I also would never have thought a fine spinning top of yours would make such insinuating moves. Mistress von Karman comes to mind.
For this reason I don't use anymore ball tips for tops with a recessed tip, for a long time.
That means if we want o follow this path at all we need to check how much spin energy we can put into a bicycle wheel or a larger steel disk by a handle like a curling stone has? If we have that energy we can estimate if we can expect substantial longer spin times after some optimization?
If you already have decided what kind of top to use, for example, a bycicle wheel, and you want the longest possible spin with it, you simply start it at the highest possible RPM. Until you don't have structural resistance problems, or balance problems, or what else, the more energy you put in it, the longer it should spin.
But if you start your project deciding first how much energy you may have for your top, and you want the longest spin, the problem is reversed; for example, if you want to make a finger top to be spun with one single twirl of the fingers, you will have to find what is the optimal size/weight for that amount of energy. I can put about 0.5 - 0.7 joule of kinetic energy in a top, with a single twirl. It doesn't matter if the top is large or little, the little top will spin very rapidly, and the large one slowly, with a single twirl, but the amount of energy transferred is about the same. I am not able to calculate theoretically what is the optimal size/weight for a given amount of energy, but I have some experience with finger tops, and I know empirically what is approximately a best size/weight for a finger top.
I show you some data; they are the longest spins of three different tops, by a single twirl, and by multiple twirls:
TOP WEIGHT DIAMETER AMI LONGEST SPIN LONGEST SPIN
Nr. grams mm kg-m
2 single twirl multiple twirls
14 165 52 0.000067 39:52 46:15
23 298 60 0.000146 28:00 58:19
9 847 101 0.001560 3:34 41:25
The Nr. 14 has an optimal size/weight for single twirl spins, and it spins longer than the other two ones with a single twirl;
by multiple twirls it doesn't spin much longer because the top is little/light and it already has high RPM after a single twirl, it is difficult to accelerate it much more with additional twirls.
The Nr. 23 is a bigger/heavier top, optimized for multiple twirl spins.
The Nr. 9 is simply too big/heavy for to be spun with fingers; with a single twirl it spins very slowly and it topples down soon after;
by multiple twirls I can start it at a decent speed but the top is still over sized even for multiple twirls.
Anyway, using a motor to start the three tops with sufficient speed and energy, I am sure that the Nr. 9 would become the best of these three.
The optimal size/weight of a finger top changes a bit depending on some other parameters, like the density of the flywheel, the kind of the contact points, the design of the top, the position of the tip, (recessed or external)... I think it would be extremely complicated to calculate it theoretically. And, in any case, there is lacking of accurate data, as Jeremy says, about both the tip friction and the air drag.