I did a little experimenting with Performance Profiling turned on. In my first stab at this, I used Generate Rows to create a row for each overall pick, parsed the draft status from the player, and joined on year/round/position in round. The Join was fairly expensive, so I tried to think of ways to reduce that, such as keeping only one record per year/round and joining on that.
With this tiny dataset, changing the join to year/round (cutting the records in the incoming draft dataset by ~95%) didn't make a significant difference, but could be something to consider with a larger dataset.