Application of the Hierarchical Equal Risk Contribution Model with Latin American ADRS Aplicación del modelo contribución jerárquica de igual riesgo con ADR latinoamericanos

Abstract

The Hierarchical Equal Risk Contribution (HERC) approach, as proposed by Raffinot (2017, 2018), is introduced here. Similar to the model proposed by López de Prado (2016), it incorporates machine learning techniques for port­folio optimization, addressing certain limitations of the Mean-Variance model by Markowitz (1952). An application of the HERC model is conducted, consid­ering Single and Ward linkage methods for hierarchical clustering of a set of assets traded on the NYSE, with companies located in Latin American countries. The results indicate that, for this set of assets, the Ward clustering and hierarchy method is characterized by being intra-country, resulting in a more compact number of clusters compared to the Single clustering method. Additionally, it demonstrates better performance, lower volatility, and a higher Sharpe ratio.