I am no fan of PER when comparing players but neither do I put win shares in some much higher regard when comparing players. The marginalization to make the sum for all players on one team equal the total wins on that team means players on bad teams will have deflated numbers and players on good teams will have inflated numbers, for the most part.
I thought the per36 numbers when comparing those players probably the best statistical comparison. Also, the eye test has to be passed. I think Sully and Butler played fairly closely in total overall performance last year with both players showing inconsistencies during the year. The per36 shows that so I would say they are pretty even.
I do, however, see more potential to grow as a player in Sully than the other players on Chicago.