I hung the top from a thread and the center of mass is indeed shifted towards the stem.
In this case it seems like the biconic top doesn't precess about its center of gravity.
I don't know the reason, but maybe I have an idea;
maybe the top is driven by the precession speed more than the center of mass wanting to stay in the center of revolution.
If so, I would expect a slower precession speed (and a larger circle described on the ground), when the top rolls on the ridge closer to the stem; this is because in this case the weight of the stem is in the sense of slowing down precession.
When the top rolls on the lower ridge instead, the weight of the stem should accelerate precession.
With an accelerated precession, the circle described on the ground would be littler, and the center of mass would become shifted towards the stem.
Different sizes of the circle described on the ground would then affect if and in what direction the hearts appear to move.
But it's just an idea, I am not sure ...
-----------------------------------------------------
Today I have tried using a full rod instead of a tube, and a tube less or more partially filled with clay at one extremity.
These objects seem always to precess consistently about the center of mass.
The kind of surface (glass, smooth stone, wood not varnished) seem not to influence the behaviour.
So, I would think that my explanation above is uncorrect.