Beats me if it’d look any better, but the tangent vector falls right out of the math.
Since we have equations in the form
Xpos = X + cos(A * t + B ) * XScale
Ypos = Y + sin(C * t + D) * YScale
(where everything but time is a constant, and B and D may even be 0), some basic calculus gives us
Xtan = -sin(A * t + B ) * XScale * A
Ytan = cos(C * t + D) * YScale * C
The tangent would be going “along” the curve. This may be fine, or maybe we’d want something facing “out”, 90 degrees away. That’s just a matter of switching the components and negating one of them, so either (-Ytan, Xtan) or (Ytan, -Xtan), then just doing the same atan2 () / deg () treatment on whichever was chosen.