The Toronto Raptors just won an NBA championship without a single lottery pick on the roster. It’s true. Kawhi Leonard, the Finals MVP, was picked 15th. Kyle Lowry, Pascal Siakam and Serge Ibaka were all picked at the back of the first round. Marc Gasol, Danny Green, Norman Powell and Patrick McCaw were second-rounders. Fred VanVleet and Jeremy Lin weren’t even drafted.
Not a single top-10 pick on the team. When asked about this stunner of a fact ahead of the NBA Finals, Gasol told reporters the draft was so important to him that he was actually asleep when he found out he was selected 48th in 2007. And now he’s a champion and NBA Defensive Player of the Year.
Said Gasol: “I don’t know what it says about the draft.”
On the topic, Lowry, who was picked 24th by Memphis in 2006, told the Toronto Sun: “Picks don’t mean anything.”
It certainly seems that way, to an extent. Not only was the Finals MVP drafted outside the lottery, but so was the presumptive regular-season MVP in Giannis Antetokounmpo, who was drafted 15th overall in 2013. The reigning NBA Defensive Player of the Year, Rudy Gobert, was drafted 27th by Denver and traded for the 46th pick and cash. The DPOY before him, Draymond Green, was a second-rounder himself.
On the other side of the equation, we’ve had two massive misfires at the No.1 pick in the last six drafts. Just 20 months after being the top pick in 2017, Markelle Fultz was traded for Jonathon Simmons and what will likely be a few second-rounders. Anthony Bennett, the No. 1 pick in 2013, is out of the league.
Lottery teams prepare countless reports, stay up ungodly hours and talk to hundreds of people to help identify the next star. Teams have more data than ever and, with social media, more insight into a player’s mindset and personality traits. And yet, the draft seems as much of a crapshoot as ever and lottery teams continue to miss on stars. Just look at the NBA champions.
As we’re getting ready for Thursday night, we have to ask ourselves: Are we getting worse at the draft?
* * *
The NBA draft has always been a difficult enterprise. Clifton McNeely, the very first pick of the very first draft back in 1947 (when it was called the Basketball Association of America/BAA) never actually played a single game in the NBA. Instead, he decided to coach high school basketball. (Imagine if Zion Williamson did that).
There have been busts and always will be busts. That’s the nature of trying to look into the crystal ball.
But are teams actually getting worse at looking into the crystal ball?
It would stand to reason that scouting technology and the rise of analytics have raised the accuracy of front offices. Gadgets have certainly helped other industries. Short-term five-day forecasts, for example, are nearly as accurate as two-day projections were three decades ago. Netflix estimates that 75 percent of viewer activity is driven purely by its recommendation algorithms geared to predict what you’ll watch next.
But the NBA draft, apparently, isn't getting any more predictive. If anything, it might be getting worse.
“It’s still a crapshoot,” said one prominent scout. “With all the available resources and new technology, we certainly haven’t gotten better.”
If the draft were perfectly predictive, the best player would always be picked first and the worst player available picked last. No undrafted players would ever step foot onto an NBA court.
Of course, that doesn’t happen. But the correlation of pick slot and win shares doesn’t seem to be getting stronger. That is, if we look into every draft since 1990, the best players -- as measured by prominent value metric win shares -- aren’t increasingly getting picked at the top, or near it.
Let’s look at the trends. Over the last five drafts, pick slot and player production (as measured by win shares) has a correlation coefficient of -0.38, indicating a moderate negative relationship between the two variables. In less nerdy terms, as picks go down from 60 to 1, we’re seeing player win share totals going in the opposite direction -- up and up and up. It’s not a perfect minus-1.0 relationship where the closer you get to No. 1, the more value is produced. (Cash winnings at a pro golf event are a perfectly negative relationship; the lower the score, the more money won).
If drafting was completely random, we’d see a correlation coefficient of zero. If the best players were always selected 60th, we’d see a perfectly positive 1.0 relationship. As you go closer to No. 60, the better the players get. Days without shaving and beard length are a perfectly positive relationship; the more days you leave the razor on the shelf, the longer the whiskers.
Where does minus-0.38 compare? The five drafts before that (2009 to 2013), that same number was minus-.50, indicating a stronger association between draft slot and player production. The five years before that, it was minus-0.41. From 1994 to 1998, that number was minus-0.52, the strongest five-year sample in the bunch.
In sum: the last five years -- where intel on players on and off the court has skyrocketed -- haven’t been the sharpest work by NBA executives. If anything, the last half-decade of drafting seems to be more random than ever.
To be fair, win shares may not be the most accurate metric to evaluate draft performance and even if it was, it tells only part of the story. Recent drafts may seem more inaccurate simply because players haven’t fully developed; Joel Embiid was considered a bust two years after getting drafted. And who knows, maybe Fultz becomes an All-Star in Orlando.
But even when there’s more information out there, it seems we’re not getting better at drafting. Executives have some theories as to why.
* * *
By now you’re probably wondering: Well, what about 2018? And you’re right. Deandre Ayton, Marvin Bagley, Luka Doncic, Jaren Jackson Jr. and Trae Young (picks Nos. 1-5, respectively) had terrific rookie seasons. For the first time since the 1984-85 season, the top five picks in the previous NBA draft have all been selected to the NBA All-Rookie first team (there was no such thing as a All-Rookie second team that season.)
But 2017 was horrible by comparison. The Philadelphia 76ers were so confident in Fultz that they traded a future first-round pick to the Boston Celtics in order to move up two slots to draft the University of Washington prospect No. 1 overall. The Sixers dumped him less than two years later after a bizarre string of events.
That same draft year, Donovan Mitchell, who has scored nearly 1,000 more points than the next-highest player in his class, was selected 13th overall. Kyle Kuzma, who is averaging 17.3 points per game in his NBA career, fell to pick No. 27 and was traded to the Los Angeles Lakers on draft night.
If the 2017 draft was redone, it’s hard to imagine 18 teams passing on Atlanta Hawks big man John Collins. Denver’s Monte Morris, who averaged 24 minutes per game for the 54-win Nuggets this season, went 51st. In a do-over draft, does Morris go higher than Fultz? That it’s even a question shows how much of a gamble the draft is.
But in talking with executives (who all declined to go on the record so close to the draft), there’s something else going on here. One prevailing theory seemed to arise: They’re not getting dumber -- the players are just getting younger.
There’s something to that. I pulled up all the draft data from Basketball Reference since 1990, when Derrick Coleman was selected No. 1 overall, and analyzed the ages of the draftees. Turns out that the average age of the 2018 draft class was 20.7 years old. That’s the lowest number of all the 29 classes I studied.
Last year was such a young draft. The top five picks were 19.9, 19.3, 19.3, 18.8 and 19.1 years old, respectively. That’s the first time in NBA history that the top five picks were teenagers. To put that in perspective, as recently as 2013 -- the Anthony Bennett year -- there were no teenagers in the top five; Bennett, Victor Oladipo, Otto Porter Jr., Cody Zeller and Alex Len were all in their 20s.
“The average age of drafted players is the lowest of all time,” noted one assistant GM. “That probably increases variance.”
That might be true, though the accuracy of the 2018 draft class figures to be an outlier. The age limit of 2005 has given rise to the one-and-done. The average age of top-five picks over the last five years is 19.3 compared to 21.9 in years 1990 to 1993. Heck, Dikembe Mutombo was the same age when he was drafted in 1991 as Bradley Beal is today (25). Michael Olowokandi was 23 when he was the No. 1 overall pick. Sure, Fultz might have been a mistake at No. 1, but at least there’s an excuse; he was barely 19 when he was drafted after a one-and-done season at Washington.
Another scout suggested one recent factor: the mock draft effect. The proliferation of mock drafts in recent years has created a false sense of general consensus. For example, NBA.com now publishes a Consensus Mock Draft that aggregates 10 different mock drafts with varying levels of intel.
Some teams use mock drafts, especially early in the season, to outline which players they watch in tournaments and on tape. As a player moves up the mock drafts or big boards, personnel people may be trapped into confirmation bias. If a player is rising, there must be a reason why. You see what you want to see.
“A lot of guys get drafted because they’ve been on a mock draft board at certain positions for a while,” said the West scout. “It’s kind of a mob mentality. Web sites and agents are gassing it up.”
Mock drafts are supposed to be a reflection of the market, which would be a helpful resource of data. But what if the market is a mirage? With so little information on these youngsters, executives may be desperate to find clues, no matter how good or bad those clues are.
* * *
The Golden State Warriors’ dynasty launched a league-wide hunt for the next Draymond Green, who was famously picked 35th in the 2012 draft. This year, front offices will surely hear from powerful people with a task to identify the next Fred VanVleet, the undrafted Wichita State product who played so well that he got a vote for NBA Finals MVP.
You could probably say the same for Malcom Brogdon, a 2016 second-round pick, who won NBA Rookie of the Year and became a 50-40-90 club member this season for the best regular-season team in the NBA.
There is one characteristic that Green, VanVleet and Brogdon share beyond the fact that they weren’t first round picks: They played all four years at college.
Perhaps we’ve invested so far in the one-and-done era that there may be what I’ll call a Diploma Bias -- guys who played all four years may be undervalued in the market. These prospects are usually labeled as “low upside” players, but VanVleet just proved his ceiling as one of the best players on a Finals team. Same with Draymond Green and Danny Green, the latter of whom went 46th after four years at North Carolina.
Sure, one-and-done players may be sexy because they did so well in just one year of experience, but it’s still, you know, one year of data. One executive sees the one-and-done problem having a compounding effect.
“Whoever plays in college next year won’t play against Zion Williamson,” said a front office executive. “That reduces their competitive environment.”
Which makes it tougher to evaluate, even with technology and analytics.
“Although there is more information, the signal in that information has dropped,” the executive said.
This isn’t to say that Cam Johnson, who is 23 years old after his senior year at UNC, should go No. 1 over Zion. But it might mean that he’s being unfairly punished for his age. For those who are looking at the next Draymond Green, it may be worth taking a look at Brandon Clarke, who is 22.8 years old but shares much of the same versatility and defensive acumen that we heard about Green. The same goes for Virginia’s defender, De’Andre Hunter, who is 21 and a half.
While everyone chases the next kid and bemoans how hard it is to draft in today’s environment, it might make more sense to zig when everyone zags. Old may be the new young.