Trading Optimally Diversified Portfolios in Emerging Markets with Neuro-Particle Swarm Optimisation

摘要

In previous work the authors have developed trading models using both particle swarm optimisation and neural networks for specific emerging markets industry sectors. Here, a more flexible model is developed that is effective across a wide range of sectors. It is discovered there is a strong dependence of the quality of returns on the minimum number of trades allowed within a given time period (a risk-minimisation measure used to maintain portfolio diversity) and that in the case of emerging markets the optimal value for this parameter may be different to the standard investment industry recommendation. Learning is then extended to include this parameter, with out-of-sample testing demonstrating very promising results.