[jakelazaroff]

This still doesn't make sense to me. If the return of inverse strategy S' were greater the loss of strategy S, couldn't you then guarantee a positive return by using both strategies simultaneously?