A rather popular belief on this topic is that h-pol is superior to v-pol. Probably that is traceable to published "far-field" patterns over a lossy ground plane generated by Method of Moments software such as NEC (Numerical Electromagnetics Code). This post is a deeper dive into the topic.
As clearly shown in the graphic clip first below, h-pol e-m waves radiated at low elevation angles tend to be canceled by their reflections from the Earth near the antenna site. Note the last sentence in the text column at the right side of that graphic. (This clip is taken from a textbook printed in the 1940s, but Physics hasn't changed in the interim.)
The graphic second below shows a NEC4.2 comparison of the E-fields existing over a zero to 45° elevation sector, 0.1 km downrange from an h-pol and v-pol radiator both radiating 100 watts. (Note that an h-distance of 0.1 km from these radiators lies well within their far field.)
The reduced fields from the h-pol radiator are clearly evident.
I think frequency band and application need to be considered as a context.
Satellite signals get randomly repolarized as they travel through the ionosphere, so I don't believe there is any advantage to one orthogonal polarization over another (e.g. horizontal vs. vertical, right-hand circular vs. left-hand circular, etc.)
UHF and higher radio waves in urban areas are subject to extreme multipath scattering that also tends to randomize polarization.