We introduce the topological spaces (L; 1 ) which are well-de…ned for any given subset of random variables L on any given non-additive probability space ( ; F; ). A (c; )-ball at X 2 L contains all random variables in L that are su¢ ciently close to X in the sense that any payo¤ di¤erences to X smaller than c > 0 happen with -probability greater than 1. We derive two main results concerning (c; )-balls. Firstly, all (c; )-balls must be open sets in (L; 1 ) whenever is continuous from below and dual-autocontinuous from above. In that case, convergence of sequences of random variables on (L; 1 ) is equivalent to convergence in non-additive probability measure with probability-one coincidence.Secondly, an open (c; )-ball cannot be a convex strict subset of L whenever L has a non-trivial local cone structure and ( ; F; ) is dual-nonatomic.