Properties of the reconciled distributions for Gaussian and count forecasts

Reconciliation enforces coherence between hierarchical forecasts, in order to satisfy a set of linear constraints. However, most works focus on the reconciliation of the point forecasts. We instead focus on probabilistic reconciliation and we analyze the properties of the reconciled distributions by considering reconciliation via conditioning. We provide a formal analysis of the variance of the reconciled distribution, treating separately the case of Gaussian forecasts and count forecasts. We also study the behavior of the reconciled upper mean in the case of 1-level hierarchies; also in this case we analyze separately the case of Gaussian forecasts and count forecasts. We then show experiments on the reconciliation of intermittent time series related to the count of extreme market events. The experiments confirm our theoretical results about the mean and variance of the reconciled distribution and show that reconciliation yields a major gain in forecasting accuracy compared to the base forecasts.