I have a copy of "Desirable and Undesirable Characteristics of Offshore Yachts" that discusses this. You are right: people had previously thought of a capsize more like a static event with the large mass of ballast resisting, but found it is actually a dynamic event where distance and shape are important in addition to mass. Figure 31 in this book shows the contribution of the ballast, hull, and rig to resisting capsize in light and heavy yachts and the rig contributes about 60% and 70% respectively.
I took some related classes back in engineering school and am not sure about your weight aloft idea. Find a table of moment of inertia for different shapes and maybe plug in numbers to see what might be effective. For a "rod about it's end" like a mast, the formula is: I = (1/3)*M*L^2. The L squared tells you the length is the most important factor by far. Adding rigidly secured mass at the top, to move the effective M of that mast further out, thereby increasing L could add capsize resistance from an inertia perspective, but you sure wouldn't want it there when the boat is heeled far over and it's pulling it down further. So you'd want to calculate how much this idea would affect the overall roll resistance of the boat (i.e. Is it worth it? Is it possible to add enough weight to make a difference?) and how would it affect the boat's recovery from a capsize/knockdown (i.e. Is it counterproductive?) There are probably more issues.
