Algorithmic optimization of cattle breed portfolios: application of the Markowitz-Freund model to breeding in Cˆote d’Ivoire

Abstract

Optimizing the cattle portfolio is a major challenge for the sustainability and competitiveness of livestockproduction in Côte d’Ivoire. This paper proposes a modelling approach inspired by the Markowitz-Freund model (Hardaker, Huirne, Anderson, and Lien 2004); applied to the optimal selection of cattle breeds. The aim is to determine the combination of breeds that maximizes overall yield while minimizing economic and biological risk, particularly in terms of susceptibility to tsetse fly. Using real data on five local breeds (N’Dama, Baoulé, Z´ebu, Métis, and Lagunaire), quadratic programming was implemented in IBM CPLEX to solve the risk-return trade-off problem, with constraints on the total proportion, diversification, and tsetse fly rate. The results show that a balanced allocation, dominated by the Baoulé and Lagunaire breeds, makes it possible to reconcile economic performance and biological robustness. This approach, at the crossroads of bioeconomic and computational intelligence, paves the way for optimized decision-making that can be adapted to African agro-economic realities.