All the problems of this dubious ( ) bend stem from the fact that it lacks the full symmetry of the Zeppelin Bend
This bend is produced using non-interlocked "b" and "d" loops and for this reason has the same symmetry as the Smith/Hunter's Bend (produced using interlocked "b" and "d" loops). This symmetry is such that R1( x, y, z ) = R2( -x, -y, z ) [ read as the section of the first rope located at point ( x, y, z ) is equivalent (twinned if you like) to the section of the second rope located at point ( -x, -y, z ) ]. The origin ( x, y, z ) = ( 0, 0, 0 ) can be taken near ( z = 0 is undefined ) the center of the bend and the z-axis is perpendicular to the plane defined by the standing parts and ends of the bend (approximately).
The aesthetically pleasing symmetry of the Zepplin Bend differs in that it has complete inversion symmetry about the center of the bend (origin), that is, R1( x, y, z ) = R2( -x, -y, -z ). However, as far as the influence of symmetry on the properties (for example, stability) of a bend, consider that the symmetry of the unstable Thief Knot is the same as that for the Zepplin Bend. As an interesting aside, a simple wrap and tuck of each of the working ends of the Thief Knot produces the Figure Eight Bend while maintaining this central inversion symmetry. In like fashion, a "b" and "q" loop which has no stability whatsoever indeed has complete central inversion symmetry prior to the simple wrap and tuck which produces the Zepplin Bend.
From this perspective, it would appear to me that symmetry alone does not explain the deficiencies of the bend in question. I might go so far as to speculate that symmetry alone cannot explain the deficiencies of any bend.
DDK