Andrew Johnson has put together an interesting article that takes an analytics approach to evaluating the top draft choices.
http://nyloncalculus.com/2016/05/23/who-should-the-celtics-take-with-the-third-pick-in-the-2016-nba-draft/The article and links to other articles explain the model and all the data that has gone into it, but basically this incorporates not only their recent year playing data but also earlier age recruiting index info and physical metrics. It does include combine data for those players that participated. The model has been trained against a large corpus of historical data to validate projections, fed by pre-draft data, of future NBA performance.
He starts with straight PAWS model results (which do not factor in scouting rank indices) for the top 10 players currently ranked by DraftExpress in this chart (they are listed left-to-right in order based on the current DraftExpress ranking):
To understand the chart, the important number is the number next to each green dot, as that is the actual model result for that player. Each colored bar section indicates different contributions to their score and whether that is a positive or negative contributor to the score. I.E. positive contributors are stacked onto that player's bar on the positive side, above zero while negative contributors are stacked onto it on the negative side, below zero.
For example, on Kris Dunn's bar, because of his age, being older hurts him in two ways: (1) his age relative to his draft peers hurts his projected upside slightly and (2) his age relative to his competition decreases the value of his production slightly. Dunn also gets pulled negative by poor rebounding and scoring, but has very strong defensive and playmaking which result in an overall solid positive number.
Conversely, Ingram, being so young compared to his peers and relative to his draft peers gets a large positive contribution from his age component. Ingram has a small negative contribution due to poor playmaking numbers, but the rest of his contributors are positive.
Simmons and Bender are the only players here for whom all contributing factors are positive.
Some quick notes:
Bender's data used in the model includes most of this season at Maccabi but does not include his most recent playoff games, which might be significant because he just played 10% of his entire minutes this season in the last two playoff games (and played well). Bender is, of course, the only one still playing and his data may not be complete until June 9th. Also, Bender's physical data (ht, wingspan, etc.) is probably due for revision. The model also does not include Bender's stellar FIBA tournament play. Basically, the results are going to likely undervalue Bender.
The Jaylen Brown and Buddy Hield results are huge red flag alerts for 'Do Not Want' because results under 4.0 correlate too strongly with 'roll player or less'. Skal is also considered a Do Not Want.
Johnson looks briefly at a 'PAWS Top-10' ranking, which brings in Valentine, Sabonis and others into the picture, but I'll let you read the article to get that. He discusses issues with those results.
He then evolves to present his 'Scouting Informed PAWS model' results which combines the PAWS model with scouting ranking data. This model has proven to be the best predictor of NBA performance.
This basically breaks the draft down into the tiers:
Tier 1: Simmons & Ingram -- clearly ahead of the pack. Arguably Simmons (9.38) is well ahead of Ingram 7.80) in this but both are clearly separate of the rest.
Tier 2: Bender, Davis, Poeltl and Stone all post nearly identical scores (5.79-5.97). Keep in mind that Bender's numbers might move if re-run after his season is finally over.
Tier 3: The rest: Dunn, Ellenson, Valentine, not separated by much. (Monte Morris, looking like a late-round sleeper PG pick, has pulled out of the draft and is going back to school).
His punch-line conclusion as to who the Celtics should pick at #3:
"There is no clear candidate I would take against the field to become a star, but given his youth and impressive FIBA play, not included in my model, I have to lean toward Bender as marginally the best upside candidate."
I concur, especially when you factor in that the model basically works against Bender by not including his most recent data, his FIBA performance and possible improvements in his physicals. And the fact that all of his contributing factors are positive is notable as well.
Deyonta Davis and Diamond Stone do extremely well in this model, held back only by their poor passing ratings.
Thoughts?