Trev,
On Peco points (and most others probably
) there is a very slight "step" at the point where the moveable blade makes contact with the fixed rail. Some wheel profiles (Older Arnold models and some Grafars are examples) have flanges that "hit" this "step" and either "split" the point blade or ride up over it. Either way it results in a derailment
This problem is often worse when the set of points is placed immediately after a long curve. You can reduce this "step" by chamfering the end of the blade with a very fine needle file but this makes the blade very thin at the end and easily damaged. What I tend to do is bend the very end of the blade slightly towards the fixed rail with a small pair of long-nosed pliers(see photo). This makes the blade "tuck into" the fixed rail better and therefore reduces the "step"
point2.JPG
This obviously won't address your derailment problems in the trailing direction. That has got me baffled
I can't really offer any advice for your coupling problems either. Adding extra weight might make coupling a little easier but I doubt it. If the couplings aren't coming together smoothly then the fault lies with the design. If you have spring-loaded couplings and they are "drooping" this can be improved by removing the springs and stretching them but the job is fiddly and I guarantee that sooner or later one of the springs will shoot over your shoulder never to be seen again
I sincerely hope that your modelling experiences take a turn for the better. You deserve some luck.
John.