Particle Swarm Optimization Based On Predatory Search For Portfolio Investment

In this paper, considering the merit and demerit of Particle Swarm Optimization,we propose a PSO based on predatory search strategy which can more effectivelyimprove convergence speed and precision by examining test function and also applieswell in portfolio optimization problems at last.Modern portfolio theory is the theory settling the basic tradeoffs betweeninvestment returns and risk avoidance. Every rational investor pursues the least risk inthe given proceeds, or the best interests in certain risk conditions.Analyzing real investment environment and practical characteristics in Chinasecurities market, this paper gives a portfolio model with full cost from applyingperspective. And, to avoid falling into local optimum easily and low searchingprecision which both exist in the standard particle swarm optimization, the PSO basedon feeding strategy is proposed and applied in portfolio model. In the improving PSO,predatory search strategy is used to control particle swarm searching space byadjusting the restriction level. Therefore we can balance the global search and localsearch.At last, we make a numerical simulation, collecting actual securities investmentdata and giving a solution to portfolio model by using the optimization the paperproposed. The numerical result suggests that PSO based on predatory search strategycan more effectively improve convergence speed and precision in comparison withstandard particle swarm optimization.