Draft Value and the Shape of the Curve

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
I mean, that's obviously the hope for [Langford] given his skill set, but expecting any #14 guy to become an All-Star creator is a huge longshot.
This post made me think: Is the difference between #14 and #9 (Hayward 2010) that drastic? Paul George was selected #10 right after him, too.

We can all think of individual examples at just about any draft position, so let me do a little data for a couple of pick ranges across 30 drafts, going in 3s for sample size:

Count of all-stars / all-NBA, 1990-2019 drafts:
picks 5-7: 18
picks 8-10: 18
picks 11-13: 6
picks 14-16: 8
picks 17-19: 9
Entire 2nd Round: 19

picks 5-7: 18 (Steve Smith, T. Gugliotta, Juwon Howard, Garnett, Ray Allen, 'Toine, Vince Carter, Wally World, Rip Hamilton, D-Wade, Chris Kaman, Devin Harris, Luol Deng, Brandon Roy, K. Love, Steph Curry, D. Cousins, Lillard)
picks 8-10: 18 (Vin Baker, Eddie Jones, McGrady, Dirk, Pierce, Marion, Joe Johnson, Stoudemire, Caron Butler, Iguodala, Bynum, J. Noah, B. Lopez, DeRozan, Hayward, P. George, Kemba, Drummond)
picks 11-13: 6 (Tyrone Hill, Terrell Brandon, Dale Davis, Allan Houston, Kobe, Klay Thompson)
picks 14-16: 8 (Chris Gatling, Peja Stojakovic, Steve Nash, Al Jefferson, Kawhi, Vucevic, Giannis, Artest)
picks 17-19: 9 (T. Ratliff, J. O'Neal, J. Magloire, Z. Randolph, David West, D. Granger, R. Hibbert, J. Holiday, Jeff Teague)
Entire 2nd Round: 19 (Antonio Davis, Cedric Ceballos, N. Van Exel, Rashard Lewis, Ginobili, M. Redd, Arenas, Okur, Boozer, Mo Williams, Korver, Millsap, M. Gasol, D. Jordan, Dragic, IT4, Draymond, K. Middleton, Jokic)

I guess that's a pretty sharp dropoff from 10 to the teens, though obviously even with 90 data points per range, this is still very noisy. And not all all-stars are created equal, either. Let's do it by WS for certain ranges:

WS > 60
picks 5-7: 13
picks 8-10: 16
picks 11-13: 8
picks 14-16: 7
picks 17-19: 3
picks 20-22: 2
Entire 2nd Round: 11

WS > 40
picks 5-7: 27
picks 8-10: 28
picks 11-13: 16
picks 14-16: 12
picks 17-19: 14
picks 20-22: 10
Entire 2nd Round: 36

WS > 20
picks 5-7: 48
picks 8-10: 45
picks 11-13: 33
picks 14-16: 29
picks 17-19: 30
picks 20-22: 27
Entire 2nd Round: 83

So that largely backs up the notion of a very sharp dropoff after #10 in terms of premium value, though not in terms of rotational value (the WS > 20 as a rough proxy). I guess Langford really does have much lower odds than Hayward did.

Notes:
- In the 2001 draft, 4 all-stars were selected between #25-38: Gerald Wallace, Tony Parker, Gilbert Arenas and Mehmet Okur
- The 2nd round has had long periods of total unproductivity. From 1991-1997 you had exactly 1 all-star (Van Exel), you've got a couple runs of 0s (04-05, 09-10), and there haven't been any since Jokic in 2014, with a much-higher fraction of selections who never played a game in the NBA. But as the WS stats indicate, you can find a solid contributor (WS > 20)
- Obviously the jury is very much out on the most recent few drafts
 

lovegtm

Member
SoSH Member
Apr 30, 2013
11,996
This post made me think: Is the difference between #14 and #9 (Hayward 2010) that drastic? Paul George was selected #10 right after him, too.

We can all think of individual examples at just about any draft position, so let me do a little data for a couple of pick ranges across 30 drafts, going in 3s for sample size:

Count of all-stars / all-NBA, 1990-2019 drafts:
picks 5-7: 18
picks 8-10: 18
picks 11-13: 6
picks 14-16: 8
picks 17-19: 9
Entire 2nd Round: 19

picks 5-7: 18 (Steve Smith, T. Gugliotta, Juwon Howard, Garnett, Ray Allen, 'Toine, Vince Carter, Wally World, Rip Hamilton, D-Wade, Chris Kaman, Devin Harris, Luol Deng, Brandon Roy, K. Love, Steph Curry, D. Cousins, Lillard)
picks 8-10: 18 (Vin Baker, Eddie Jones, McGrady, Dirk, Pierce, Marion, Joe Johnson, Stoudemire, Caron Butler, Iguodala, Bynum, J. Noah, B. Lopez, DeRozan, Hayward, P. George, Kemba, Drummond)
picks 11-13: 6 (Tyrone Hill, Terrell Brandon, Dale Davis, Allan Houston, Kobe, Klay Thompson)
picks 14-16: 8 (Chris Gatling, Peja Stojakovic, Steve Nash, Al Jefferson, Kawhi, Vucevic, Giannis, Artest)
picks 17-19: 9 (T. Ratliff, J. O'Neal, J. Magloire, Z. Randolph, David West, D. Granger, R. Hibbert, J. Holiday, Jeff Teague)
Entire 2nd Round: 19 (Antonio Davis, Cedric Ceballos, N. Van Exel, Rashard Lewis, Ginobili, M. Redd, Arenas, Okur, Boozer, Mo Williams, Korver, Millsap, M. Gasol, D. Jordan, Dragic, IT4, Draymond, K. Middleton, Jokic)

I guess that's a pretty sharp dropoff from 10 to the teens, though obviously even with 90 data points per range, this is still very noisy. And not all all-stars are created equal, either. Let's do it by WS for certain ranges:

WS > 60
picks 5-7: 13
picks 8-10: 16
picks 11-13: 8
picks 14-16: 7
picks 17-19: 3
picks 20-22: 2
Entire 2nd Round: 11

WS > 40
picks 5-7: 27
picks 8-10: 28
picks 11-13: 16
picks 14-16: 12
picks 17-19: 14
picks 20-22: 10
Entire 2nd Round: 36

WS > 20
picks 5-7: 48
picks 8-10: 45
picks 11-13: 33
picks 14-16: 29
picks 17-19: 30
picks 20-22: 27
Entire 2nd Round: 83

So that largely backs up the notion of a very sharp dropoff after #10 in terms of premium value, though not in terms of rotational value (the WS > 20 as a rough proxy). I guess Langford really does have much lower odds than Hayward did.

Notes:
- In the 2001 draft, 4 all-stars were selected between #25-38: Gerald Wallace, Tony Parker, Gilbert Arenas and Mehmet Okur
- The 2nd round has had long periods of total unproductivity. From 1991-1997 you had exactly 1 all-star (Van Exel), you've got a couple runs of 0s (04-05, 09-10), and there haven't been any since Jokic in 2014, with a much-higher fraction of selections who never played a game in the NBA. But as the WS stats indicate, you can find a solid contributor (WS > 20)
- Obviously the jury is very much out on the most recent few drafts
The result isn’t as counter-intuitive when you realize that Hayward himself had long odds to beat. He was decent-ish his first 4 years, and even then the Jazz weren’t stoked about paying him and made him go get an offer sheet in RFA.

Life outside the top 5 is rough.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
The result isn’t as counter-intuitive when you realize that Hayward himself had long odds to beat. He was decent-ish his first 4 years, and even then the Jazz weren’t stoked about paying him and made him go get an offer sheet in RFA.

Life outside the top 5 is rough.
this data says that life in 8-10 is just as good as life in 5-7. Obviously neither of them will be as good as 1-4, but I was surprised by that result - I thought the downward slope of value would be much more uniform and exponential.
 

BillMuellerFanClub

Member
SoSH Member
Aug 1, 2006
1,388
Does the availability of minutes on non-tanking teams impact the accumulation of experience and WS over time? Not sure how much you can discern from this if you view the development of talent through the lens of integrating new players in a system with few opportunities. In other words, how are picks treated within teams that are routinely in contention and picking 15+ in terms of workload or development? Would be interesting to isolate the success of late-round picks when that pick was made by a team with a record worse than the original owner of the pick, or where picks selected had an opportunity to play significant minutes very early in their career.

I would call this the Siakam Argument; drafted at 27 and averaged 15 minutes played per game his rookie season on a winning team (.600+ in his tenure), and has increased each year since.
 

Big John

New Member
Dec 9, 2016
2,086
So is all-stardom or lack thereof the key factor that determines the effectiveness of a draft pick? How about win shares or PER or something else that measures a player's contribution to winning? There are far too many players who put up gaudy offensive numbers on losing teams and get selected for the ASG because someone from the franchise had to go for marketing purposes.
 

PC Drunken Friar

Member
SoSH Member
Sep 12, 2003
14,539
South Boston
this data says that life in 8-10 is just as good as life in 5-7. Obviously neither of them will be as good as 1-4, but I was surprised by that result - I thought the downward slope of value would be much more uniform and exponential.
Could this be because, usually there are no real surprises in the top 3 or 4 picks? Most people know who will go 1-4 in some order well before the draft. And then 5-10 is a little more cloudy? Like the GM of pick 6 might like his guy but GM of the 10th pick would take the player he got at 10 at 6 if he had the chance.
 

Devizier

Member
SoSH Member
Jul 3, 2000
19,463
Somewhere
Win shares might undersell the curve because teams with top picks are bad and have fewer “shares” available. I wonder if something truly generic like minutes played isn’t the best metric.
 

moondog80

heart is two sizes two small
SoSH Member
Sep 20, 2005
8,091
Even with a sample going back to 1980, there is a lot of noise in the data. The 9 pick does way better than the 8 pick, but simply switch Dirk Nowitzki and Larry Hughes in the 1998 draft and that difference is more than halved.

Is there any logical explanation why pick X+1 might be inherently more valuable than pick X? I guess maybe the first pick of the second round might fare better than the picks immediately in front of them over the long haul, because the team picking 29 may have made a promise to pick a certain guy and thus unable/unwilling to pick the guy who unexpectedly drops? But nobody makes a promise with the first pick of the second round, so that pick never has any encumbrance?
 

Devizier

Member
SoSH Member
Jul 3, 2000
19,463
Somewhere
Even with a sample going back to 1980, there is a lot of noise in the data. The 9 pick does way better than the 8 pick, but simply switch Dirk Nowitzki and Larry Hughes in the 1998 draft and that difference is more than halved.

Is there any logical explanation why pick X+1 might be inherently more valuable than pick X?
There isn't but you could probably apply an exponential decay curve and be done with it.
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,945
New York, NY
Win shares might undersell the curve because teams with top picks are bad and have fewer “shares” available. I wonder if something truly generic like minutes played isn’t the best metric.
Minutes are going to have this issue as well. Bad teams have more minutes up for grabs, and further are more incentivized to play high picks. I'd suggest using VORP instead of win shares if you want to stick with Basketball Reference data.

As others have suggested, I would also suggest using an exponential model here. That'll solve these issues with respect to pick X vs. X+1, and give you a smooth curve.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
So after the discussion in the Ainge-drafting thread turned towards "what counts as a win vs loss", I went and re-did the analysis that WBCD posted earlier, so I could try and answer the question of wins-and-losses, and how Ainge stacks up, a little more quantitatively.

Background on analysis design:
  • I chose the draft years 1989 through 2017, to exclude both too-recent ones as well as ones where there were enough fewer teams that relative pick position starts to become incomparable (1989 is when the league expanded to 27 teams, two seasons after it had only 23, and also the first year of only having 2 rounds). You could argue to start in 1985 with the lottery and salary cap, or 1983 with the emergence of international players, or 1977 with the jump from 18 to 22 teams. At some point, though, the effect of "having more data" is swamped by the fact that the league and draft were a very different place in those earlier years.
  • As a result, within the first 2 rounds of those drafts, I have 1442 players who spent any time at all playing in the NBA, out of 1682 total draft selections (86%). Players who never played in the league are assigned a performance value of 0 for purposes of evaluating the draft choice's quality.
  • For putting draftees into draft-order buckets, I used overall selection number, rather than round-plus-number or any adjustments, despite the difference in years of control or teams in the league. This means that (say) every 28th pick is bucketed together, regardless of whether that pick fell within the 1st or 2nd round based on how many teams were in the league. I felt this was fairer because each selection made was based on the same steady depletion of a draft's talent, and despite the talent depth and reach of the NBA expanding over the years, this kept the decision-making by GMs more apples-to-apples.
  • I stuck to WS as a value metric (rather than VORP as I'd have preferred, per bowiac), because that's what Bk-Ref Play Index's search was willing to return in search results. I could use minutes played, or any combination of the stats it gives me, if people agree that that would be better.
  • The study WBCD posted included all-star selections as a secondary value metric, but I felt that was too subjective. Frankly, all-NBA selections would be a better choice, as the award has existed in its 3-team format since 1988, and there's no fan component to the voting. We could also include all-defensive selections. If there's an easy dataset available for that, I might well give it a shot.
  • Some mild data cleaning was required to update franchises who have moved or names have changed. NJN => BRK, VAN => MEM, the whole NO / Charlotte thing, etc.
  • The lowest career WS right now for players in those 29 drafts is Josh Jackson, at -2.4 WS. 5th-from-bottom is Cryin' Adam Morrison at -1.4.

What does it mean for a pick to be a win or a loss?
I think there's one easy and fair approach: let's look at the average WS curve by pick, smoothed out to a logarithmic curve, and then rate every pick a "win" or a "loss" based on whether the pick ends up exceeding or falling short of the WS benchmark for that selection number. Since that smoothed WS average is a stand-in for "expectations" at that pick number, this is just a statement of whether or not any given pick exceeded or fell short of expectations.

Discussion of problems with this, and counterarguments / mitigations:
There are, of course, a couple obvious objections to this:

1) It depends on players piling up a career's worth of WS, and players whose careers aren't finished yet are getting undersold on this count. But then again, we just have one bar to clear, a binary judgment on the pick, so we only really need to be directionally correct. So I've corrected for this by pro-rating: I've calculated the average career length (in years) by pick number from among all draftees for drafts through 2010 only, smoothed that out a bit (logarithmic regression), and for any player with fewer years of service than the average of their pick, I've added in an estimated number of WS-to-be, by using their WS/48 for their career-to-date and figuring out how many WS they would be expected to end up with on that current pace, if they were to have an average-length career.

So for example, Anthony Davis as a #1 pick is expected to have an average career length of 12. He's played 7 seasons. For the 5 "missing" seasons, given his minutes averages, I calculate we're missing ~12,300 minutes, which at his 0.218 WS/48 means he's short 56.2 WS. Adding that to his 78.8 of actual WS accumulated, his "projected" total is 135.0, which flips him from a push (as a #1 pick) to a decisive win.

Now, this is even less fair to very recent picks, but that's partly addressed by ending with the 2017 draft. And sure, WS/48 isn't perfect either, as young players can expect to grow in value as they reach their primes. But it helps correct a decent fraction of the injustice here, and particularly for those who are already showing promise of being stars.

2) A drafting team isn't entitled to all the value created by a player in their career. Really, you'd prefer to do something like what PFR does in their draft tables, having a "Draft AV" that shows AV accumulated while the player remained on the drafting team. But given the frequency with which teams make player-for-player trades in the NBA (certainly relative to the NFL), it's pretty fair to assume that on average they recapture any value traded away with equal value coming in. If possible, I'd focus on just the first 7 years of control, as that's how long you can make someone stay on your team who might otherwise leave. But I can't easily adjust for that data (and even if I could, , and so I'll deal. There aren't enough really-late-bloomers in the league to really make a pick be highly undervalued this way, to the point of flipping a loss (for the drafting team) into a win (by a player who ends up having a great career but mostly for teams other than the drafting team).

3) "Age at drafting time" is another confounding variable. All else equal, older players reach their peak sooner and thus return more value to their drafting team. But it's also true that "amount of experience playing in the NBA" is another dimension to development curve, and neither tells the full story. And furthermore, a lot of the highest, premium talent that's entered the league has known they were ready for the NBA at an earlier age. I've had to assume that these effects all cancel each other out, but I can at least bucket things by draft-age to see if there are major trends in "win rate".

4) There's skew to the distribution of value even around a single pick number, with a lot of extra "weight" to the average coming from the few biggest winners in each bucket, so overall there will be more Losers than Winners. In fact, with just a plain ol' average, only 31% of picks are judged a Win, and some like Jaylen Brown are judged a Loss. As a result, the benchmark / threshold rate I've used is actually 75% of the smoothed average. This yields a 48% Win Rate for 1st-rounders (1-27), and 28% Win Rate for second-rounders. Part of the difference is that to be rated a Win, you still have to have a WS > 0, and for a steadily-increasing fraction of second rounders, they never play in the league at all. If someone wants to argue that I should instead use a median WS value (for each selection number) instead of mean, and then smooth that, so we get roughly 50/50 W-L, I'll listen.

5) This obviously ignores whether the team the player was on "used him correctly". i.e., some players may be very valuable and end up having great careers, but as a result of coaching, roster composition, their own immaturity, or many other reasons, they never put it together while on the drafting team - and perhaps never would have. Chauncey Billups, to take a familiar example. And that'd be more of an issue if we were trying to evaluate how good a player was, rather than whether a selection was a W or L for the drafting GM. But since that GM had to integrate the pick into their team and try to get on-court value out of them, I think it's fair to just judge by what their production was like for the team that did play them. This intermingles with Problem #2 somewhat, and in theory you'd really like to isolate how the player did just for the drafting team. But to do that, you'd probably also need some estimate of value returned to a team when trading away a player, and that's just a rabbit hole I lack the time to go down.

6) This data assigns a player to a team that actually cast the pick, even if they were "drafting for someone else", i.e. there was a trade immediately before or after. So Trae Young is counted as DAL here and Doncic as ATL. edit: this is now corrected thanks to scottno below.

All that accounted for, here's what this looks like:

Career Length

28087

This is the average number of seasons, used to gross-up the WS of younger players so their WS totals are pro-rated to that of an average career for their pick #. I smoothed it to be a 12-year max threshold for the top 5 picks, then this regression formula for the rest. Note both the line (prediction / smoothing) being below the actual number of seasons for all of the 20s, and also the sharp dropoff from the high 20s to the low 30s - it seems that there may be an effect where players are getting some end-of-career mileage out of being a "first round pick" rather than the steady decline that you'd expect. Also, a linear regression fit better, but didn't fit with an intuitive sense of the data.

WS "Win Benchmark"

28088

Here's the average Career WS, once it is (1) grossed-up to account for young players' "missing" seasons, (2) smoothed down to a log regression (R^2 of 92%!), and then (3) lowered to a 75% Threshold benchmark. Which lets us then evaluate each player as a Win or Loss. If we had a very high N, you'd expect this to be much closer to a smooth and steady line. As it is, it's not all that unsmooth.


Win Rate

28089

There's the chart by pick number, going from >60% win rate for the very top picks, down to the 40s for most of the 1st round, hovering around 30% for the first half of the second round, and then down to ~20% for the back half of the second round. The orange line is a 5-point moving average (-2 to +2 picks).

It's fair to point out that this merely indicates whether the player cleared the threshold or not, and at what frequency. It makes no distinction between someone barely clearing it (e.g. Porzingis at #4 or Tyler Zeller at #17) or someone who blew it out of the water (your Manu Ginobili or Isaiah Thomas). We can look at a pick or group of picks and ask how much they collectively exceeded or fell short of expectations.


OK all that's great, but how has Ainge done?

Well there's the question we care about, right? Here's Ainge's full 42-pick drafting history, sorted by "amount of win or loss".

(edit 1/19: replaced the data with a list corrected for draft-day trades)
28114

The numeric columns after their name are, in order: Actual WS, "Missing" WS as calculated by grossing up to average career length, Total WS (actual + missing if any), Benchmark WS, and then Over-Under (Total minus Benchmark).

So he's hit on his 3 top-10 picks, missed big on a few of his mid-1st rounders, and acquired Rondo, Perk and Olynyk in draft-day trades for a +45 WS win (relative to pick #21), +17 and +13.5, respectively. No huge misses with top picks, no huge wins with really low picks. You can surely quibble with a lot of how these numbers come out - getting any non-zero value out of a second rounder seems like a win, frankly.

How does that stack up against other franchises? Well, only including picks from 2003-present:

28115

That's Danny right up there at 9th out of 30. Now of course, the top of the rankings are dominated by a handful of picks swinging things:
- NOP: Chris Paul, David West, Anthony Davis
- OKC: Durant, Harden, Westbrook, Ibaka
(and at the other end, teams taking some big Ls...)
- PHO: Josh Jackson at #4 (-51), Dragan Bender at #4 (-35), and Robin Lopez as their biggest W (+25).
- DAL: Dennis Smith at #9 (-31 so far), a bunch of small losses, and Josh Howard the only real W (+28).

Ainge certainly has a LOT of picks: counting draft-day acquisitions, I'm pretty sure he's had the most number of draft choices made in this span. And in terms of excess value generated, he's at 9th out of 30 on a total value basis (13th on a per-pick basis). By this measure, Danny has been above average, though he hasn't, of course, had a run like New Orleans or OKC.

28116

OKC: That same crew, plus Gary Payton, Rashard Lewis at #32, Shawn Kemp...
SAS: Duncan, Parker at #28, Ginobili at #57, etc.
LAL: Kobe, Eddie Jones, Vlade, plus I have to count Marc Gasol taken at #48
(and the big Ls)
LAC: Olowokandi at #1, Darius Miles at #3, Bo Kimble at #8, and so on
 
Last edited:

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
Maybe it would be way too hard to do, but that above chart seems completely useless if it's going by who a team technically drafted, and not who they ended up with. Ainge didn't really "draft" bell and jones in 2003 for example, he knew on draft day he was getting banks and perkins, just changing that alone goes from 9.7 WS to 34.5 in one season.

Also as you said it leaves out his 2nd most valuable pick because he technically bought the pick, even though he made the selection. He's also hit on 3/3 top 10 picks, you can't count green for him or foye against him.
 

wade boggs chicken dinner

Member
SoSH Member
Mar 26, 2005
30,482
So after the discussion in the Ainge-drafting thread turned towards "what counts as a win vs loss", I went and re-did the analysis that WBCD posted earlier, so I could try and answer the question of wins-and-losses, and how Ainge stacks up, a little more quantitatively.

Background on analysis design:
  • I chose the draft years 1989 through 2017, to exclude both too-recent ones as well as ones where there were enough fewer teams that relative pick position starts to become incomparable (1989 is when the league expanded to 27 teams, two seasons after it had only 23, and also the first year of only having 2 rounds). You could argue to start in 1985 with the lottery and salary cap, or 1983 with the emergence of international players, or 1977 with the jump from 18 to 22 teams. At some point, though, the effect of "having more data" is swamped by the fact that the league and draft were a very different place in those earlier years.
  • As a result, within the first 2 rounds of those drafts, I have 1442 players who spent any time at all playing in the NBA, out of 1682 total draft selections (86%). Players who never played in the league are assigned a performance value of 0 for purposes of evaluating the draft choice's quality.
  • For putting draftees into draft-order buckets, I used overall selection number, rather than round-plus-number or any adjustments, despite the difference in years of control or teams in the league. This means that (say) every 28th pick is bucketed together, regardless of whether that pick fell within the 1st or 2nd round based on how many teams were in the league. I felt this was fairer because each selection made was based on the same steady depletion of a draft's talent, and despite the talent depth and reach of the NBA expanding over the years, this kept the decision-making by GMs more apples-to-apples.
  • I stuck to WS as a value metric (rather than VORP as I'd have preferred, per bowiac), because that's what Bk-Ref Play Index's search was willing to return in search results. I could use minutes played, or any combination of the stats it gives me, if people agree that that would be better.
  • The study WBCD posted included all-star selections as a secondary value metric, but I felt that was too subjective. Frankly, all-NBA selections would be a better choice, as the award has existed in its 3-team format since 1988, and there's no fan component to the voting. We could also include all-defensive selections. If there's an easy dataset available for that, I might well give it a shot.
  • Some mild data cleaning was required to update franchises who have moved or names have changed. NJN => BRK, VAN => MEM, the whole NO / Charlotte thing, etc.
  • The lowest career WS right now for players in those 29 drafts is Josh Jackson, at -2.4 WS. 5th-from-bottom is Cryin' Adam Morrison at -1.4.

What does it mean for a pick to be a win or a loss?
I think there's one easy and fair approach: let's look at the average WS curve by pick, smoothed out to a logarithmic curve, and then rate every pick a "win" or a "loss" based on whether the pick ends up exceeding or falling short of the WS benchmark for that selection number. Since that smoothed WS average is a stand-in for "expectations" at that pick number, this is just a statement of whether or not any given pick exceeded or fell short of expectations.

Discussion of problems with this, and counterarguments / mitigations:
There are, of course, a couple obvious objections to this:

1) It depends on players piling up a career's worth of WS, and players whose careers aren't finished yet are getting undersold on this count. But then again, we just have one bar to clear, a binary judgment on the pick, so we only really need to be directionally correct. So I've corrected for this by pro-rating: I've calculated the average career length (in years) by pick number from among all draftees for drafts through 2010 only, smoothed that out a bit (logarithmic regression), and for any player with fewer years of service than the average of their pick, I've added in an estimated number of WS-to-be, by using their WS/48 for their career-to-date and figuring out how many WS they would be expected to end up with on that current pace, if they were to have an average-length career.

So for example, Anthony Davis as a #1 pick is expected to have an average career length of 12. He's played 7 seasons. For the 5 "missing" seasons, given his minutes averages, I calculate we're missing ~12,300 minutes, which at his 0.218 WS/48 means he's short 56.2 WS. Adding that to his 78.8 of actual WS accumulated, his "projected" total is 135.0, which flips him from a push (as a #1 pick) to a decisive win.

Now, this is even less fair to very recent picks, but that's partly addressed by ending with the 2017 draft. And sure, WS/48 isn't perfect either, as young players can expect to grow in value as they reach their primes. But it helps correct a decent fraction of the injustice here, and particularly for those who are already showing promise of being stars.

2) A drafting team isn't entitled to all the value created by a player in their career. Really, you'd prefer to do something like what PFR does in their draft tables, having a "Draft AV" that shows AV accumulated while the player remained on the drafting team. But given the frequency with which teams make player-for-player trades in the NBA (certainly relative to the NFL), it's pretty fair to assume that on average they recapture any value traded away with equal value coming in. If possible, I'd focus on just the first 7 years of control, as that's how long you can make someone stay on your team who might otherwise leave. But I can't easily adjust for that data (and even if I could, , and so I'll deal. There aren't enough really-late-bloomers in the league to really make a pick be highly undervalued this way, to the point of flipping a loss (for the drafting team) into a win (by a player who ends up having a great career but mostly for teams other than the drafting team).

3) "Age at drafting time" is another confounding variable. All else equal, older players reach their peak sooner and thus return more value to their drafting team. But it's also true that "amount of experience playing in the NBA" is another dimension to development curve, and neither tells the full story. And furthermore, a lot of the highest, premium talent that's entered the league has known they were ready for the NBA at an earlier age. I've had to assume that these effects all cancel each other out, but I can at least bucket things by draft-age to see if there are major trends in "win rate".

4) There's skew to the distribution of value even around a single pick number, with a lot of extra "weight" to the average coming from the few biggest winners in each bucket, so overall there will be more Losers than Winners. In fact, with just a plain ol' average, only 31% of picks are judged a Win, and some like Jaylen Brown are judged a Loss. As a result, the benchmark / threshold rate I've used is actually 75% of the smoothed average. This yields a 48% Win Rate for 1st-rounders (1-27), and 28% Win Rate for second-rounders. Part of the difference is that to be rated a Win, you still have to have a WS > 0, and for a steadily-increasing fraction of second rounders, they never play in the league at all. If someone wants to argue that I should instead use a median WS value (for each selection number) instead of mean, and then smooth that, so we get roughly 50/50 W-L, I'll listen.

5) This obviously ignores whether the team the player was on "used him correctly". i.e., some players may be very valuable and end up having great careers, but as a result of coaching, roster composition, their own immaturity, or many other reasons, they never put it together while on the drafting team - and perhaps never would have. Chauncey Billups, to take a familiar example. And that'd be more of an issue if we were trying to evaluate how good a player was, rather than whether a selection was a W or L for the drafting GM. But since that GM had to integrate the pick into their team and try to get on-court value out of them, I think it's fair to just judge by what their production was like for the team that did play them. This intermingles with Problem #2 somewhat, and in theory you'd really like to isolate how the player did just for the drafting team. But to do that, you'd probably also need some estimate of value returned to a team when trading away a player, and that's just a rabbit hole I lack the time to go down.

6) This data assigns a player to a team that actually cast the pick, even if they were "drafting for someone else", i.e. there was a trade immediately before or after. So Trae Young is counted as DAL here and Doncic as ATL. If I had a list, I could easily make the adjustments here to give drafting credit to the team who actually first put him on the court. Trae and Luka are excluded of course, as they're 2018 Draft and we end in 2017. And we can correct for that on a case-by-case basis when looking at a specific GM's history, or for specific picks. I've gone and corrected it for the 100 most-valuable players, just so it doesn't count Kobe as a Charlotte Hornet New Orleans Pelican, or Dirk as a Milwaukee Buck. But we couldn't draw conclusions about "which teams draft the best" without correcting this across the board.

All that accounted for, here's what this looks like:

Career Length

View attachment 28087

This is the average number of seasons, used to gross-up the WS of younger players so their WS totals are pro-rated to that of an average career for their pick #. I smoothed it to be a 12-year max threshold for the top 5 picks, then this regression formula for the rest. Note both the line (prediction / smoothing) being below the actual number of seasons for all of the 20s, and also the sharp dropoff from the high 20s to the low 30s - it seems that there may be an effect where players are getting some end-of-career mileage out of being a "first round pick" rather than the steady decline that you'd expect. Also, a linear regression fit better, but didn't fit with an intuitive sense of the data.

WS "Win Benchmark"

View attachment 28088

Here's the average Career WS, once it is (1) grossed-up to account for young players' "missing" seasons, (2) smoothed down to a log regression (R^2 of 92%!), and then (3) lowered to a 75% Threshold benchmark. Which lets us then evaluate each player as a Win or Loss. If we had a very high N, you'd expect this to be much closer to a smooth and steady line. As it is, it's not all that unsmooth.


Win Rate

View attachment 28089

There's the chart by pick number, going from >60% win rate for the very top picks, down to the 40s for most of the 1st round, hovering around 30% for the first half of the second round, and then down to ~20% for the back half of the second round. The orange line is a 5-point moving average (-2 to +2 picks).

It's fair to point out that this merely indicates whether the player cleared the threshold or not, and at what frequency. It makes no distinction between someone barely clearing it (e.g. Porzingis at #4 or Tyler Zeller at #17) or someone who blew it out of the water (your Manu Ginobili or Isaiah Thomas). We can look at a pick or group of picks and ask how much they collectively exceeded or fell short of expectations.

OK all that's great but how has Ainge done?

Well there's the question we care about, right? Here's Ainge's full 43-pick drafting history, sorted by "amount of win or loss".

View attachment 28090

The numeric columns after their name are, in order: Actual WS, "Missing" WS as calculated by grossing up to average career length, Total WS (actual + missing if any), Benchmark WS, and then Over-Under (Total minus Benchmark).

So he's hit on 4 out of 5 of his top-10 picks, missed big on a few of his mid-1st rounders, and acquired Rondo in a draft-day trade (not listed) for a +45 WS win relative to pick #21. No huge misses with top picks, no huge wins with really low picks. You can surely quibble with a lot of how these numbers come out - getting any non-zero value out of a second rounder seems like a win, frankly.

How does that stack up against other franchises? Well, so there's the caveat again that But having done that, here's the rankings, only including picks from 2003-present:

View attachment 28097

That's Danny right up there at 19th out of 30. Now of course, the top of the rankings are dominated by a handful of picks swinging things:
- NOP: Chris Paul, David West, Anthony Davis
- OKC: Durant, Harden, Westbrook, Ibaka
- DEN: Carmelo, Gobert, Jokic at #41, Jameer Nelson?!
- SAS: Kawhi, George Hill, Dragic at #45
(and at the other end, teams taking some big Ls...)
- PHO: Josh Jackson at #4 (-51), Dragan Bender at #4 (-35), and Robin Lopez as their biggest W (+25).
- DAL: Dennis Smith at #9 (-31 so far), a bunch of small losses, and Josh Howard the only real W (+28).

Ainge certainly has a LOT of picks: counting draft-day acquisitions for Rondo and Olynyk, I'm pretty sure he's had the most number of draft choices made in this span. But in terms of excess value generated, he's at 19th out of 30 on a per-pick basis, and still only 17th on an overall outperformance basis. By this measure, Danny has been average.

View attachment 28098

SAS: Duncan, Parker at #28, Ginobili at #57, etc.
OKC: That same crew, plus Gary Payton, Rashard Lewis at #32, Shawn Kemp...
LAL: Kobe, Eddie Jones, Vlade, plus I have to count Marc Gasol taken at #48
(and the big Ls)
LAC: Olowokandi at #1, Darius Miles at #3, Bo Kimble at #8, and so on
Wow, great info. Really appreciate reading this, thanks.

If it's not too much trouble, could you please create separate win totals for 1st round picks versus second round picks. I know that's arbitrary but 2nd round picks seem to me to be more akin to guessing that I'm interested to see "win" rate fir 1st round picks.

Thanks again for the time you spent on this.
 

Imbricus

Member
SoSH Member
Jan 26, 2017
4,810
How does that stack up against other franchises? Well, so there's the caveat again that But having done that, here's the rankings, only including picks from 2003-present:
Just seconding, this is great (hope you're not heading to divorce court or whatever after doing all this work). On quick inspection, it passes the eye test, as we all know by now Phoenix has had terrible drafts, after getting high picks, and the Spurs and Raptors (and 76ers) seem to draft well. Some thoughts I had:

* I wonder (following up on your note above on NOP, OKC, etc.) to what extent the top 5 and bottom 5 teams here are influenced by the fact they've been poor teams, and thus had a lot of high picks, which tend to be either a big success or a big flop? If so, one would tend to discount their placement (on either the high or low end) as being the result of smart drafting, and attribute it more to luck. It would also make the Spurs' ranking more impressive, as over that span the worst record they ever had was 50 wins, so they were hitting with low picks.

* I wonder how volatile the ranking would be if separated out by decade since, say, 1970. As in, do some teams consistently draft well, or if you look at the big picture, is there a moderately high degree of luck involved? I suppose separating out by decade doesn't really answer this, because you don't have the same front office for 50 years, but maybe some organizations have built a long-term culture around effective drafting?

Anyway, great stuff. I've always thought that NBA teams should pour money into getting the draft right, because it's such an obvious place to get a winning edge for a reasonable investment.
 

HomeRunBaker

bet squelcher
SoSH Member
Jan 15, 2004
30,096
Maybe it would be way too hard to do, but that above chart seems completely useless if it's going by who a team technically drafted, and not who they ended up with. Ainge didn't really "draft" bell and jones in 2003 for example, he knew on draft day he was getting banks and perkins, just changing that alone goes from 9.7 WS to 34.5 in one season.

Also as you said it leaves out his 2nd most valuable pick because he technically bought the pick, even though he made the selection. He's also hit on 3/3 top 10 picks, you can't count green for him or foye against him.
Right idea, wrong execution? It’s a shame those resources were put into this chart with incorrect data as it excludes Rondo, Perkins, Marcus Banks and Glen Davis as Celtics draft picks while including Troy Bell and Randy Foye.
 

RetractableRoof

tolerates intolerance
Lifetime Member
SoSH Member
Dec 1, 2003
3,836
Quincy, MA
I'm not sure who (and apologize for not knowing), but someone here on MBPC had done another bit of work around the draft and had done something interesting. They had broken the draft up into chunks of 3, when doing their analysis. (Treating picks 1, 2,3 as on group, 4,5,6 as the next) in order to smooth out the data a bit. Do you think this would add any value to what you've done here?

Is it possible the person presenting chunked data was you?

There is a lot of work here, thanks for sharing your effort with us. I do agree with scottyno as well, it doesn't really matter who they drafted if they flipped the player(s) before a game was even played. Almost like the 'net' of the draft versus the 'gross' of the draft.

Wow, this looks like a ton of work...
 

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
Right idea, wrong execution? It’s a shame those resources were put into this chart with incorrect data as it excludes Rondo, Perkins, Marcus Banks and Glen Davis as Celtics draft picks while including Troy Bell and Randy Foye.
Yeah, I love the idea in theory, I'm not sure if it's perfect (any metric that considers bradley a "loss" for the celtics seems wrong for example), but it should give a decent idea of how GMs actually do.

The biggest possible issue is probably going about how to determine how much credit a GM actually gets. Etwaun Moore for example is having a really solid career for a #55 pick, an easy "win" with the above chart for the Cs, but he basically never played with the celtics and was dumped after a year, so I'm not sure if Ainge deserves credit for that or not. Maybe it could only give credit for what the guy did on the team that drafted him, adjusted for years on the team, but then you lose a lot of a guy like Jefferson's value, who was mostly the centerpiece of the KG trade based on potential.
 

benhogan

Granite Truther
SoSH Member
Nov 2, 2007
20,111
Santa Monica
The biggest possible issue is probably going about how to determine how much credit a GM actually gets. Etwaun Moore for example is having a really solid career for a #55 pick, an easy "win" with the above chart for the Cs, but he basically never played with the celtics and was dumped after a year, so I'm not sure if Ainge deserves credit for that or not. Maybe it could only give credit for what the guy did on the team that drafted him, adjusted for years on the team, but then you lose a lot of a guy like Jefferson's value, who was mostly the centerpiece of the KG trade based on potential.
Then you start measuring different aspects of what a GM/team does, other than DRAFTING. You're adding in DEVELOPMENT (Moore) and then TRADING (Jefferson). The ultimate trifecta is if a team drafts well, develop the prospect and then the player is good for the team or used for trade bait.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
I'm not sure who (and apologize for not knowing), but someone here on MBPC had done another bit of work around the draft and had done something interesting. They had broken the draft up into chunks of 3, when doing their analysis. (Treating picks 1, 2,3 as on group, 4,5,6 as the next) in order to smooth out the data a bit. Do you think this would add any value to what you've done here?

Is it possible the person presenting chunked data was you?
Uh... so, like, in the first post of this very thread? :D The analysis I did yesterday was more of a followup to that, to try and answer this value question more cohesively so we had a common basis to discuss Ainge in the other thread.

I definitely think there's value in smoothing here (in fact, I did it twice for these purposes), because there's just not enough data to take each pick on its own and act like the difference between N and N+1 or N-1 is meaningful. So for that win-rate trendline I did a 5-pick moving average (which is cleaner than the buckets-of-3 approach, statistically), and for setting the baselines here I did some logarithmic regression (fitting the curve, because the individual points vary but there's no reason that, for example, Pick #24 (average WS: 23.1) should have a higher standard to clear than the 16th pick (average WS: 19.1). And it turns out the curve fits very nicely indeed, despite the statistical noise - I was super stoked to see an R^2 of 92.6%.

I do agree with scottyno as well, it doesn't really matter who they drafted if they flipped the player(s) before a game was even played. Almost like the 'net' of the draft versus the 'gross' of the draft.
Yeah, when attributing a player's eventual on-court value to a franchise, I thought the following approach was fair:
  • If team X drafted them and then they played their rookie season on team X, their whole career's value is attributed to Team X's drafting skill, because they were able to use and, if desired, dispose of that asset as they saw fit.
  • If team X traded the pick used to select them as a draft-day trade, but due to NBA rules had to formally make the selection on behalf of team Y and then trade him to team Y, I gave the credit for that career to Team Y. Because it was Team Y who was taking the risk on an unknown asset at that point, and their drafting that led them to believe the player would be more valuable than what they gave up. I was able to go implement this for the top 100-or-so picks before running out of steam, but the franchise rankings would be more accurate if I had the time to do the top 300-400 or so (before you really get down to the 'who cares' level of career impact).
  • For the hybrid case, where Team X drafts them, they play in the G-league a bit or ride the bench for a while, and then are traded to Team Y before they really have a chance to prove themselves one way or the other (the Marc Gasol case), I still allocated them to Team X, because they spent some time getting evaluated by coaches in practice and games as a part of the organization post draft. So if they decided the trade was worthwhile at that point, that's them harvesting their estimation of value. But it's a fuzzy boundary, and you could argue differing attribution on a case-by-case basis.

I'm trying to use that definition to define the "net of the draft", as you put it. If they trade the pick day-of-draft, I think it's the acquiring team - the one whose organization the player joins - that it's fair to judge on the question of how well they scouted and used that pick.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
If it's not too much trouble, could you please create separate win totals for 1st round picks versus second round picks. I know that's arbitrary but 2nd round picks seem to me to be more akin to guessing that I'm interested to see "win" rate fir 1st round picks.
Sure, that's relatively easy with Excel. And like I mentioned in a footnote, the Win Rate for 1st-rounders is closer to 50-50 while the Win Rate for 2nd-rounders is in the 20s-to-30s (when they win, they can win big, but the odds are not nearly as good).

28117

For first-rounders only, 1989-2017, Boston is ranked 5th on an overall-value basis, with a 51% win rate (slightly above-average), and on a per-pick basis is ranked 8th (he's had a LOT more picks). The rankings don't change a ton, though, from the 1st-and-2nd-round totals.

Filtering down even further to "only 1st rounders in the Ainge Era" - and, holy Small Sample Size Alert, batman - we find that Ainge's 21 first-rounders (in 15 years) rank him 6th on a total-value basis, 12th on a per-pick basis. OKC, NOP, CHI and SAS still top the charts, but SAS falls from 1st because Ginobili and Dragic are now excluded.

(also, footnote, if anyone wants the excel model to tinker with, let me know, I'll throw it on a webserver somewhere, nothing I did was all that deep).
 
Last edited:

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
  • If team X traded the pick used to select them as a draft-day trade, but due to NBA rules had to formally make the selection on behalf of team Y and then trade him to team Y, I gave the credit for that career to Team Y. Because it was Team Y who was taking the risk on an unknown asset at that point, and their drafting that led them to believe the player would be more valuable than what they gave up. I was able to go implement this for the top 100-or-so picks before running out of steam, but the franchise rankings would be more accurate if I had the time to do the top 300-400 or so (before you really get down to the 'who cares' level of career impact).
So guys like Kawhi, and Gobert are with their right teams, but not Rondo because he fell outside the top 100? Better than nothing, but it still seems like it's missing a big part of the picture. Just eyeballing the Celtics list, it looks like they drafted 1 winner for another team and 7 losers, before we even get into who others drafted for them, which I imagine is massively positive.

Not sure if it would help make things easier, but if you search on basketball reference by team draft history it lists next to the player if they were dealt in a draft day trade, like boston here:

https://www.basketball-reference.com/teams/BOS/draft.html
I think this is newly added, but maybe I'd just somehow never seen it before.
 

RetractableRoof

tolerates intolerance
Lifetime Member
SoSH Member
Dec 1, 2003
3,836
Quincy, MA
Uh... so, like, in the first post of this very thread? :D The analysis I did yesterday was more of a followup to that, to try and answer this value question more cohesively so we had a common basis to discuss Ainge in the other thread.
Damn, can you identify the person who isn't sleeping? :D Sorry about that, but I did guess it was your effort :)

Yeah, when attributing a player's eventual on-court value to a franchise, I thought the following approach was fair:
  • If team X drafted them and then they played their rookie season on team X, their whole career's value is attributed to Team X's drafting skill, because they were able to use and, if desired, dispose of that asset as they saw fit.
  • If team X traded the pick used to select them as a draft-day trade, but due to NBA rules had to formally make the selection on behalf of team Y and then trade him to team Y, I gave the credit for that career to Team Y. Because it was Team Y who was taking the risk on an unknown asset at that point, and their drafting that led them to believe the player would be more valuable than what they gave up. I was able to go implement this for the top 100-or-so picks before running out of steam, but the franchise rankings would be more accurate if I had the time to do the top 300-400 or so (before you really get down to the 'who cares' level of career impact).
  • For the hybrid case, where Team X drafts them, they play in the G-league a bit or ride the bench for a while, and then are traded to Team Y before they really have a chance to prove themselves one way or the other (the Marc Gasol case), I still allocated them to Team X, because they spent some time getting evaluated by coaches in practice and games as a part of the organization post draft. So if they decided the trade was worthwhile at that point, that's them harvesting their estimation of value. But it's a fuzzy boundary, and you could argue differing attribution on a case-by-case basis.
I'm trying to use that definition to define the "net of the draft", as you put it. If they trade the pick day-of-draft, I think it's the acquiring team - the one whose organization the player joins - that it's fair to judge on the question of how well they scouted and used that pick.
Yep, I was right - that's a metric ton of work. I think your rules of attribution are reasonable. Is there an updated list of how far you did take the Celtics drafts (via the top 100?) Or IS the 43 you list the after you got through the top 100? Because if that is the AFTER there are some pretty big swings in there for the Cs that aren't accounted for (as scottyno pointed out, Marcus Banks is a larger loss for Ainge, and Perkins is a larger gain I think). In fact if I'm understanding your numbers correctly, just swapping Perkins for Bell results in a 27+ point swing for Ainge which jumps him to 14 on WS Over/Under list. I'm guessing that's not the only swing in the data for Ainge either given his prolific trading.

WBCD made the comment about the 1st round having more value to him, and in that context - I think the data adjustment has to be completed through the 1st round for these traded 'credits' or 'net' draft adjustments in order to get a more apples to apples view. Perkins in Round 1 for Bell (as well as Banks) as an example is such a wild swing given the small data sizes.

Finally, I think one of Ainge's strengths is his ability to extract draft picks from these teams in these trades. If he is able to use them as currency (saving the Irving deal when IT4's hip was a sticking point), or drafting a few scratch tickets hoping for the 2nd round big win you alluded to. As such, I tend to rate them as no cost bonus shots for the organization, and shouldn't be viewed in any way as a negative. They only cost short money, sometimes create good will for the Celtics (drafting a kid out of Providence with no real shot), etc. So given that, I value the accumulated "WS over/under" more highly than an "WS average per pick" given that it in essence drags down Ainge's numbers because he takes a shotgun approach (2016) to 2nd round assets. One could argue he also does that with 1st round picks, and sometimes in a weaker draft he's left juggling and taking someone who's willing to go overseas for his development (Yabusele) rather than losing someone on the end of his roster he still considers a potential asset. Obviously, if he mismanages those assessments, then he's mismanaging the draft - so in the first round I totally understand the criticism.

Thanks again, for the work, and the interesting thread.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
okay, with the help of the draft-day-trades indicators on the Bk-Ref pages that scottyno points out, and some decent sorting choices, I bet I can get through allocating all of those draft-day trades back to 1989 in about an hour or so. That seems worthwhile to do, because right now, yeah, the signal is swamped by the noise of not having that accounted-for.

(edit: but seeing the notation of "ATL draft-day-traded Stacey Augmon to ATL", I start to wonder if there are a bunch of edge cases that I'm not even aware of...)

Finally, I think one of Ainge's strengths is his ability to extract draft picks from these teams in these trades. If he is able to use them as currency (saving the Irving deal when IT4's hip was a sticking point), or drafting a few scratch tickets hoping for the 2nd round big win you alluded to. As such, I tend to rate them as no cost bonus shots for the organization, and shouldn't be viewed in any way as a negative. They only cost short money, sometimes create good will for the Celtics (drafting a kid out of Providence with no real shot), etc. So given that, I value the accumulated "WS over/under" more highly than an "WS average per pick" given that it in essence drags down Ainge's numbers because he takes a shotgun approach (2016) to 2nd round assets. One could argue he also does that with 1st round picks, and sometimes in a weaker draft he's left juggling and taking someone who's willing to go overseas for his development (Yabusele) rather than losing someone on the end of his roster he still considers a potential asset. Obviously, if he mismanages those assessments, then he's mismanaging the draft - so in the first round I totally understand the criticism.
I think that the bolded is a reasonable approach. The only reason my default ranking was by value-per-pick was because of the fact that some franchises came into being more recently and as such didn't have as long a track record, so per-pick made sense to me. But especially if we're starting in 2003, with 30 teams, none of that comes into play (except residual futures picks and trades of future picks from previous years), and total value ("over-under") is probably the right comparison metric.

I still think the biggest flaw here is the use of Career Win Shares as the one-size-fits-all stand-in for "value to the drafting team", and wish I had a better one to grind with. Is there a readily-available dataset for download that someone knows about? To try and get VORP, I've tinkered with Play Index (I can always pull down a new dataset of all our players and then vlookup to bolt on new stats for them), but for some reason this search doesn't work, even though Lebron's career VORP is >100 it just comes back with him leading the results with a 33 VORP. I'm using Debut Year as a filter, rather than Draft Year or limiting the seasons played. But yeah - have I found a bug or am I just not searching correctly? edit: it was the latter, I was only getting Playoff VORP, fixed now. May add this shortly, so we can look at it through that lens as well.

And speaking of "no cost bonus shots", thinking of these acquired picks from trades as free rolls of the dice... that kinda makes Romeo Langford the ultimate exemplar of the category, doesn't it? Ainge went from his first choice in the draft to his first choice in the draft, meanwhile picking up the #14 pick two years later. Nice piece of business there.
 
Last edited:

RetractableRoof

tolerates intolerance
Lifetime Member
SoSH Member
Dec 1, 2003
3,836
Quincy, MA
okay, with the help of the draft-day-trades indicators on the Bk-Ref pages that scottyno points out, and some decent sorting choices, I bet I can get through allocating all of those draft-day trades back to 1989 in about an hour or so. That seems worthwhile to do, because right now, yeah, the signal is swamped by the noise of not having that accounted-for.

And speaking of "no cost bonus shots", thinking of these acquired picks from trades as free rolls of the dice... that kinda makes Romeo Langford the ultimate exemplar of the category, doesn't it? Ainge went from his first choice in the draft to his first choice in the draft, meanwhile picking up the #14 pick two years later. Nice piece of business there.
Yes, that's kind of the perfect example. Now, on one hand I'd like to give him credit for the strategy, but in reality it only has value when the player comes to fruition. Yabusele in the end didn't deliver anything of significance - though Zizic aided his ability to land Irving. I can't think of any reasonable way to enhance a GMs score for acquiring extra rolls of the dice without double counting them in some way, and the noise wouldn't be worth it.

1 hour to update all that data??? I couldn't do that WITH sleep, never mind now... lol And I was thinking similar things, if you keep your model in place, you could (relatively) easily add in the data for any metric you think might be more telling. A fun project that may end up even more interesting.

Edit: going forward, I wonder if the 2nd round picks will change value again, as the G league becomes more of a developmental entity for the NBA teams. I wouldn't be surprised to see the NBA think about adding a 3rd round at some point.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
Right now I'm wondering what franchise attribution to give Dwight Powell. Not traded on draft day, but traded the day after draft day to CLE, he then signs a contract two months later, and then a month after that, he's traded to Boston from there, while in preseason (Sept 25th) and without having appeared in a game.

Bk-Ref calls that CHA -> CLE -> BOS, but that's the first re-routing symbol I've seen from them where the move didn't happen on draft day itself.

Likewise Andrew Wiggins, trade was 2 months later. CLE or MIN?

edit: I'm leaning towards a boundary condition of "if the trade was within a week after the draft, assign to acquiring franchise, otherwise retain with drafting franchise". Jae Crowder was the example that sold me, a one-day-later trade really means you were drafting for that other team, not that you'd decided to take a bet on the guy but were suddenly bowled over by a trade proposal. If you trade him within a few days of drafting him, you were probably working on that trade before or during the draft.

edit2: and it's not that much data entry, frankly, RR. It's only a small fraction of the picks that got a draft-day-trade treatment. You just have the two windows side-by-side and take the updates on faith (though I'm still doing spot checks, hence the Powell case).
 
Last edited:

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
Another thing to consider, and I know there's no possible way to factor this in to a chart like that, is the expectation for each pick given the strength of the draft class. By this model Lebron and Anthony Bennett would both have the same "expected career" if I understand it correctly, but one of them was maybe the highest touted prospect ever, and the other was basically picking a name out of a hat because no one thought any of that class was worthy of being #1. A lot of it just comes down to what year did you have your ping pong luck in. It might be bias, but I feel like that Ainge's lottery luck has been in years where it wasn't as helpful, getting the #1 pick when there was no #1 lock superstar, and getting the #3 in what was perceived as a 1/2 person draft, and then missing out on KD.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
Another thing to consider, and I know there's no possible way to factor this in to a chart like that, is the expectation for each pick given the strength of the draft class. By this model Lebron and Anthony Bennett would both have the same "expected career" if I understand it correctly, but one of them was maybe the highest touted prospect ever, and the other was basically picking a name out of a hat because no one thought any of that class was worthy of being #1. A lot of it just comes down to what year did you have your ping pong luck in. It might be bias, but I feel like that Ainge's lottery luck has been in years where it wasn't as helpful, getting the #1 pick when there was no #1 lock superstar, and getting the #3 in what was perceived as a 1/2 person draft, and then missing out on KD.
The expected career length is only used for purposes of pro-rating a currently-young player's career out to the average career length of a player drafted at their position. Anthony Bennett still sucks even if you pro-rate his career to 12 seasons; he goes from 0.5 WS to 1.5 WS in the pro-rating, but then up against a benchmark for #1s of 55.3 WS, and rates out -53.8 as a result.

And you're right, I'm not sure we can correct for "strength of the draft class", because that's something qualitative, a narrative interpreted later into the story of that year. We could, perhaps, introduce draft-year rankings though: re-defining a Win as, if you're selected at pick N, and your career WS ranked against your peers drafted in the same year makes you the Nth-or-higher-ranked player in that draft, then you're a Win. If you're below the Nth-most-valuable player in your class, you were a Loss. I *think* that should result in a 50% win rate regardless of all else.
 

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
The expected career length is only used for purposes of pro-rating a currently-young player's career out to the average career length of a player drafted at their position. Anthony Bennett still sucks even if you pro-rate his career to 12 seasons; he goes from 0.5 WS to 1.5 WS in the pro-rating, but then up against a benchmark for #1s of 55.3 WS, and rates out -53.8 as a result.
No doubt he sucks under any system, but like, there was no Lebron to pick in that draft, it was just bad luck to have the #1 pick in such a shitty draft. You're basically just going off of "genenric #1 overall pick should be worth X winshares" right? So Lebron and Bennett would be starting from the same point of being expected to have similar careers, even though at the time everything assumed Lebron would be a franchise player and no one in Bennett's class would. Just looking at the top 10 in that draft it's possible that every player in it could have been considered a miss at #1, I assume #1 would be expected to have something like 50-60 WS based on #3 being 40.5.

I know there's no way to factor that in, just interesting to look at, especially for teams whose draft results would swing so heavily based on one pick.
 

RetractableRoof

tolerates intolerance
Lifetime Member
SoSH Member
Dec 1, 2003
3,836
Quincy, MA
No doubt he sucks under any system, but like, there was no Lebron to pick in that draft, it was just bad luck to have the #1 pick in such a shitty draft. You're basically just going off of "genenric #1 overall pick should be worth X winshares" right? So Lebron and Bennett would be starting from the same point of being expected to have similar careers, even though at the time everything assumed Lebron would be a franchise player and no one in Bennett's class would. Just looking at the top 10 in that draft it's possible that every player in it could have been considered a miss at #1, I assume #1 would be expected to have something like 50-60 WS based on #3 being 40.5.

I know there's no way to factor that in, just interesting to look at, especially for teams whose draft results would swing so heavily based on one pick.
Maybe you just eliminate Lebron from the system? lol, or arbitrarily cap his value at "2nd highest in slot +10" or something so it doesn't blow out all other picks at the slot position. Yeah, I know it's not how it's done... lol
 

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
Maybe you just eliminate Lebron from the system? lol, or arbitrarily cap his value at "2nd highest in slot +10" or something so it doesn't blow out all other picks at the slot position. Yeah, I know it's not how it's done... lol
It's more then just Lebron, if every team would have drafted the same guy then how much credit does a team really even deserve? I guess teams deserve a small amount of credit for drafting Lebron, Davis, Wall, Griffin, and Simmons, and any other no brainer consensus #1s I might be forgetting from 2003+ instead of trading the pick or doing something really dumb. Yeah those guys were all hits and ended up being franchise players to different degrees, but that didn't take any skill, any of us would have done the same thing.
 

RetractableRoof

tolerates intolerance
Lifetime Member
SoSH Member
Dec 1, 2003
3,836
Quincy, MA
It's more then just Lebron, if every team would have drafted the same guy then how much credit does a team really even deserve? I guess teams deserve a small amount of credit for drafting Lebron, Davis, Wall, Griffin, and Simmons, and any other no brainer consensus #1s I might be forgetting from 2003+ instead of trading the pick or doing something really dumb. Yeah those guys were all hits and ended up being franchise players to different degrees, but that didn't take any skill, any of us would have done the same thing.
I get what you are saying, but look at the Tatum/Fultz draft. At least 2/3 of the league would have grabbed Fultz. Ainge didn't, and got another asset for his trouble. Now Fultz is making a come back (young kid deserves it), so he would still justify the draft slot, but it's a funky one for evaluating which GM deserves what credit.
 

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
I get what you are saying, but look at the Tatum/Fultz draft. At least 2/3 of the league would have grabbed Fultz. Ainge didn't, and got another asset for his trouble. Now Fultz is making a come back (young kid deserves it), so he would still justify the draft slot, but it's a funky one for evaluating which GM deserves what credit.
Yeah, I make a distinction between most of the league would have taken X, and every team would have taken X and no one else was really even considered, but I admit it's a gray area because there's really no way to know for sure if any player was as much a no brainer #1 as they appeared in the media. Probably every mock had Fultz going #1, so if the Cs had just taken him instead of trading the pick and stating after the fact that they would have taken Tatum #1 anyway, maybe Fultz might have looked like the no brainer #1 too.
 

In my lifetime

Well-Known Member
Lifetime Member
SoSH Member
Dec 18, 2003
959
Connecticut
I just want to add the IF has really done an outstanding job. Not only in the initial analysis, but also in being willing, after reading comments to go back and modify. This is one of the best member analysis done on material in years. It answers the 2 basic questions of this thread and the Ainge drafter thread.
 

HowBoutDemSox

Member
SoSH Member
Aug 12, 2009
10,103
And speaking of "no cost bonus shots", thinking of these acquired picks from trades as free rolls of the dice... that kinda makes Romeo Langford the ultimate exemplar of the category, doesn't it? Ainge went from his first choice in the draft to his first choice in the draft, meanwhile picking up the #14 pick two years later. Nice piece of business there.
Until he earns a sobriquet more related to his play, I'm suggesting he be referred to as Romeo "Found Money" Langford.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
Not sure if it would help make things easier, but if you search on basketball reference by team draft history it lists next to the player if they were dealt in a draft day trade, like boston here:

I think this is newly added, but maybe I'd just somehow never seen it before.
I had to use archive.org to get Chicago's page, and that taught me that the feature was implemented sometime this summer (the May snapshot didn't have it but the September one did). So you're not going crazy after all!
It's more then just Lebron, if every team would have drafted the same guy then how much credit does a team really even deserve? I guess teams deserve a small amount of credit for drafting Lebron, Davis, Wall, Griffin, and Simmons, and any other no brainer consensus #1s I might be forgetting from 2003+ instead of trading the pick or doing something really dumb. Yeah those guys were all hits and ended up being franchise players to different degrees, but that didn't take any skill, any of us would have done the same thing.
Well, that's the alternative approach we could take from my last post. Line up everybody from the same draft, order by (projected) win shares, and rank 'em. So if the best guy in the draft went #9 (Dirk, 1998), he's now "#1", and so on 1-60 in each year. Then the GM gets credit for drafting someone at or below the player's resultant level of performance relative to their own draft, in proportion to how much he was over- or under-drafted.

The issue is, over-drafting the #1 best player at #20 and drafting the #41 best player at #60 are both a "+19" when clearly they're not the same value. How do we reconcile that? There's two choices we could use to take this alternate approach:

1) Use the same sliding scale of smoothed average pick value, so if you took the #1 best player in the draft at #21, you get credit equal to 73.7 (smoothed avg WS for #1) - 19.0 (smoothed avg WS for #21) = +54.7. But taking the #40 best player at #60 gets you (7.5 - 0.2) = +7.3.
2) Use the actual value of the player at that slot in their own draft. If it's 1998 and you draft Pierce (150.0) at #10, when the 10th-best player in that draft was Nesterovic (39.9), you get (150.0 - 39.9) = +110.1. But that's a lot noisier than it perhaps should be. The many people passing on Giannis are all going to take massive penalties despite there being a strong element of chance to the development of these prospects, so the first approach would be arguably more fair.
 
Last edited:

RetractableRoof

tolerates intolerance
Lifetime Member
SoSH Member
Dec 1, 2003
3,836
Quincy, MA
I had to use archive.org to get Chicago's page, and that taught me that the feature was implemented sometime this summer (the May snapshot didn't have it but the September one did). So you're not going crazy after all!

Well, that's the alternative approach we could take from my last post. Line up everybody from the same draft, order by (projected) win shares, and rank 'em. So if the best guy in the draft went #9 (Dirk, 1998), he's now "#1", and so on 1-60 in each year. Then the GM gets credit for drafting someone at or below the player's resultant level of performance relative to their own draft, in proportion to how much he was over- or under-drafted.

The issue is, over-drafting the #1 best player at #20 and drafting the #41 best player at #60 are both a "+19" when clearly they're not the same value. How do we reconcile that? There's two choices we could use to take this alternate approach:

1) Use the same sliding scale of smoothed average pick value, so if you took the #1 best player in the draft at #21, you get credit equal to 73.7 (smoothed avg WS for #1) - 19.0 (smoothed avg WS for #21) = +54.7. But taking the #40 best player at #60 gets you (7.5 - 0.2) = +7.3.
2) Use the actual value of the player at that slot in their own draft. If it's 1998 and you draft Pierce (150.0) at #10, when the 10th-best player in that draft was Nesterovic (39.9), you get (150.0 - 39.9) = +110.1. But that's a lot noisier than it perhaps should be. The many people passing on Giannis are all going to take massive penalties despite there being a strong element of chance to the development of these prospects, so the first approach would be arguably more fair.
I see the logic, and the execution(s), but don't agree with this unless it's an interesting additional column in a multiple form of measurement tool. My biggest concern (beyond your example) is GMs are at different life stages in the construction of their teams. The Warriors of 3 years ago might theoretically be drafting solely on position of need, regardless of ceiling, to extend their window. The Nets in that same draft would likely be drafting best player available or even best ceiling available. So this approach while offering fun math, would seem seem to reward a BPA strategy versus a team with a specific need.
 

Eddie Jurak

canderson-lite
Lifetime Member
SoSH Member
Dec 12, 2002
44,475
Melrose, MA
I had to use archive.org to get Chicago's page, and that taught me that the feature was implemented sometime this summer (the May snapshot didn't have it but the September one did). So you're not going crazy after all!

Well, that's the alternative approach we could take from my last post. Line up everybody from the same draft, order by (projected) win shares, and rank 'em. So if the best guy in the draft went #9 (Dirk, 1998), he's now "#1", and so on 1-60 in each year. Then the GM gets credit for drafting someone at or below the player's resultant level of performance relative to their own draft, in proportion to how much he was over- or under-drafted.

The issue is, over-drafting the #1 best player at #20 and drafting the #41 best player at #60 are both a "+19" when clearly they're not the same value. How do we reconcile that? There's two choices we could use to take this alternate approach:

1) Use the same sliding scale of smoothed average pick value, so if you took the #1 best player in the draft at #21, you get credit equal to 73.7 (smoothed avg WS for #1) - 19.0 (smoothed avg WS for #21) = +54.7. But taking the #40 best player at #60 gets you (7.5 - 0.2) = +7.3.
2) Use the actual value of the player at that slot in their own draft. If it's 1998 and you draft Pierce (150.0) at #10, when the 10th-best player in that draft was Nesterovic (39.9), you get (150.0 - 39.9) = +110.1. But that's a lot noisier than it perhaps should be. The many people passing on Giannis are all going to take massive penalties despite there being a strong element of chance to the development of these prospects, so the first approach would be arguably more fair.
I think ranking the players is a step in the right direction if we are evaluating GM drafting performance, as it eliminates problems of variance in draft quality by year, etc. But I wonder if you should include some smoothing in the rankings. For example, if the 5th through 8th best players in a given draft year all fall within a small number of WS, maybe when you do the rankings they should all be ranked the same.

I also think there's a problem with the idea of crediting the GM for the value add when he drafts a player who should have been off the board. The GM who takes the best player in the draft year at #20 made the right move (drafting the best player on his board), for which he deserves credit, but also benefited from things not under his control (19 other teams passing on the best player). Maybe there should be a cap on how much credit a GM gets from drafting a guy who shouldn't have been there. At the extreme, maybe the system should only penalize GMs: they can either take the best player on the board, or not, and if they do not, they should get penalized for the expected amount of value lost. With a bigger penalty to a team like 2013 Cleveland that uses the #1 pick to draft the #39th best player (Bennett) than would be asigned to the Wizards, who drafted the 6th best player (Otto Porter) at pick #3 even though each of the top 5 were still available.
 
Last edited:

RetractableRoof

tolerates intolerance
Lifetime Member
SoSH Member
Dec 1, 2003
3,836
Quincy, MA
I think ranking the players is a step in the right direction if we are evaluating GM drafting performance, as it eliminates problems of variance in draft quality by year, etc. But I wonder if you should include some smoothing in the rankings. For example, if the 5th through 8th best players in a given draft year all fall within a small number of WS, maybe when you do the rankings they should all be ranked the same.

I also think there's a problem with the idea of crediting the GM for the value add when he drafts a player who should have been off the board. The GM who takes the best player in the draft year at #20 made the right move (drafting the best player on his board), for which he deserves credit, but also benefited from things not under his control (19 other teams passing on the best player). Maybe there should be a cap on how much credit a GM gets from drafting a guy who shouldn't have been there. At the extreme, maybe the system should only penalize GMs: they can either take the best player on the board, or not, and if they do not, they should get penalized for the expected amount of value lost. With a bigger penalty to a team like 2013 Cleveland that uses the #1 pick to draft the #39th best player (Bennett) than would be asigned to the Wizards, who drafted the 6th best player (Otto Porter) at pick #3 even though each of the top 5 were still available.
This is incredibly stupid and childish. One of the worst comments I've seen in the port cellar.

Fake edit: Sorry, I was replying to the wrong post. lmao
 

wade boggs chicken dinner

Member
SoSH Member
Mar 26, 2005
30,482
I just want to add the IF has really done an outstanding job. Not only in the initial analysis, but also in being willing, after reading comments to go back and modify. This is one of the best member analysis done on material in years. It answers the 2 basic questions of this thread and the Ainge drafter thread.
Second this. Really great analysis.

And to me the discussion makes it clear why evaluating drafting ability is so difficult. Was drafting LBJ a win? While drafting Giannis is a coup for MIL, it was only made possible because other teams passed on him. Plus one thing we will never know - if MIL had a higher pick, would they still have drafted him?

Note: I don't count when organizations find a structural flaw like SAS realizing European players were undervalued or BOS/MFY/etc. realizing that they could throw overslot $ to increase their chances.

I think the axiom that the only way to draft better is to have more bites at the apple.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
Right now I'm wondering what franchise attribution to give Dwight Powell. Not traded on draft day, but traded the day after draft day to CLE, he then signs a contract two months later, and then a month after that, he's traded to Boston from there, while in preseason (Sept 25th) and without having appeared in a game.

Bk-Ref calls that CHA -> CLE -> BOS, but that's the first re-routing symbol I've seen from them where the move didn't happen on draft day itself.

Likewise Andrew Wiggins, trade was 2 months later. CLE or MIN?

edit: I'm leaning towards a boundary condition of "if the trade was within a week after the draft, assign to acquiring franchise, otherwise retain with drafting franchise". Jae Crowder was the example that sold me, a one-day-later trade really means you were drafting for that other team, not that you'd decided to take a bet on the guy but were suddenly bowled over by a trade proposal. If you trade him within a few days of drafting him, you were probably working on that trade before or during the draft.
As a followup, I landed on about 2 weeks as my "Draft-day trade" window, to change attribution of the draft credit from the franchise that actually used their draft selection, to the franchise that let him turn into a pro and get some real pro tape on file. After about 2 weeks post-draft, you're "developing", you're not "drafting" someone.

So the Tyus Jones 2015 trade is actually a great illustration of the boundary conditions here. On draft day 2015, MIN traded two second-rounders (Cedi Osman and a nobody named Raheem Christmas) to CLE for Tyus Jones. Osman, obviously, Cleveland chose to keep. However, a month later (6/25 -> 7/23), Cleveland decided to trade Christmas to Indiana for a 2019 2nd-rounder. So obviously, MIN was drafting Christmas "on behalf of Cleveland", and it was Cleveland that really made the selection. But just as obviously, waiting a month to trade Christmas means that Cleveland clearly had been working him out and decided he wasn't a fit, so on he went (and thence to the G-league). His performance isn't a reflection on Indiana's drafting, it's a reflection on Cleveland's.

(I've also found a few instances where Bk-Ref listed multiple trades off a draft, and had them in the wrong order, so I've had to audit a lot of the listings there, but it's also helped me consider where to draw that boundary for attribution)
 

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
Anecdotally, when I've had this discussion with others about Ainge's draft history in the past, I've always considered a pick a hit or a miss relative to the 5-10 guys selected after him. If you got one of the best several guys then it's a hit, even if they didn't end up so great the other guys you could have picked didn't end up any better. I imagine Ainge would perform better looking at things that way, because he hasn't found any superstars to inflate his numbers, just a lot of decent players.

If after all of this is done Ainge still ends up somewhere around 10th or so, I suppose what that tells us is that finding a few stars in the later picks is better than a bunch of solid players, and Ainge's approach could be a bit sub-optimal, thoguh of course it's not that simple, if we knew the types of players to target later who tend to fall and then become future stars they wouldn't be falling.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
48,204
Thank you IF for your work and this discussion. Even if the approach needs some refinement, it is far more interesting and informative than casual, ex-post takes on Ainge's drafting ability based on him missing out on Antetokounmpo and/or Gobert.

As others have noted, evaluating a GM simply on their drafting results has limitations. For example, context matters - where the team is in terms of competitiveness as well as the GMs job security. Then there are GMs/organizations who may not be great at talent evaluations but are good at aggregating talent using picks and young players as trade chips. Frankly, I find that skill more valuable because acquiring realized talent at a fair value requires less luck imo.

In any event, great work and thank you for sharing it with us. Also, kudos for the thoughtful responses.
 

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
Having fun looking at the chart more, it's absolutely astounding how bad Phoenix has been, especially when you consider that the chart is still giving them credit for at least some combination of Rondo, Gortat, Deng, and Nate Robinson, all guys they traded. Even more fun when you consider that if you looked at the 5-10 year period before 2003 they were probably one of the best with Amare, Marion, Nash, and Finley. I guess getting absolute zero of out the 2016 and 2017 drafts including 2 top 5 picks is kind of bad.

Is there an easy way to find out the best and worst team drafts of the period? Though it's probably just some boring answer like the Giannis bucks, and the Bennett cavs.

This is going to be awesome once it's totally done, please post somewhere if possible where we can see the breakdown by players and drafts, thanks.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
21,753
Pittsburgh, PA
OK I've finished the franchise re-assignments for draft-day trades, thanks be to toddler naptime. And I'm glad I did, because if you think about it, being so desirous of a prospect that you're willing to trade for the rights to them day-of-draft, to the point of having some other team "draft for you", says a lot about your drafting philosophy and priorities - you could almost argue those situations should count more for/against a GM.

First and most importantly, the Bk-Ref index name for Jonathan Bender is "bendejo 01", and if you've been staring at this shit as long as I have, that makes you giggle.

Secondly, I've updated the main post above with the new data (still using WS as a benchmark). This shows that once you account for draft-day (or near-draft-day) trades, Ainge rises to 9th on an overall-value basis (13th on a per-pick basis).

28118

Thirdly, I'm willing to do this for VORP or any other BK-ref advanced metric, as long as it's a counting stat. Would that add insight here?
 

scottyno

late Bloomer
SoSH Member
Dec 7, 2008
11,304
So it looks like Boston and Utah are the only teams in the top 10 that never made a top 2 pick (though a few of those teams botched that pick and still ended up top 10 to their credit), and one of a couple teams to be in the top 10 despite not drafting a superstar. That seems about right to me, I'd certainly take a couple of those teams over Danny if we're going purely on draft evaluation (I think Toronto has to be #1 right now), a couple of them just got lucky at the right time, and a couple of them the guys who drafted the value for them are long gone.
 

RetractableRoof

tolerates intolerance
Lifetime Member
SoSH Member
Dec 1, 2003
3,836
Quincy, MA
Thanks for the update. Very interesting data, I think it's fair to say that Ainge IS top 1/3 of the league, even higher if one considers (as Scottyno notes) that he's not had a top 2 pick. I think if one considers that short of trade swindles, since the Celtics don't appear to be interested in tanking - they aren't seeing a #1 pick anytime soon. And when they did get one, he managed to make more of it than was available on the surface.

A couple of thoughts: a) If it's not too difficult, can we see an update of what WBCD asked for earlier - a first round only set? I don't imagine Ainge moves much this time. b) Without having the raw data for all the teams, it's pretty obvious that by not having picks #1, #2 Ainge never had a chance at acquiring the Davis +78 WS. Similarly other monster #1, #2 WS payoffs. c) This hopefully will continue to show Ainge in a good light as the 3 youngish picks of Smart, Tatum, Brown are obviously good picks that will inch the Celtics numbers higher going forward. The same holds true for Zion and NOP.

I repeat myself, but thank you for presenting us with this data (and revised versions of it). It does begin to show what I felt was Ainge's performance, even given the difficulties of answering the question set. Well done.