The process of catching the champion Golden State Warriors continues (or in some cases starts) Thursday with the NBA draft, in which teams will be doing everything they can to identify the players who can elevate their franchises.
Separating the players who will be All-Stars from those who will merely be role players is part of a front office's job, and one that we've set out to quantify using our NBA-caliber player-percentage metric below.
In the chart below, we've ranked the likelihood of members of the 2017 draft class becoming NBA-caliber players based on a projection of their statistical plus-minus (SPM) in Seasons 2 through 5 in the league. (Seasons 2-5 roughly represents the time frame players are under team control for below-market prices but omits rookie years, which are often not reliably predictive of future success).
After we've determined whether a player is NBA-caliber, we've drilled down further to predict whether they'll become one of three levels of players:
All-star (gray bar, see below): Chance a prospect averages better than 2.2 SPM (statistical plus-minus) in Seasons 2-5.
Starter (red bar, see below): Chance a prospect averages between minus-0.3 and 2.2 SPM in Seasons 2-5.
Role player (black bar, see below): Chance a prospect averages between minus-1.4 and minus-0.4 SPM in Seasons 2-5.
(For more on the model's inputs, outputs and performance projections of past classes, see FiveThirtyEight's 2015 write-up or this explainer on ESPN's draft model.
Lonzo Ball is the top prospect, according to ESPN's NBA draft model, with a projected average SPM of 1.47 in Seasons 2-5 of his career. Ball is also the least likely bust among all prospects, with a 28.6 percent chance of him not playing at least at a replacement level in the NBA. Despite all the debate about Ball, the model has him as the lowest-risk player in the draft. The UCLA product's adjusted assist rate of 30 percent is fifth best of any player in this draft, and in spite of his unique shooting style, his adjusted eFG percentage of 64 is the highest for a collegiate point guard prospect in our data set (2001 to present).
Jonathan Isaac of Florida State is right behind Ball in terms of his NBA-caliber projection, though the components that make up his overall number are much different. He's only ninth on the list when looking at his All-Star likelihood, but our model projects Isaac as the most likely starter-caliber player. He's an elite defensive rebounder and shot-blocker for his position, but Isaac lacks the elite shooting potential of players with higher ceilings.
The most interesting player in this draft may be point guard Frank Ntilikina of France. Ntilikina is just 18 years old, and in just under 600 minutes in French league Pro-A this season, he shot an above-average 43 percent from behind the 3-point line, equaling an eFG percentage of 57. His efficient play at such a young age in Europe makes him the most likely member of this class to play at an All-Star level, but his limited minutes and the unknowns that accompany the fact that he is coming from Europe, also give him a 59 percent chance of being a bust according to our model.
The consensus No. 1 prospect of scouts around the league seems to be Markelle Fultz, but our model does not support that assessment; Fultz is No. 7 on our list. Among point guards in this draft, Fultz was above-average in block percentage and assist percentage but as a shooter was much closer to Kobi Simmons -- who ranks last among collegiate prospects in adjusted eFG percentage -- than to Lonzo Ball, who ranks third. The Philadelphia 76ers made a major move over the weekend that pointed to them selecting Fultz, but the organization might want to take pause after looking at the analytic evidence.
That said, when poring over the projections, it is important to remember that every team will have different goals when drafting. The Sixers and Lakers are far from contention and should be more willing to take a risk on a player who perhaps has a higher ceiling (All-Star percentage, represented by the gray bar above) even if there is a greater risk he doesn't pan out. The Celtics, despite having the No. 3 pick, already have a strong team and would seem a better candidate to target a player who is low-risk and most likely to contribute. Boston doesn't necessarily need a potential All-Star to build around; they need a player who can contribute right away.
Beyond the lottery
Isaac Humphries (Kentucky), Thomas Bryant (Indiana), Caleb Swanigan (Purdue) and Tyler Lydon (Syracuse) are all top-30 players in our projections, though none can be found within the top 29 picks of Chad Ford's latest mock draft.
Humphries (not listed in either round of the mock) had a very high adjusted offensive rebound percentage while at Kentucky, and rebounding is the collegiate skill that most consistently carries over to the NBA, according to the projections. Humphries has little chance of becoming an All-Star, but the model gives him a 28 percent chance of being above a replacement-level player. His likelihood of being a role player should be attractive to teams at the end of the first round and into the second.
Bryant (No. 59 in the latest Ford mock), the 6-foot-10 center from Indiana, took 18 percent of his shots from 3-point range, after adjusting for opponents and pace. In the modern NBA, a big man who is also unafraid to shoot from outside is a valuable commodity. Bryant has a 29 percent chance of being a role player in the NBA, according to our model.
Swanigan (No. 43) had a monster 2016-17 season at Purdue, but as a prospect he's polarizing. Swanigan grades out as above average in rebounding, 3-point shooting and assists for the power forward position, but he turns the ball over too often and was well below average in steals and blocks for a power forward.
Lydon (No. 30) projects better on defense than on offense but still had a decent eFG percentage and had a good turnover rate while at Syracuse.
A look at our complete rankings of NBA-caliber players for this class:
For more from ESPN Analytics, visit the ESPN Analytics Index.