Index fund optimization using a hybrid model: genetic algorithm and mixed-integer nonlinear programming

Díaz, José Antonio Acevedo; Cortés, María Teresa González; Hernández, Juan Ángel Agudelo; Clavijo, Óscar; Ardila, Carlos J.; Cabrales, Sergio

doi:10.1080/0013791x.2019.1633450

Cited by 11 publications

(1 citation statement)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, the strategy directly considers the trade-off between transaction costs and similarity in terms of normalized value development. Díaz et al (2019) propose a hybrid model for solving the multi-period index tracking problem, which includes limits on the number of stocks, floor and ceiling constraints, diversification by sector, and transaction costs. Their model combines a genetic algorithm, used to select stocks, and a mixed-integer nonlinear programming, to estimate the weights.…”

Section: Introductionmentioning

confidence: 99%

Polynomial goal programming and particle swarm optimization for enhanced indexation

Kaucic

Barbini²,

Verdù

2019

Soft Comput

View full text Add to dashboard Cite

Enhanced indexation is an investment strategy that aims to generate moderate and consistent excess returns with respect to a tracked benchmark index. In this work, we introduce an optimization approach where the risk of under-performing the benchmark is separated from the potential over-performance, and the Sharpe ratio measures the profitability of the active management. In addition, a cardinality constraint controls the number of active positions in the portfolio, while a turnover threshold limits the transaction costs. We adopt a polynomial goal programming approach to combine these objectives with the investor's preferences. An improved version of the particle swarm optimization algorithm with a novel constraint-handling mechanism is proposed to solve the optimization problem. A numerical example, where the Euro Stoxx 50 Index is used as the benchmark, shows that our method consistently produces larger returns, with reduced costs and risk exposition, than the standard indexing strategies over a 10-year backtesting period.

show abstract

Section: Introductionmentioning

confidence: 99%