Andres Perezchica grew up in California and was at UC Berkeley during the Warriors’ We Believe era. He then studied a master’s program in Finance in Monterrey, Mexico, where he became intrigued by the growing field of Sports Economics. He is currently living back in California.
During the 2009-10 NBA season, the Golden State Warriors tied a league record for call-ups from the NBA D-League with five. These five players combined to produce 0.81 wins (numbers taken from the Wins Produced site of Andres Alvarez). Over the course of their time on the team, they earned a combined $1.164 million.
However, last year the Warriors employed three other players that had gone undrafted and got their start in the D-League. These eight players produced 9.6 wins and were paid $6.07 million. This comes out to 1.58 wins per million dollars.
This Warriors team is a very good example of how useful the D-League can be to an NBA team. One wonders, can this possibly be a cheaper alternative to drafting players late in the first round, where their contracts are guaranteed?
In Table 1, the Warriors are split between players drafted late in the first round and players who started in the D-League. Each player’s Wins Produced [WP], WP48 [Wins Produced per 48 minutes] and WP per million are also shown. Devean George was the only Warrior on the team last season that was drafted late in the first round (outside of the lottery). George produced 1.36 wins and earned a salary of $1.6 (for a WP/mill mark of 0.85).
Although this is a very small sample, I wanted to see if it was an indication of a trend that spanned the NBA. If this is a trend and not an aberration, it would lead one to wonder if an organization would be better off employing players from the D-League rather than drafting a player late in the first round. Using Andres Alvarez’s data on Wins Produced, I looked at each player that went undrafted and was called up to the NBA from the D-League. This group of players was then compared to each player that went drafted from picks fifteen through thirty of the first round. The results of this analysis are reported in Table 2.
Just to review the table…. over the last five years, 77 different players were called up from to the NBA from the D-League. These players combined to play 40,619 minutes in the NBA. In these minutes, they combined to produce 77.9 wins. Their average wp48 comes out to 0.092, a mark that is essentially average. Combined, these players earned a salary of $42.029 million over this time. So this group of D-Leaguers produced 1.853 wins per million. Meanwhile, there have been 80 players that were drafted with picks 15-30 of the first round across the last five years. These 80 players combined for 237,812 minutes of playing time and produced 359.36 wins. This translates into a WP48 mark of 0.070, which is slightly below average. These late-first round picks were paid $293.521 million; and given the Wins Produced for these players, produced 1.224 wins per million dollars spent.
While there are roughly the same number of players that are called up from the D-League as players that are drafted in the latter half of the first round, the drafted players played almost six times the minutes given to the D-Leaguers. This may be largely due to the guaranteed three-year contract all first-round picks receive (editor’s note: published research – as noted in Stumbling on Wins – argues that there is a link between draft position and minutes played in the NBA). It is likely that teams play their first-round picks more minutes because of this, even if the players prove to be ineffective.
For further evidence of this is bias, of the 77 players called up from the D-League over the last five years, 27 never played in the NBA or were out of the league after their first season. In contrast, there are only five players who were drafted in the latter half of the first round who never played in the NBA.
In the minutes they have been given, though, the D-League call-ups have produced slightly more than the late-first round selections. They have a higher wp48 mark, posting a number that is just a shade below average. And the D-leaguers have also produced 0.629 more wins for every million dollars they were paid.
While the D-Leaguers out-produced the latter first rounders in these respects, it is a small sample size and it is unknown whether these players would maintain their level of production if their minutes increased to match the draft picks’ minutes. That being said… if the production of the D-Leaguers remained stable with an increase in minutes, it appears that an NBA organization may be wiser to trade away draft picks in the latter first round and search the D-League for players to fill out their roster. In this case, the team would be able to follow a player’s development without being forced into a three-year contract. In addition to potentially seeing more on-court production, an average salary earned by a D-Leaguer is under $0.5 million, while a player drafted from picks 15-30 earned an average of just under $1.5 million.
In sum — and this surprising result bears repeating — avoiding drafting a player outside of the lottery in the first round could potentially save an organization about one million dollars for each employee. And again, this savings may come with an even higher level of on-court production.
– Andres Perezchica
Wages of Wins Journal 
Chicago Tim
July 25, 2010
Fascinating article. So can we track the stats of D-leaguers anywhere?
some dude
July 25, 2010
Sounds like a clear case of OVB.
When people draft someone late in the 1st round, the player has to “prove his worth” in NBA games, whereas the D-leaguer proves some of his worth in D-league games. In other words, D-leaguers have a sample already known to GMs so when they’re called up, they’re more likely to produce than a random late 1st rounder with an unknown quality about them.
The argument becomes “The best D-leaguers are on average better than late 1st rounder.” Well, duh. We would expect that seeing as how we don’t call up poor D-league performers, no?
Now, what is the STD of the WP48? I have a hard time believing the STD for D-leaguers is larger. D-leaguers probably, with few exceptions, have a cap on their quality. I’m sure there are more Tony Parker types as 1st rounders than as D-leaguers.
Robert
July 25, 2010
Fascinating analysis, Andres. It is an example of the great quality of postings on this site. I believe Don Nelson himself asked why worry about free agents or draft picks when you have the d-league.
The coaches there, I believe are better qualified to teach the NBA game and the players there not glamour picks who feel entitled. It may say something about both the age of the players drafted first and the poor quality of college coaching by the guys with $5k suits and the huge shoe contracts.
Your data show this concretely.
dan
July 26, 2010
echoing the comments above — i really dug this post. in the future i’d love to see more posts that go beyond simply rehashing the wp48 numbers and really get at the heart of how teams can strategically weigh their assets. thanks for the fresh insight, andres.
Rockets fan
July 26, 2010
To echo a question I’ve raised elsewhere — but haven’t seen addressed — what is a reasonable estimate of the WS’s error margin? There are some obvious problems with the metric. (For example, it can’t attach a number to plays where a defender’s defense makes an offensive player miss a shot, it values all assists the same, and it does not account for charges drawn.) To be clear, I’m not saying the metric is bunk. But, I think it’s beyond dispute, that it isn’t perfect. Given that it’s not perfect, how imperfect is it? Is .7 really worse than .9? Can the WS make such fine distinctions? I don’t know, and I’d be interested in reading an answer.
One of my biggest complaints about this blog is that, even though we all accept there’s some error of margin, in nearly all posts the implicit assumption is that a higher WS necessarily means the individual contributed more wins. In other words, a player with a .150 WS will be treated as obviously better than a player with a .135. I’m not sure that’s the case. Sorry to (try to) highjack a thread, but I feel like my question comes up in nearly every post — including this one.
. . .
Another comment that is a bit more on topic. Even if the average d-leaguer is as good, if not better, than a late first round draft pick, are there more late first round draft picks who end up being very good players? If so, it might be worth the risk to take a first round player.
Thomas
July 26, 2010
I think if anything this probably shows that many non-lottery picks would likely gain by spending some time in the D-league. I think given how many players are coming in with only a year or two of college experience, this could certainly be beneficial to the league. Baseball certainly has no problems sending its top draft picks or even slumping veterans back to the minors, and I think that is something the NBA could use as well. I’m sure there are salary cap implications here that I’m discounting, but if the league wanted they could certainly change the rules to encourage use of the D-leagues and put a better product on the floor, while giving D-league fans a chance to see some bigger names.
brgulker
July 26, 2010
Wow, this is very well done! Great, great work!
I’ve made comments to friends in passing about wishing my Pistons utilized the “farm league” to develop players, but honestly, they were purely fleeting. Now I’ll have another scientific argument to bore my friends with!! ;)
bduran
July 26, 2010
Rockets fan,
Arturo at Aruto’s sillly little stats talked about this. I believe he said that the stats in the regular box score explain 85% of the variation in wins.
Great post, i’ve been wondering about this for some time.
BV
July 26, 2010
A lot of this may have to do with the fact that D-league call ups are desperate to perform to their absolute best, much more so than late first round picks who have guaranteed contracts. As David Thorpe of ESPN says hunger to succeed is a skill.
Chicago Tim
July 26, 2010
By the way, D-League coaches might be worth a look, too. Phil Jackson started coaching in lower-level professional leagues.
leon
July 26, 2010
Either way this undermines the ability of the nba scouts to process college performance and translate that into successful late round draft selections.
Nick
July 26, 2010
I would say this seems to be like a reasonable conclusion.
Guys drafted late in the first round, usually aren’t the consensus stars. So once you eliminate for the probability of those first few picks, being spectacular, the randomness of the draft comes into play.
And here, it makes sense that a D-Leaguer, who is often a late first round pick or a second rounder, would be better in the NBA. First of all, no team is going to sign a D-Leaguer who isn’t performing. So you would eliminate most of the REALLY terrible D-Leaguers from the average.
As has been mentioned, a late first rounder probably has more potential to be great, but, the article here suggests, that the potential to be great…isn’t all that high, and that it makes more sense to build out the fringe of your roster with D-League guys to save money for the REAL big ticket players.
It’s an interesting concept, the Warriors have done a GOOD job at building that, but they just haven’t had the stars to make this work. We’ll see if the David Lee era brings different results.
One thing to note, is that there is additional benefit to what GS has been doing. Morrow signed 3 year 12 million dollar contract with the Nets. By bringing in guys on the cheap, and raising there value, some other team might spend money on that player. Allowing your team, if you don’t resign him, to get an advantage on the market. Then, once again you scrape the D-League for the talent, and recycle. It wouldn’t be a large gain, but to me, if may make a couple of differences depending on which team bites on the player.
But back to the original point, it comes down to the “real lottery” system. Most people would rather roll the dice with a late first, and hope for the star pick, even if overall, it makes more sense to sit out the late first, and stick with D-Leaguers (don’t buy lottery tickets, just keep the money).
Buckeroo
July 26, 2010
This really is a thought-provoking post; I would think player personnel folks, especially in lower-revenue markets, could benefit from this type of thinking and information. Good work, and keep it up! Look forward to seeing an expansion on this research soon.
arturogalletti
July 26, 2010
Bdduran/Rockets Fan
The post is
http://arturogalletti.wordpress.com/2010/07/01/win-regression-for-the-nba-2/
Plain old boxscores explain about 85% of the variation in wins.
Rockets Fan, your questions are really on point.
Shawn Ryan
July 26, 2010
-Rockets fan
I’ve taken a play out of Andres Alvarez’s playbook and made a blog post out of your question. You, and of course anyone else interested, can view said post by clicking on my name.
Jason E
July 26, 2010
Picks 15-30 get more than the NBA minimum and have it guaranteed for a couple of years. They’ll ride through this putting up terrible garbage time minutes for a few years, or roughly 16 “10-game contracts”. The cumulative effect of a couple of years of garbage time minutes for guys who are stuck on the payroll drags down the group; d-leaguers who cannot perform do not continue to get even the garbage minutes, while those who can perform are essentially locked in at the NBA min for a few years and swamp the d-league sample when it comes to minutes.
reservoirgod
July 27, 2010
I appreciate that people enjoyed this post because it’s different than the typical posts but I don’t understand the line of thought that sees so much value in it. D-Leaguers play for bad teams and late 1st rd picks don’t play for good teams. If the value comes from the suggestion that good teams could save a few hundred thousand dollars by trading their 1st rd picks & signing D-Leaguers – who cares? I don’t think championships are being decided by late 1st rd picks & D-Leaguers. Now, if they ARE being decided by those players, THAT would be a valuable story…
khandor
July 27, 2010
In this instance, reservoirgod is right on the money.
re: “This Warriors team is a very good example of how useful the D-League can be to an NBA team. One wonders, can this possibly be a cheaper alternative to drafting players late in the first round, where their contracts are guaranteed?”
Actually, what the GSW are a prime example of is a business operation which is perfectly content to cheat its own fanbase by “filling out” its roster with players that have no hope whatsoever of developing into a legitimate contending team in the NBA … while reaping huge long term capitol gains for their owner[s].
Utah Jazz fans should be thrilled that a useful player like Sundiata Gaines was added to their roster last last, as a D-League call-up.
GS Warrior fans, on the other hand, should be outraged that their team decided to use D-League players for as much as half of their entire roster for certain games last season … while operating under the guise of a legitimate NBA team.
khandor
July 27, 2010
Rockets fan,
Amen.
Tom Mandel
July 27, 2010
This is very imaginative and valuable statistical research! Thank you.
Tom Mandel
July 27, 2010
Having said that, let me raise an issue:
1. We know from SoW that after pick 1 there is no statistically significant correlation between where in the draft a player was picked and his productivity (measured by WP48).
2. In particular, from http://www.wagesofwins.com/Draft10910.html we know that of the top WP48 rookies in the ’09 draft, half of them came from the bottom half of Round 1.
3. This is a strong indication that, outside of the first pick, the bottom half of Round 1 yields just as good players as the top half.
4. Given that players in the bottom half of Round 1, despite providing nearly equivalent productivity, are substantially less expensive than those in the top half (no doubt throughout their career but especially their rookie contracts), surely we must conclude that these picks are more valuable than top half picks.
5. From Andres’ argument above we learn that in addition to this surprising result, these same bottom half of round 1 picks are *less valuable* than call-ups from the D-League!
Isn’t it interesting that the same statistical measures can be employed to show how *valuable* these picks are and how *little value* they have. And that to conform these results to each other, we must conclude that D-League call-ups are more valuable than all but the first pick in the draft.
Feel free to point out any way in which I am misusing either the data or logic behind either Dave’s analysis of draft productivity or that of Andres.
Tom Mandel
July 27, 2010
Rockets Fan asks what is WP48’s “margin of error,” and Shawn Ryan responds with a blog post on the subject.
I’m sorry, but both of you misunderstand statistical facts. Let me see if I can make clear what I mean:
1. Is .09 really better than .07? asks Rockets Fan. What can he mean? *Obviously* .09 is better — by .02 !
2. That can’t be what he means to be asking. He must mean is “a player rated .09” really better than “a player rated .07”? Right?
There is no answer to this question, because there is no such thing as “a player.”
If you take a random sample of players who rate .07, and an equally large random sample at .09 — the latter should be a better group. That’s it. That’s all the .02 *statistical* difference means.
Asking how a statistical formula compares to reality is *the wrong question* in other words. What makes WP48 useful is that it is better than any other formula we have in accounting for players’ roles in producing their team’s record. It’s better than PER, EFF, Win Shares, or any other offering. There’s no other measure of its utility. And no way to ask “how ‘real’ is WP48?” It’s not a picture of reality.
Tom Mandel
July 27, 2010
In his blog post, Shawn makes the classic error of conflating “precision” with “truth” — the more minutes we have for a player, he says, the better able we are to use WP48 to say how good he is.
No. The more minutes we have the more precisely we can determine his WP48. Also the more precisely we can determine his PER. And his EFF. The better we can *predict* his future WP48 (and PER and EFF) too.
A better understanding of how good a player is certainly comes to us as minutes increase! But we need no recourse to a particular statistical formula or methodology for that to be true. If we think scoring is all-important, more minutes tell us how well a player scores.
What WP48 does is tell us what a player’s box score stats mean in respect of wins/losses over the range of the data.
Jason E.
July 27, 2010
@khandor:
As a Warriors fan, I’ve felt cheated for most of my conscious lifetime, but their ‘decision’ to use D-leaguers last year doesn’t add to that. The Warriors experienced a rather ridiculous level of injuries last year. Mid-season, with no room under the cap, I’m not really clear on what else they *could* have done. It wasn’t like there were guys who were going to turn them into contenders who were waiting for roster spots but were blocked by the d-league call ups. Perhaps you meant something else, but your reasoning that it was somehow a choice in order to cut costs doesn’t really follow from what actually happened.
For what it’s worth, I feel far more cheated that Monta Ellis was allowed to hoist up 22+ shots a night with such a low rate of return on his possessions rather than sharing the ball a bit with some of the more efficient scorers –whether those scorers have draft-night pedigrees or were culled from the d-league.
Leon
July 27, 2010
A key thing is that (also mentioned in SoW) that decision makers aren’t drafting the right players. There are many players, not even picked by a single team with two “choices” for players who are trumping those players picked as the teams first choice.
If it is such the case that those decision makers can’t make the right choices then maybe a better option is for them to trade away the picks that they will (inevitably) waste.
I know anecdotal (and full of hindsight) evidence doesn’t trump scientific evidence, but surely somebody could see that the difference between Andrea Bargnani (1st pick) and Brad Miller (undrafted) wasn’t that large?!
Eliot
July 27, 2010
From what I’ve seen, the D-League has a lack of solid bigs (most likely due to the short supply of tall people) so any big who has near NBA level talent will shred the league without much effort. However, it seems pretty well stocked in guards, and the Warriors seem to have realized this getting several rotation level players over the years as noted in the post.
khandor
July 28, 2010
————————–
re: It’s not a picture of reality. – Tom Mandel
————————–
I simply cannot deny that this specific sentence had me ROTFLOLWKAS.
————————–
re: The Warriors experienced a rather ridiculous level of injuries last year. Mid-season, with no room under the cap, I’m not really clear on what else they *could* have done. It wasn’t like there were guys who were going to turn them into contenders who were waiting for roster spots but were blocked by the d-league call ups. Perhaps you meant something else, but your reasoning that it was somehow a choice in order to cut costs doesn’t really follow from what actually happened. – Jason E.
————————–
Did the Warriors last year experience a greater injury toll than the Portand Trail Blazers?
Hmmm …
I think not.
IIRC, at no time last season, did the Trail Blazers’ actual line-up resemble that of Golden State’s, re: the gross number of D-League call-ups used in a real NBA game.
In general, it’s what’s done … and, how it’s done … throughout the entire year … and, across several years together … that determines whether, or not, a specific franchise should be properly characterized as having “cheated” their own fanbase, re: the quality level of the players on their roster.
Everything we do in life involves making a choice; and, no single choice exists in a vacuum.
dan m
July 29, 2010
I can see an argument for calling up D-leaguers if you need a spot minute role-player. That doesn’t mean that D-leaguers are more valuable than late 1sts round picks.
The issue is teams are stuck in guaranteed contracts with late 1st round picks so they use different criteria in selecting them than they would for D-leaguer callups. Teams are looking for players that have “upside”, “potential”, some sort of NBA starter skill that can be developed.
There are going to be far more busts and benchwarmers in a group picked for potential and upside than a group picked to fill a specific spot role on the bench of a team.
The real comparison should be between D-leaguers and 2nd round picks. This is an apples to apples comparison where both set of players aren’t on guaranteed contracts and both set of players are being picked for a specific role that the team hopes they can fill.