hudson 3 Wilt Chamberlain* Larry Faust John Drew Mike Conley Ed Pinckney 4 Chris Paul Paul George Andrei Kirilenko Brook Lopez Larry Johnson 5 Neil Johnston* Jeff Hornacek Bob McAdoo* Rudy Tomjanovich Gordon Hayward 6 LeBron James Frank Ramsey Elton Brand Draymond Green J.J. Redick 7 Kareem Abdul-Jabbar* Robert Parish Ray Allen* Archie Clark Rashard Lewis 8 James Harden LaMarcus Aldridge David Lee Chris Webber James Worthy* 9 Magic Johnson* Tyson Chandler Terry Porter Dan Roundfield Vince Carter 10 Kawhi Leonard Paul Westphal Brad Miller Calvin Murphy Alvan Adams 11 Anthony Davis Willis Reed Tom Heinsohn* Donnie Freeman Al Jefferson 12 Kevin Durant Brad Daugherty Shawn Marion Jason Kidd Donald Sidle 13 Charles Barkley* David Thompson Patrick Ewing* Tim Hardaway Red Robbins 14 Jerry West* Detlef Schrempf Dennis Rodman* Michael Redd Billy Knight 15 Bob Pettit* Kyle Lowry Bill Laimbeer Ryan Anderson Darryl Dawkins 16 Tim Duncan* Andre Drummond Julian Hammond Mack Calvin Terry Cummings 17 John Stockton* George Gervin Gary Payton* Dave Robisch Toni Kukoč 18 Shaquille O'Neal* Paul Pierce Anderson Varejão Swen Nater Deron Williams 19 Oscar Robertson* Al Horford Dominique Wilkins* Buck Williams Hot Rod Williams 20 Stephen Curry Mark Price Kiki Vandeweghe Hersey Hawkins Tiny Archibald* 21 Karl Malone* Cedric Maxwell Elgin Baylor* John Havlicek Otis Thorpe 22 Larry Bird* Chris Bosh Wes Unseld* Clifford Ray Spencer Haywood* 23 Yao Ming* Joakim Noah Eddie Jones Greg Monroe Walter Davis 24 Ed Macauley* Russell Westbrook Derrick Favors Kevin Martin Corey Maggette 25 Artis Gilmore* Walt Bellamy Jerry Lucas* Marcus Camby Arthur Becker 26 Dirk Nowitzki Jimmy Jones Horace Grant Vlade Divac Tree Rollins 27 Bill Russell* Marques Johnson Peja Stojaković Byron Beck P.J. Brown 28 Julius Erving* Dwyane Wade Steve Mix Jack Twyman Wally Szczerbiak 29 Dolph Schayes* Rick Barry Terrell Brandon Grant Hill Nikola Vučević 30 Manu Ginóbili Steve Nash Shawn Kemp Happy Hairston Gilbert Arenas 31 Adrian Dantley* Kyrie Irving Ricky Pierce Sam Perkins Gus Williams 32 Jimmy Butler Don Nelson Scottie Pippen* Anthony Mason Kerry Kittles 33 Sidney Moncrief* Connie Hawkins Jack Sikma* Mel Daniels Bill Cartwright 34 Paul Arizin* Alonzo Mourning Roger Brown* Zydrunas Ilgauskas George McGinnis* 35 Sam Jones* Blake Griffin Calvin Natt Bob Cousy Alex English* 36 Harry Gallatin* Chet Walker Jim Eakins Doc Rivers Willie Wise 37 Kevin Garnett* Pau Gasol Jeff Foster Chris Mullin Allen Iverson* 38 Dan Issel* Amar'e Stoudemire Amir Johnson Rasheed Wallace Mario Elie 39 Giannis Antetokounmpo Vern Mikkelsen George Hill Steve Smith Danny Granger 40 Kevin McHale* Cliff Hagan Danilo Gallinari Nene Hilario Ty Lawson 41 Bailey Howell* Kobe Bryant Isaiah Thomas Tony Parker Jamaal Wilkes* 42 Bill Sharman* Larry Nance Carlos Boozer Steve Kerr Earl Monroe* 43 George Yardley* Zelmo Beaty Brent Barry Ben Wallace Anfernee Hardaway 44 Clyde Lovellette* Kevin Love Larry Jones Dave Cowens Brandon Bass 45 Kevin Johnson Dwight Howard Marc Gasol Billy Cunningham Taj Gibson 46 Hakeem Olajuwon* Clyde Drexler Maurice Cheeks* Serge Ibaka Ray Felix 47 Walt Frazier* Moses Malone Mehmet Okur Sam Cassell Cliff Levingston 48 Reggie Miller* DeAndre Jordan Marcin Gortat Dale Davis Dan Majerle 49 Chauncey Billups Bob Lanier Larry Costello David West Andre Iguodala 50 Damian Lillard Bobby Jones James Silas Tom Owens Fred Brown Win Share Snake Draft.xlsx (13.25 KB) |
|
|
|
Post by killybing on Dec 5, 2020 15:24:39 GMT
Set up a minimum and maximum win shares that can be in each class. problem solved. easy.
|
|
|
|
Post by eric on Dec 5, 2020 16:20:34 GMT
Here is the snake draft for the top 250 in win shares. Excel file attached thanks, great work. i think you caught some ABA guys in there but it still works as someone who isn't that bothered by draft class variation i don't think my opinion is as relevant here, but fwiw i'm not sure this really evens things out. of course we can never tell how players will develop but in 5.0 that 19 class was monstrous - oscar and horford were megastars, dominique was a star, buck williams was decent. or the 29 class, the worst guy there is rick barry and he was a borderline max player. or the 16 class, tim duncan is a top ten player, drummond and cummings were both stars, if the other two guys there had played in the nba at all who knows how good the class gets then there's the 5 class that is straight trash - although we apparently never had neil johnston who i would guess would have been decent at least, the other four are all below league average. the 32 class has butler who was a star but the others, woof like i said though, i always thought draft class variation was overstated anyway, so i'm interested to hear if other people think this exercise got us anywhere |
|
|
|
Post by eric on Dec 5, 2020 16:21:13 GMT
Set up a minimum and maximum win shares that can be in each class. problem solved. easy. i don't think class depth is the issue people have though as much as the tippy top level talent, and a class wide metric won't necessarily capture that |
|
|
|
Post by killybing on Dec 5, 2020 16:31:08 GMT
Set up a minimum and maximum win shares that can be in each class. problem solved. easy. i don't think class depth is the issue people have though as much as the tippy top level talent, and a class wide metric won't necessarily capture that
Ok then do it on a player-by-player basis.
Each class has to have at least 1 X winshare player, 1 Y winshare player, 1 Z winshare player and then the rest of the class has to add up to N Potential numbers could be 12, 10, 8 and 50 (for single season values)
|
|
|
|
Post by Logan on Dec 5, 2020 17:17:16 GMT
i'll say this much, none of the classes in druce's list are anywhere near as bad as the 2000 class for example. we miiiight even be able to expand it to top 300 win shares to get more classes out of it.
biggest issue i see with this system is how to add the upcoming irl classes to it. probably coulr just add them to the main pool though.
|
|
|
|
Post by Logan on Dec 5, 2020 17:18:08 GMT
also fully randomized classes are really cool and exciting imo. its something we've never done that isnt ridiculous.
|
|
|
|
Post by Druce on Dec 5, 2020 18:42:53 GMT
Here is the snake draft for the top 250 in win shares. Excel file attached thanks, great work. i think you caught some ABA guys in there but it still works as someone who isn't that bothered by draft class variation i don't think my opinion is as relevant here, but fwiw i'm not sure this really evens things out. of course we can never tell how players will develop but in 5.0 that 19 class was monstrous - oscar and horford were megastars, dominique was a star, buck williams was decent. or the 29 class, the worst guy there is rick barry and he was a borderline max player. or the 16 class, tim duncan is a top ten player, drummond and cummings were both stars, if the other two guys there had played in the nba at all who knows how good the class gets then there's the 5 class that is straight trash - although we apparently never had neil johnston who i would guess would have been decent at least, the other four are all below league average. the 32 class has butler who was a star but the others, woof like i said though, i always thought draft class variation was overstated anyway, so i'm interested to hear if other people think this exercise got us anywhere Part if it is how the profiles are written as well. Take the Neil Johnston guy for example, if he’s written as a stud then that increases the likelihood he ends up being good. Picking which qualities to highlight and which to downplay would make a significant difference in how good a particular draft class could be |
|
|
|
Post by eric on Dec 5, 2020 20:04:40 GMT
thanks, great work. i think you caught some ABA guys in there but it still works as someone who isn't that bothered by draft class variation i don't think my opinion is as relevant here, but fwiw i'm not sure this really evens things out. of course we can never tell how players will develop but in 5.0 that 19 class was monstrous - oscar and horford were megastars, dominique was a star, buck williams was decent. or the 29 class, the worst guy there is rick barry and he was a borderline max player. or the 16 class, tim duncan is a top ten player, drummond and cummings were both stars, if the other two guys there had played in the nba at all who knows how good the class gets then there's the 5 class that is straight trash - although we apparently never had neil johnston who i would guess would have been decent at least, the other four are all below league average. the 32 class has butler who was a star but the others, woof like i said though, i always thought draft class variation was overstated anyway, so i'm interested to hear if other people think this exercise got us anywhere Part if it is how the profiles are written as well. Take the Neil Johnston guy for example, if he’s written as a stud then that increases the likelihood he ends up being good. Picking which qualities to highlight and which to downplay would make a significant difference in how good a particular draft class could be sure but i mean he was a mega star irl so the odds are good he makes the 5 class at least a little better, since right now it's pure trash |
|
|
|
Post by eric on Dec 5, 2020 20:05:19 GMT
i don't think class depth is the issue people have though as much as the tippy top level talent, and a class wide metric won't necessarily capture that Ok then do it on a player-by-player basis.
Each class has to have at least 1 X winshare player, 1 Y winshare player, 1 Z winshare player and then the rest of the class has to add up to N Potential numbers could be 12, 10, 8 and 50 (for single season values)
it occurred to me that we should really compare the top 5 from each draft class we actually did to the spreads we're seeing here, if i get a chance later today i'll run those numbers and we'll also be able to see if we already satisfy those criteria |
|
|
|
Post by Druce on Dec 5, 2020 21:18:04 GMT
I will also say that while part of this is about balancing classes, a bigger portion for me personally is that I think it’d just be a really fun change
|
|
|
|
Post by eric on Dec 5, 2020 23:40:32 GMT
okay so here's where we end up
the randomized classes from druce have a standard deviation in top five WS/48 of .033
we have software draft pages going back to 2001 except for a few missing ones for a total of 43 classes so we're in the same sample size range, and those have a standard deviation of .021, or less spread between the good ones and bad ones. the only classes that didn't produce a player with a career average of 10+ WS per season (82*36 mp) were...
9 kenyon martin
8 bernard king
9 rajon rondo
and they had individual season highs of 13, 11, and 11 respectively
so i think given the vagaries of TC and GM, the way we've been doing it is already satisfying the constraints people are looking for in terms of class disparity, and doing so better than any other option so far suggested - the only way to guarantee a player will end up as a 12 WS type is to build them as a 16 WS type, and that's lebron level, and that's not something we should have on average in every class
.
i recognize that some people have said they want to do it the new way for its own sake, i just want to make clear that people should expect this to increase class disparity, not decrease it
|
|
|
|
Post by Logan on Dec 5, 2020 23:51:39 GMT
my thing here is i'm not looking at average or single season peak ws, i'm looking at career totals. disirregardless, druce is correct in saying that a player's quality is largely dependent upon how the profile is written. a lot of this would be giving up recognizable names to be our elite players rather than whatever an olden polynice is (yes i know he would be in this group but hey i still barely know who the guy is)
the big appeal for me is that this is a new approach we havent tried yet that isnt going to have a negative impact on the league.
|
|
|
|
Post by eric on Dec 6, 2020 0:12:52 GMT
my thing here is i'm not looking at average or single season peak ws, i'm looking at career totals. disirregardless, druce is correct in saying that a player's quality is largely dependent upon how the profile is written. a lot of this would be giving up recognizable names to be our elite players rather than whatever an olden polynice is (yes i know he would be in this group but hey i still barely know who the guy is) the big appeal for me is that this is a new approach we havent tried yet that isnt going to have a negative impact on the league. i think career totals isn't a helpful way to look at it because career length is so randomized for us, like wilt retiring at 33 put him at the same career ws level as dwight and dwight wasn't worthy to carry wilt's jock strap as far as negative impact my big concern is that we try it for a few years, the league decides they hate it, then what do we do for the classes that happened to have their great players taken already? like if we took MJ from 1984 and lebron from 2003 that's not a huge deal, there's plenty of top end talent left in those classes. but there are a lot of top heavy classes too, like if we take KG from 1995 that immediately becomes one of the worst classes unless the Dragan Tarlacs and Brent Barrys [sic] of the world pop, so it immediately becomes one of the worst classes again i'm aware that a few people have moved beyond the class imbalance issue in this thread, but there are still people who think that's really important, so i'm very leery of putting in a change that will in the best case exacerbate it and in the worst case dramatically exacerbate it my argument for wheel was never that there wasn't any class imbalance or that class imbalance didn't matter, just that there wasn't as much as people seemed to believe there was, and i think the numbers we've seen have borne that out. no one likes getting a #1 pick in a class that looks bad, but when we look at it on a long term league wide level there's legit talent in every draft class, and almost always legit star talent |
|
