Skip to Main Content
Recently quite much attention was given to the investigation of Particle Swarm Optimization algorithm (PSO). It was proved that PSO algorithm has exhibited good performance across wide range application problems. This paper proposes the use of PSO algorithm for decision making model updating. The decision making model is used to generate one-step forward investment decisions for stock markets. The artificial neural networks (ANN) are used to make the analysis of historical daily stock returns and to calculate one day forward possible profit, which could be get while following the model proposed decisions, concerning the purchase of the stocks. Subsequently the Particle Swarm Optimization (PSO) algorithm is applied in order to select the ldquoglobal bestrdquo ANNs for the future investment decisions and to adapt the weights of other networks towards the weights of the best network. Different from our previous works this paper presents a new variant of PSO algorithm where the clusters of particle are identified in the search space. Knowing clusters the centers of clusters are substitutes for the best particle. Also this paper introduces variation of regular PSO algorithm where decision and particle training is made based on the performance of worst particle. Experimental investigations have shown that average performance per a fixed number of iterations can be improved by substituting cluster centers for the individualpsilas best positions. Also experimental investigation on decision making using worst particle shows that better results than using regular PSO can be achieved.