In AWT most general time arrow is so called cosmological time arrow related to omnidirectional Universe expansion, which is manifestation of dispersive character of energy spreading. While the general understanding is, Universe is facing thermodynamical death, it's not true at all as such conclusion is observer dependent. It's observer, not a Universe, who suffers by entropic processes and entropy of Universe as a whole remains constant from exsintric perspective. Thermodynamical death of Universe is just a consequence of Simillia simillibus observatur principle.
We cannot neglect fact, one half of Universe evaporates and separates by antigravity (radiation pressure), while the second one agglomerates by gravity. In AWT boundary between insintric and exsintric observational perspective is divided by observer distance scale, which corresponds to wavelength of cosmic microwave background (CMB scale) at 1.73 cm. Above such scale thermodynamic time arrow for material object becomes reversed and driven by gravity. So what we are observing are two thermodynamical processes separated by CMB/human scale into exsintric and insitric perspective. Material objects, which are large then 1.73 cm tends to agglomerate in their gravity field into larger ones. This is essentially negentropic process, related to inverse time arrow, whereas object smaller then CMB photons are evaporating into radiation, whis is indeed common entropic process. For particles of energy the whole situation remains reciprocal: large photons are dissolving like tachyons in CMB, while smaller one are condensing into solitons, i.e. material particles. No process violating CPT symmetry was observed so far.
If we consider material particles as the only observable part of Universe, thermodynamical time arrow becomes dual for 3D space-time, so we can propose more general, cosmological time arrow, which is independent to entropy of Universe (which remains the same in this case), but it's defined by the combination of the above processes. At the moment, when we would observe separation of large objects while the smaller ones would condense, we could say, not just thermodynamical, but cosmological time arrow gets reversed, too. It corresponds the propagation of observable objects across space-time brane, composed of mutually interacting gradients of Aether foam density, so that entropic processes are always balanced by these negentropic ones. From macroscopic (the past) or microscopic (the future of space-time expansion) perspective Universe is behaving like randomly undulating Aether gas, where formation of density fluctuations balances their dissolution and Universe appear atemporal, albeit it's still full of random motion. Every time arrow observed is therefore a local effect only.

While thermodynamical time arrow appears broken above CMB scale for material objects, it's just an effect of inverse geometry, in which dispersion of information occurs. Thermodynamical time arrow still remains valid here due the dispersive nature of energy spreading. In AWT gravity is just thermalization as being observed from exsintric perspective of Le-Sage Aether model. All forces are of dispersive nature, which is behaving like shielding Duillier-LeSage force from dual perspective (dual force to gravity is pressure of radiation, i.e. the only force, which can defy gravity). But we are observing universe from both perspectives, so we shouldn't omit gravity when talking about entropy from general perspective. Note that near CMB scale inverse square law for gravity becomes violated due the cosmic microwave radiation and gravity becomes effectively repulsive force bellow this scale, because of prevailing pressure of CMB radiation. This manifests by violation of equivalence principle in 3D and by weak deceleration assigned to dark matter. Again, it's just an result of perspective inversion - as a finite size fluctuation of Aether we're observing the same Le Sage gravitation "from inside". From local perspective time arrow and sign of gravity are related mutually by trivial projective geometry of mutual interactions of density fluctuations via transversal waves.














