Comment by esafak

Sure. My model is to predict a random date in the future, the window of which I can pick with an LLM.

How much better is it than my trivial estimator?