Most modern discussions of Bell's theorem take microscopic causality (the arrow of time) for granted, and raise serious doubts concerning realism and/or relativity. Alternatively, one may allow a weak form of backwards-in-time causation, by considering "causes" to have not only "effects" at later times but also "influences" at earlier times. These "influences" generate the correlations of quantum entanglement, but do not enable information to be transmitted to the past. Can one realize this scenario in a mathematical model? If macroscopic time-asymmetry is introduced by imposing initial conditions, such a model can not be deterministic. Stochastic Quantization (Parisi and Wu,1981) is a non-deterministic approach known to reproduce quantum field theory. Based on this, a search for models displaying quantum nonlocal correlations, while maintaining the principles of realism, relativity and macroscopic causality, is proposed.