Good call...
http://fivethirtyeight.com/features/projecting-the-top-50-players-in-the-2015-nba-draft-class/
Rozier = 0.7% chance star. 18% chance at being a starter. 50% chance he busts.
Hunter = 0.4% star. 12 % starter. 49% bust.
These guys effectively have less than 1% chance of being stars in this league. Sure, I have no problem packaging d-league talent for a shot at a real superstar. If I got word that embiid had a better than 50% chance of recovering fully, I'd trade these mediocre rooks gladly. We'll be lucky if any of them turn into rotation players.
All of this being based on statistics, which in turn are based entirely on draft position.
It's a statistical system that works fairly well if you are talking about selecting a random player at a random spot in a said range. When you are taking about an individual player with individual attributes, then it all falls apart.
For example, the stats would tell you that Sully had less than 20% chance of becoming an NBA starter - yet we all knew how skilled and talented he was from day one, hence why he was expected to go high lottery prior to news of the back issues.
That's the issue with this statistic - it's useful, but only if you use it in the right way.
The statistic is not designed to predict the the probability of a specific/known player becoming X/Y/Z good, in the manner you are trying to use it. To say "hey, John Doe was drafted at #11, so there is an X% change he's going to be a star" is not an effective use of the statistic.
The statistic is designed to predict the probability of you drafting a player who is Z/Y/Z good if you are drafting at position #A. For example, to say "we have the #16 pick this year, so history suggests we have a 50% chance of picking a bust. We should try to trade up!" is an effective use of the statistic.
For a scientific experiment to be considered conclusive, you need to ensure that there is only one variable - if an apple grows better when placed outside on a crowded street (rather than the fridge) you don't know if the reason why the apple grew better was:
1) Sunlight versus no sunlight
2) Outside temperature versus fridge temperature
3) Less moisture in the air outside vs in the fridge
4) Change in air pressure, etc
Once you know who the player is, the statistics (based on draft position) become to a large degree irrelevant, because you end up with too many variables such as:
* Motor - guys who don't give up on plays are more likely to earn rotation spots, despite skill level
* Work ethic - guys who gym rats are more likely to develop
* Physical - guys with exceptional physical gifts are more likely to become individually dominant
* Basketball IQ - guys with high IQ often become decent rotation players at the very least
* Medical history - many guys with star potential have had careers ruined by injuries
* Team fit - How well does the player compliment his new team-mates / coaches
* Team need - how needed at this player's skills, and hence how likely is he to get minutes?
So, to predict that Rozier has a 50% chance of busting based purely on his draft position, isn't an accurate / realistic use of the statistic.