Invariant probability measures on holomorphic functions on the unit disk

We study probability measures on the space of holomorphic functions on the unit disk that are invariant under pre-composition by conformal automorphisms and give no mass to constant functions. We determine the slowest possible growth of functions supported by such measures. Let $M_f(R)$ denote the maximum modulus of $f$ on the hyperbolic ball of radius $R$ centered at the origin. We show that given a non-decreasing function $G:[1,\infty)\to(0,\infty)$, there exists an $\textrm{Aut}(\mathbb{D})$-invariant probability measure $μ$ on $\textrm{Hol}(\mathbb{D}, \mathbb{C}) \setminus \{\textrm{constant functions} \}$ such that $\log^+ M_f(R)=O_f(G(R))$ for $μ$ almost every $f$ if and only if $\int_1^\infty (R/G(R)) \, dR <\infty$. For the lower bound, we use a mass transport argument to show that almost every trajectory of the gradient flow of $\log |f|$ has finite energy. To establish sharpness, we construct holomorphic functions of slow growth on infinite Riemann surfaces.