Predicting HCH using expert advice — LessWrong