De novo caller clamps allele balance at 0.20

I’ve been trying to loosen the de novo caller criteria to include variants that are not present in the parents, but can be present in the child at an allele balance < 0.20. However, using the following command (or other iterations of it), it seems that the allele balance is somewhat clamped at 0.20.

de_novo_asc = hl.de_novo(mt, pedigree, pop_frequency_prior=mt.raw_AF_e, min_child_ab=0.000001, min_dp_ratio=0.00001, max_parent_ab = 0.001)

The distribution of the allele balances are as follows:

Is this the expected behavior? From previous studies we would expect more weight on variants between the 0.20 and 0+ range.