Well, I can't tell you that because I haven't studied this myself fully yet.
But I made the following observation when developing my swingtrading portfolio long-only options system (ie. it trades many titles at the same time, keeping single positions upto 10 days):
If the SL per position is set to less than 10% or more than 15% (depending on HV) then the performance starts to degrade significantly.
So, that region seems to be an ideal area for SL for this environment. But as said I haven't studied it fully yet for my own case,
just made that important observation and took simply the midpoint at about 12.5% SL (depending on HV).
But of course it is obvious to see that my environment is a little bit different than what you and the other studies have tested.
And: SL in this sytem does not necessarily mean a real stop. At such an SL level some other substrategies get applied like scaling-in etc.
(HV=historical vola of the underlying)