Data-driven Particle Swarm Optimization Algorithm And Its Application In Portfolio Management

Swarm intelligence is a kind of intelligent behavior in which a group of multiple simple individuals accomplishes corresponding tasks through simple cooperation among themselves.Swarm intelligence algorithm has the ability of self-organizing and adaptive global optimization to a certain extent,and is widely used to solve various optimization problems.Particle Swarm Optimization(PSO)algorithm is a kind of swarm intelligence algorithm,which simulates the predatory behavior of birds.The particle swarm optimization algorithm model is easy to understand,has few parameters,and is easy to operate.However,the unique search mechanism leads to insufficient population diversity in the later stage,and has problems such as premature convergence.Solving the portfolio problem is to provide an optimal portfolio investment scheme,which is a complex nonlinear programming problem.The application of swarm intelligence algorithm in this area has advantages.This article introduces the related concepts of optimization problems,optimization algorithms,swarm intelligence algorithms,and portfolios.Inspired by the conjugate gradient method,this paper proposes a datadriven particle swarm optimization algorithm.The algorithm further utilizes the historical data of particle swarm to guide the particle flight by using the historical data as the driver and collaborating the corresponding information of the particle contemporary.By testing on different benchmark functions,the effectiveness of the proposed strategy is verified and the parameters are analyzed and tested.The proposed algorithm is further compared with other excellent variant particle swarm optimization algorithms on the function test sets.The results show that the proposed algorithm has the best comprehensive performance.Finally,the proposed algorithm is used to solve the portfolio optimization problem,and simulation experiments are conducted using the mean-variance model and the mean-VaR model as examples.The results show that the proposed algorithm has improved compared to the standard particle swarm optimization algorithm,which verifies the feasibility of the algorithm in portfolio problems and the applicability of practical problems.