Examinando por Materia "Valoración de activos"

Mostrando 1 - 3 de 3
  • No hay miniatura disponible
    Publicación
    Evaluation of investment alternatives in the Colombian bonds Market Using Machine Learning Algoritms
    (Universidad EAFIT, 2025) Gómez Plaza, Sergio Daniel; Cardona Llano, Juan Felipe
    The bonds market in Colombia, according to the Colombian Stock Exchange (BVC), represents the largest segment of the country's capital market, where fixed-income instruments account for 80% of the average daily trading volume on exchange systems. Sovereign and corporate bonds are the most sought-after by investors, as they offer stable returns and low risk. However, valuing these assets is not always straightforward due to volatility and economic fluctuations.This study explores how machine learning algorithms can enhance the valuation of these investment options in the Colombian market. Various approaches, such as neural networks and decision trees, will be tested to determine which best predicts the behavior of these assets. These models are expected to support better investment decision-making, particularly during periods of uncertainty. The goal is to leverage artificial intelligence to develop more effective and well-suited tools for an informed market.
  • No hay miniatura disponible
    Publicación
    Identificación no paramétrica de factores para el corte transversal de retornos de acciones latinoamericanas
    (Universidad EAFIT, 2022) Zuluaga Rendón, Simón; Agudelo Rueda, Diego Alonso
  • No hay miniatura disponible
    Publicación
    Maf Challenge - Retornos de equilibrio para multiactivos
    (Universidad EAFIT, 2025) Giraldo García, María Alejandra; Aristizábal Arango, Daniel; Giraldo Restrepo, María Alejandra; Mejía Escobar, Juan Camilo; Grajales Correa, Carlos Alexander; Durango Gutiérrez, María Patricia
    This study develops a quantitative model to estimate equilibrium returns in multi-asset portfolios by integrating the Arbitrage Pricing Theory (APT) with the machine learning algorithm Extreme Gradient Boosting (XGBoost). The model incorporates monetary, real, financial, and expectation-based macroeconomic factors—such as inflation, employment, monetary policy, liquidity, credit spreads, and consumer sentiment—classified as leading, coincident, or lagging. The inclusion of lag structures captures the gradual transmission of macroeconomic shocks, while XGBoost identifies nonlinear patterns and complex interactions among variables. This hybrid framework allows for the estimation of factor sensitivities (betas) and risk premia (λ), producing return estimates consistent with a multifactor equilibrium. The methodology is applied to liquidity instruments, sovereign bonds, high-yield credit, emerging market debt, REITs, and global equities. Results show that both macroeconomic fundamentals and investors’ expectations play a central role in determining expected returns, highlighting the relevance of combining economic theory with modern machine learning techniques.