The Top 50 NBA Draft Prospects, According To Our CARMELO Projections

The NBA doesn’t waste much time before moving on. The 2018-19 season has been over for less than a week, and the Toronto Raptors are still picking up the debris from their jubilant¹ championship parade. And yet, Thursday’s NBA draft will mark the de facto beginning to the 2019-20 season. So we at FiveThirtyEight are also wasting no time: We’ve fired up our CARMELO projection system and run the numbers looking ahead to the next season and beyond. We’re still making some tweaks and improvements to the way we’re projecting veteran players, so we’ll be rolling out those numbers sometime in the next few weeks. But for rookies, we have the data, and the CARMELO computer is all ready to go. Let’s take a look at the best statistical prospects whose names should be called from the podium by Adam Silver.

First, though, a little refresher on how this works. CARMELO (the Career-Arc Regression Model Estimator with Local Optimization) is our system for predicting the career of each NBA player, based on how things tended to pan out for similar players from the past. For rookies, we use a database of college stats (adjusted for pace and strength of schedule) since 2001 provided to us by ESPN’s Stats & Information Group, plus biographical information like a player’s height, weight, age and — before the draft — scouting rankings.² Those latter few categories actually matter quite a bit, so older players and those regarded poorly by the scouts will need much, much better college stats to achieve the same projection as a younger player who scores better on the eye test.

Like we did last year, we’ll be ranking the members of this draft class on each player’s “upside” wins above replacement — the number of wins he’s expected to add above a minimum-salary replacement (at the same position) over the first seven seasons of his career, zeroing out seasons in which he is projected for negative WAR.³ One big disclaimer: These rankings don’t include players from overseas leagues, such as Sekou Doumbouya (who played in France), nor do they include players who didn’t accumulate enough time in college, like potential Top 5-pick Darius Garland, who logged only 139 minutes as a freshman at Vanderbilt because of a knee injury. We don’t have a good sample of data on these kinds of players, so CARMELO can’t really render an assessment right now (though we will eventually assign them ratings for our team depth charts next season).

Anyway, let’s move on to the rankings:

Unsurprisingly, Duke’s Zion Williamson ranks No. 1 in our prospect ratings with a massive upside WAR of 36.7 projected wins over his first seven pro seasons. To give that number a sense of scale, last year’s top projected prospect — Texas’s Mohamed Bamba — led the pack with an upside rating of just 21.8 WAR, so it’s fair to say that Williamson is a significantly better prospect than we’ve seen the past few years (if not much longer than that).

We’re not exactly going out on a limb with our projection: Zion is the top choice in basically every mock draft on the planet. But it’s always nice when the numbers back up the overwhelming consensus of opinion. During his lone season at Duke, Williamson rated better than the median Division I player in every major facet of the game according to the advanced metrics, headlined by a 93rd percentile usage rate and a 100th percentile true shooting percentage. His top comparable player — another former Blue Devil, Jahlil Okafor, whose career has largely underwhelmed — might be concerning, but it mainly speaks to how unique Zion is. The 0-100 scaled “similarity score” between Williamson and Okafor is just 37.1, meaning they’re not very similar at all. (By comparison, the similarity between Zion’s teammate RJ Barrett and model-namesake Carmelo Anthony is a healthy 60.1, which is normal for a top comp.) Simply put, Williamson is a unicorn. We’ve seldom seen a player quite like Zion, who is very short for his position but has great stats across the board (even in terms of rebounds, blocks and steals) and scores so often with such incredible efficiency.

Another unsurprising result is the presence of Murray State’s Ja Morant at No. 2 overall. Morant enjoyed a breakout sophomore campaign with the Racers, culminating in a triple-double in an impressive NCAA Tournament win over Marquette. To be sure, Morant has flaws in his game (he needs to work on efficiency in terms of both shooting and turnovers), and his upside projection isn’t on Williamson’s level, but he would have been CARMELO’s best prospect of last season, and some of Morant’s top comps — such as John Wall and Derrick Rose — offer a glimpse into his star potential.

After Williamson and Morant, there is a huge drop-off before the next group of prospects. Jarrett Culver of Texas Tech, Jaxson Hayes of Texas and Barrett form a clear-cut second tier below Williamson and Morant, and each comes with his own strengths and weaknesses. Hayes is a low-usage, high-efficiency big man who can protect the rim but is a work in progress on offense beyond finishing plays made by others. (Brandan Wright comes to mind as an archetype.) Barrett is a low-efficiency, high-usage swingman with unimpressive defensive indicators, in the mold of Anthony, Andrew Wiggins or Brandon Ingram. Barrett does come with a good amount of upside, in terms of potential value several years into his career, but he also looks like this draft’s best example of a high-risk/high-reward prospect who may just turn into an inefficient high-volume scorer (that classic bane of every stathead). And Culver is a nice all-around wing whose best-case comps include Harrison Barnes and Rudy Gay. Noted bust Joseph Forte isn’t the most encouraging comparison, but Culver’s versatility is a big plus as a small forward prospect.

The rest of the draft class drops off steeply after Barrett at No. 5 in our rankings. Some of the players who might be drafted highly but our model isn’t as fond of include Virginia’s De’Andre Hunter (fifth in the scout rankings vs. 14th in our projections), Duke’s Cam Reddish (seventh vs. 12th), Indiana’s Romeo Langford (11th vs. 20th) and Gonzaga’s Rui Hachimura (18th vs. 36th). By contrast, players who might be underrated in the draft include Shamorie Ponds of St. John’s (49th by the scouts vs. 23rd in our rankings), Auburn’s Chuma Okeke (41st vs. 17th), Jontay Porter of Missouri (42nd vs. 18th), Matisse Thybulle of Washington (28th vs. 13th), Tyler Herro of Kentucky (17th vs. ninth) and Iowa State’s Talen Horton-Tucker (21st vs. eighth).

Just for the sake of comparison (and transparency), here’s a version of our upside WAR rankings that doesn’t include the scouting rankings as an input,⁴ which can be viewed as a sort of “stats only” ranking of the prospects this year:

There are a few differences near the top: Hayes and Horton-Tucker leapfrog Morant as the best prospects behind Williamson (CARMELO really appears to love Horton-Tucker), while Culver and Barrett tumble in favor of efficient, rim-protector-type bigs such as Porter, Charles Bassey of Western Kentucky and Oregon’s Bol Bol, the 7-foot-2 of son of Manute Bol. I wouldn’t start using these rankings to make my picks if I were an NBA general manager — it’s been shown that scouting rankings are easily the most predictive component any draft projection can add. But they are still somewhat interesting in terms of helping to identify undervalued players whom the eye test alone might miss.

And it’s telling that Wiliamson still emerges as the clear-cut No. 1 prospect regardless of whether we’re looking only at statistics or a hybrid between metrics and scouting information. The guy is probably going to be pretty dope in the NBA, and Morant might not be far behind as the likely No. 2 pick. After that, there are a few more solid choices and then a whole bunch of uncertainty. That’s pretty standard for the NBA draft, where potential value drops off quickly after the first pick or two — but this year’s class might be unusually top-heavy even by basketball’s normal standards.