Nylon Calculus: Lonzo Ball, Markelle Fultz, and early NBA Draft projections

There has been a lot of talk about the prospective 2017 NBA Draft class, which is shaping up to be one with great depth at the top. The big question is at the very top of the class, for most of the year the athletic dynamo Markelle Fultz has been the scouts’ favorite prospect, while Lonzo Ball has clambered up the ranks to be a solid number two.

Ball Is Life?

My models flips that, based on its projections that Ball is the best prospect in this draft and on par with the best prospects in the last four drafts. My P-AWS draft model has Ball at 9.96 and Fultz at 8.1, which roughly translates to Fultz having an average expected outcome of a border line All-Star and Ball having an average expected outcome of a clear All-Star.

Both Ball and Fultz are point guards, which is somewhat helpful trying to look at them head to head, as it gives us a better basis to compare stats. Essentially, the model sees Fultz and Ball as roughly equal by a numbers of measures; in scoring Ball’s efficiency is balanced by Fultz’s greater volume, rebounding has a slight edge to Fultz, and blocks and steals to Ball. It’s in distribution that Ball creates separation with 25 percent more assists and 23 percent fewer turnovers. The younger Fultz does make up some ground with the age adjustment however. All those can be seen on the chart below focused on the two top prospects.

One thing that does differentiate the two players is the state of their teams. In short, Ball’s UCLA Bruins are very good with a 24-3 record, Fultz’s Washington Huskies are not with a 9-18 record. There are arguments to be made on both sides of this as to how the team context should affect our evaluation of each player as a prospect.

On the one hand, Fultz clearly has a less talented supporting crew, which can force worse shots and lead to more turnovers as he is the clear focus of any defense. On the other hand players going to better teams in the NBA often fair a little worse in box score stats as they lose opportunities and share the ball more. In addition, the 538 draft model the last few years has flagged a defensive rating and offensive rating by the team as an import indicator for the individual player, being on a bad team increasing the possibility of being a bust.

Next: Good defense isn't enough to win in the NBA anymore

Those context issues aside, I also looked at the error rate of the model in the past in order to do a monte carlo simulation of 500 runs to quantify how often the model would expect Ball to be the better player, as measured by target variable. This is, of course, assuming the end of season model projection is similar to the current one. In the simulations, Ball was the better player in 72 percent of the runs. If I factor in the scouting preference for Fultz, that falls to 70 percent, the small move reflective of the fact that Ball is not that far behind as the second overall rated prospect, as of now.