Selecting Forecastability Threshold

Background: The forecastability profile quantifies the maximum achievable reduction in expected log loss across all forecast horizons given an information set.

Question / Future Work: Determining the precise methodology for selecting the decision-relevant threshold $\varepsilon$ that defines when forecastability translates into a practically informative horizon set ($\mathcal{H}_{\varepsilon}$) remains an open practical question. The authors state the framework is valid for any positive threshold but do not prescribe a specific selection method based on modeling cost, error tolerance, or decision value.