Essays on probability, dependence, and modeling risk where standard methods break. Looking for tutorials and technical notes? Those live in Notes.
Supervised learning almost always predicts the mean. But the question that usually matters is the whole conditional distribution — what's possible, not just what's expected.
