It’s not too early for Sixers fans to panic about Markelle Fultz. And not just because 2017’s no. 1 pick now shoots free throws like he’s scared to hold the ball or because he was most recently spotted on a basketball court hoisting off-handed 3s.
It’s not too early for Sixers fans to panic about Markelle Fultz because draft busts can reveal themselves incredibly early in their rookie seasons. NBA players are tremendous athletes and, like any young entrant into any professional field, spend years honing their craft, cultivating new talents, and gaining valuable experience in the workplace. But even accounting for all that development, clues for how a given draftee’s career will evolve appear surprisingly quickly in his rookie autumn: Within just 10 games, highly touted, otherwise-equal draft picks begin to sort themselves by future potential, meaning it’s almost never too soon to identify a bust.
That notion might seem controversial, but it’s supported by decades’ worth of data. If a top-10 draft pick starts slow, he is likely to continue lagging behind his peers for years. Ten games is practically nothing in the grand scheme of an NBA season, let alone a career, but that tiny sliver of a sample still speaks volumes about a young player’s bona fides.
Before diving into these conclusions and their underlying data, let’s outline some parameters. First, this analysis examines top-10 picks based on the thought that anyone picked outside the top 10 doesn’t generate the same “bust” smears. Second, it generally examines their play during their first four seasons in the NBA (the standard length of a first-rounder’s contract) because a failed prospect who bounces back as a veteran still engenders the “bust” label. And third, it uses two advanced statistics designed to encompass a player’s overall performance. Basketball doesn’t break down so neatly into single numbers as, say, baseball does with WAR, but these metrics work on a macro level. To encapsulate career value, the analysis will rely on win shares, which estimate how many wins a player added or subtracted to his team’s total. Win shares cohere with the eye test, too: By this metric, the half-dozen best top-10 picks in the lottery era (from 1985’s draft onward) are LeBron James, Tim Duncan, Dirk Nowitzki, Kevin Garnett, Shaquille O’Neal, and David Robinson; the half-dozen worst (not counting players still on their rookie contracts) are Nikoloz Tskitishvili, Adam Morrison, Bobby Hurley, Jonny Flynn, Chris Washburn, and Rafael Araujo. That’ll do.
But win shares don’t vary much on a game-to-game basis, so game score will fit for shorter spans of games. This metric, which was created by John Hollinger, condenses a box score line into one number and, according to Basketball-Reference’s glossary, is supposed to “give a rough measure of a player’s productivity for a single game.” An obvious caveat is that because game score considers only numbers that appear in a box score, it doesn’t account for a player’s entire impact on the court, but it works as the best proxy for short-term player value.
Let’s jump into the numbers and start with an illustrative example of how powerful a predictor a rookie’s first set of games can be. Here are the worst top-10 picks from 1985 through 2013 (the last draft class whose picks have played four full seasons) by cumulative game score through 10 career games, with the absolute worst at the bottom:
9. (tie) Tyrus Thomas
9. (tie) Joel Przybilla
9. (tie) Jonathan Bender
8. Rafael Araujo
7. Anthony Bennett
6. George McCloud
5. DeSagana Diop
2. (tie) Mike Sweetney
2. (tie) Doug Smith
2. (tie) Darko Milicic
1. Mouhamed Sene
That’s also a list of draft busts, without exception. More than half were traded by their initial teams before their rookie contracts were up, and while the likes of Przybilla and McCloud recouped some value later in their careers—Przybilla on his third team, the unwanted McCloud after a stint in Italy—they still qualify as busts. McCloud is the only one of those 11 players ever to average even 11 points per game in a season.
Other players who barely missed appearing on that game score anti-leaderboard include Shawn Respert, Patrick O’Bryant, Chris Washburn, Jan Vesely, Hasheem Thabeet, Luke Jackson, Rodney White, and Nikoloz Tskitishvili. Bust, bust, bust, bust, bust, bust, bust, and super-bust.
For the whole group of rookies from 1985 to 2013—not just top-10 picks—the correlation between a player’s cumulative first-10 game score and his total win shares after four years is .63, on a scale in which zero represents no connection and 1 a perfect relationship. But we would expect that relationship to be pretty strong, because widening the player pool past top-10 picks accounts for fringe second-rounders and undrafted rookies who paired a silent first 10 games with a quiet career.
What is unexpected is that the relationship remains nearly as strong when restricting the sample to just top-10 picks, who are theoretically all on equal footing in their team’s future plans and in a similar place based on talent. For just top-10 picks, the correlation between game score after 10 games and win shares after four years is .58. There are exceptions, but the basic trend generally holds: The worse a player’s first 10 games, the less chance he has of becoming even a productive player, let alone the star his draft position would suggest.
For a point of comparison, a movie’s production budget and first-weekend box office earnings also have a .58 correlation. While the likes of Shaquille O’Neal and David Robinson—both no. 1 picks who have the best first-10 game scores in the sample—are in Marvel’s stratosphere, having secured both strong starts and overall careers, the latest no. 1 pick is approaching unseen indie territory.
Fultz isn’t quite as low as Milicic or Bennett, or even the Vesely-Thabeet “more fouls than made shots” level. But his 2.4 cumulative game score thus far—albeit through just four career games rather than 10 due to an injured shoulder—is in the range of only three other top draft picks in the lottery era; it’s equal to Andrea Bargnani’s first-10 score and within reaching distance of Kwame Brown’s and Pervis Ellison’s on either side. Expanding beyond top picks, Fultz’s neighbors in first-10 score include Kirk Hinrich (good! If not nearly as great as a no. 1 pick should promise), plus Randy White, Mark Macon, and Joe Alexander (bad! Very bad!).
Fultz’s company—particularly of fellow no. 1 picks—is damning. Brown, Bargnani, Ellison, and the aforementioned Bennett posted by far the worst first-10 scores among no. 1 draft picks selected from 1985 through 2013. Those four also rate among the worst top picks ever; among those 29 no. 1 selections, the surprise top-pick Bennett has recorded the fewest career win shares, and Brown, Bargnani, and Ellison rank fourth-, fifth-, and sixth-worst, respectively. (Second- and third-worst are Michael Olowokandi and Greg Oden, who also turned in lackluster, though not disastrous, sets of 10 games.)
If win shares aren’t convincing enough, this group also lags by the simplest basketball statistic: Bennett, Brown, and Ellison rank worst, second-worst, and fifth-worst, respectively, in the sample of top picks in career points per game. Bargnani jumps to “only” 10th-worst by this metric, but also has sufficient deficiencies elsewhere in his profile to confirm his bust status.
Beyond Fultz, this pattern is discouraging for a handful of other young players. Since the 2013 draft class, the worst first-10 game scores among top-10 picks belong to 2014’s Nik Stauskas and Noah Vonleh, 2015’s Mario Hezonja, 2016’s Dragan Bender and Thon Maker, and 2017’s Zach Collins, who has made just one shot through four games played. Stauskas and Vonleh both continue the trend, as they’re already on their second teams and averaging fewer than four points per game this year, while Hezonja’s scoring average has decreased in every season so far. Bender, Maker, and Collins still have time to turn their careers around, but their forebears with disastrous debut stretches don’t provide any optimism that they’ll be able to manage.
(As an aside, many of these listed names suggest that the first-10 method discriminates against international, “project”-type players like Bender and Maker, or Milicic and Sene further back. But Kristaps Porzingis did just fine in his first 10 games, as did Pau Gasol before him. Even for international players, it behooves to show at least some flashes early on.)
This research isn’t all negative for Philadelphia’s Process faithful. Just as Fultz’s early performance is reason for alarm, Ben Simmons’s is cause for celebration—just as an inferior first-10 game score signifies a shot at busting, a top-tier first-10 game score signifies a chance at stardom.
Simmons has the ninth-best first-10 score of any rookie since 1983–84, which is how far back Basketball-Reference’s Play Index extends. The 10 best players other than Simmons include seven Hall of Famers (Michael Jordan, O’Neal, Robinson, Ralph Sampson, Patrick Ewing, Hakeem Olajuwon, and Dikembe Mutombo), two more potential Hall of Famers (Grant Hill and Paul Pierce), and Brandon Jennings, who tossed a 55-point performance into his first batch of 10 games.
It’s ludicrous to map a Hall of Fame trajectory for Simmons based on his first eighth of an NBA season. He still hasn’t made a single shot outside 16 feet. But that’s mighty impressive company he’s keeping.
Beyond Simmons, this finding is good news for Chicago’s Lauri Markkanen, who scored in double digits in each of his first 10 games; none of the 33 top-10 picks from 1985 through 2013 with a higher first-10 score proceeded to suffer a bust-worthy career. And more broadly, it’s good news for most of the 2017 draft’s top 10. Besides Fultz and Collins—both of whom still haven’t reached 10 games, and therefore have room to boost their first-10 score—no rookie from the top of that class is in the danger zone.
But even a draft class otherwise filled with positive contributors can’t fully escape the stain of a calamitous top pick, meaning Fultz’s development remains the most important issue for an intriguing crop of rookies. Fultz might be just fine once his injured shoulder recovers (game score notably doesn’t fold in an injury factor) and as his game evolves to match NBA flow, but the early evidence doesn’t impart much confidence. And that’s the vital point here: The early evidence is all we have so far—and it turns out that the early evidence is a meaningful predictor of later performance.