@nickpang, the twist might move slightly over time but could never get beyond either of the pulleys so it would be a stable system. The odds are it would remain halfway between both pulleys since that is the position of least torsion and resistance.
@Mechaniker’s link mentions putting the twist on the slack side of the pulleys instead of the tension side, which makes sense. Either way, the twist could not move to the other side.