Objective of the analysis is to represent a new approach to estimate loss given default (Lgd) alternative to the existing estimation models, improving performance compliant in terms of regulation. The model proposed incorporates the empirically observed bimodality of the historical distribution of losses
