An instructive example: polynomials $x(t), y(t)$ of degrees $8, 12$ such that $x^3 - y^2$ is a nonzero polynomial of degree at most $6$ (and at least $5$ by the usual ABC/Wronskian/Riemann-Hurwitz argument). These correspond to rational functions $f = x^3/y^2$ of degree $24$ that give rational maps $f : {\bf P}^1 \to {\bf P^1}$ branched over $4$ points, with monodromy generators of type $3^8$ above $0,$ type $2^{12}$ above $\infty,$ and type $1^6 18$ above $1,$ plus a simple transposition above the image of the extra zero of $f'.$ On the elliptic-surface side, we get elliptic K3 surfaces $Y^2 = X^3 - 3 x X + 2 y$ with a singular fiber of type at least $I_{18}$ at $t=\infty,$ so the moduli space has dimension $20-\rho = 20 - (2+17) = 1$ which (as usual for K3’s) is consistent with the parameter count.
Start by scaling and translation of $t$ to put $x$ in the form $$ x = (t^4 + a_2 t^2 + a_3 t + a_4)^2 + 2 (r_5 t^3 + r_6 t^2 + r_7 t + r_8 $$ (this corresponds to what we called “$x = \zeta^2 + \eta$” in the setting of Hall’s conjecture). Then expand $x^{3/2}$ in a Laurent series about $t=\infty,$ getting $$ x^{3/2} = t^{12} + 3 a_2 t^{10} + 3 a_3 t^9 + 3 (a_4 + a_2^2) t^8 + (3 r_5 + 6 a_2 a_3) t^7 + (3 r_6 + 6 a_2 a_4 + 3 a_3^2 + a_2^3) t^6 + \cdots = \sum_{n=0}^\infty c_n t^{12-n} $$ for some polynomials $c_n$ $(n=0,1,2,3,\ldots)$ of weight $n$ in the $a_i$ and $r_j.$ We set $y = \sum_{n=0}^{12} c_n t^{12-n},$ and require $c_n=0$ for $n=13,14,15,16,17.$ The weights suggest that only the first of these will be linear in any of our parameters $a_i,r_j$, and indeed $c_{14}$ already includes a term $3r_j^2/2;$ but thanks to the special structure of this cube-minus-square system the equations $c_{13} = c_{14} = c_{15} = 0$ are simultaneously linear in $(a_3,a_4,r_8),$ so we solve them and clear denominators to get two equations in the remaining parameters $a_2,r_5,r_6,r_7.$
At this point we might expect to have a complicated pair of equations
whose intersection gives a moduli curve that has many singularities
and/or a high genus. But in fact we get the union of two simpler curves,
and this is already visible in $c_{17},$ which is now just
$r_5 (2 a_2 r_5^2 - 3 r_6^2 - 3 r_5 r_7).$ Consider first $r_5 = 0.$
Since $r_5$ occurs in the denominators of the formulas for $a_3,a_4,r_8$
we must go back to the original equations $c_n=0,$ but fortunately
this is easy. First $c_{13}$ simplifies to $3 r_6 r_7,$
and if $r_6=0$ then $c_{14}=3 r_7^2/2$ so $r_7 = 0$ anyway.
Then $c_{14} = 3 r_6 (2 r_8 - a_2 r_6) / 2,$ and if we again take $r_6=0$
then $c_{16} = 3 r_8^2/2$ which means that the $r_j$ all vanish
and we are in the degenerate