You're kind of there, but I think it's helpful to drop the notion of wave-particle duality you have since it's confusing you. Watch Feynman's videos about it (o

You're kind of there, but I think it's helpful to drop the notion of wave-particle duality you have since it's confusing you. Watch Feynman's videos about it (or just read his book called QED) if you want clarification, but I think it's best to for the most part think of the particles described in particle physics as particles with special rules and behaviors that wouldn't make sense if they were just tiny spheres, avoid thinking in terms of "waves", and learn about those special rules as you can. Anyway one way you can think of the direction is that it's on the most probable path based on the weighted sum of all possible (even absurd) paths through the local area, see https://en.wikipedia.org/wiki/Path_integral_formulation https://en.wikipedia.org/wiki/Path_integral_formulation. At the classical level, absurd paths cancel each other out, and in general the path a photon takes from A to B will be along the curve representing the least amount of time needed. (Hence the "bend" of light as it enters water since photons propagate through water at a different speed.) Whether one actually observes the photon going from A to B depends on that configuration's probability.