I don't know of anything that's guaranteed to be effective out of the box. But as a rule, simpler is better. Any assumptions should be picked apart with a fine-tooth comb and tested for robustness. But in the end, there's not a whole lot that you can do: just cross your fingers and hope.

You could probably model something that uses sampled variance less aggressively

Something bayesian