Cassie, you might be interested to learn about conformal prediction. Unlike Bayesian approach it doesn’t have any assumptions, is model and parameter free, produces valid (unbiased) predictions regardless of the sample size https://github.com/valeman/awesome-conformal-prediction