You've got a ball at rest 500 meters up, and you drop it. It falls downwards at 9.8 m/s^2. After 10 seconds, it's traveling 98 m/s towards the ground and has falled 490 meters (10 meters above the ground).
Now freeze time and reverse it!
What do you see if you reverse the "movie"? Well, the ball changes direction. It's still 10 meters above the ground, but is now heading upwards at 98 m/s. It continues up, but slows down at -9.8 m/s until it finally comes to a stop 500 meters up.
In other words, running the movie backwards looks like just like the normal laws of physics -- attractive gravity and all. Weird, huh?
If you place a ball on an inclined plane then it rolls downwards to a position of lower potential energy, (under a Newtonian consideration) the gravitational force acting to accelerate the ball.
If you reverse time, unless you know how gravity acts in reverse then how do you know that the ball will roll up the hill rather than stay in place or roll down.
You're saying gravity remains attractive, ergo that if you reverse time the ball still rolls down, the universe still expands, etc..
Interestingly, if we ignore only the point at centre, it can be shown that an object that falls in a black hole, goes right through it and is ejected at the other side. All in finite proper time. This situation is non-physical because it takes an infinite amount of time for the photon to reach the event horizon. Even if we could reverse the photon inside the black hole, it would still take an infinite amount of time to crawl back out. So even in theory, black holes are still inescapable.
Gravity affects light, so particles going backwards in time wouldn't suddenly experience gravity as a repulsive force. The curvature of spacetime is what matters. (Light experiences zero subjective time.)
The strong interaction affects protons and neutrons, not electrons or positrons. "Zero effect" is implicitly time-invariant.
As to the weak interaction, it is mainly a (reversible) "given inputs, get outputs" process. Both β+ decay and electron capture have an analogous process for positrons. The catch is that when you swap signs on the electron, you have to make corresponding changes to the rest of the equation - e.g. if there's a proton for the electron version, it needs to be an antiproton to work for a positron.