A New Technique Based on the Algorithm of Particle Swarm to Solve Prediction Problem in Stock Market and Portfolio Selection

Abstract

Over the last few decades, the average person's interest in the stock market has grown exponentially. This demand coupled with advances in trading technology has opened up the markets, so that nowadays anybody can own stocks, and use many types of software to perform the aspired profit with minimum risk. Consequently, a lot of attention has been devoted to the analysis and prediction of future values and trends of the financial markets, and due to large applications in different business transactions, stock marketprediction has become a hot topic of research.This thesis consists of six chapters. First chapter presents a theoreticalbackground of the computational intelligence techniques and theircharacteristics and applications. Fast review of different computationalintelligence paradigms such as artificial neural networks, evolutionarycomputation, swarm intelligence, artificial immune systems, fuzzysystems, bacterial foraging optimization, adaptive bacterial foragingoptimization, and genetic algorithm were presented. Second chapterreviews the particle swarm optimization algorithm in details with itscharacteristics and problems, it also introduce a description of improvedversions of Particle Swarm Optimization. The different modifications on PSOare divided into two main categories, internal and externalmodifications. Third Chapter presents the stock markets problem and reviews the previous work such as predictionmethods in that field. Also, some common indicators, which related to ourwork, were provided. fourth chapter a computationalintelligence techniques called Particle Swarm Optimization (PSO) isdeveloped using the physical principle “Center of Mass”, which Wedenoted by PSOCoM. This modification gives a new efficient searchtechnique. It gets benefit from this physical principle to move theparticles to the new best predicted position. Then the evaluation of theproposed technique is presented. Fifth chapter the proposed particle swarmoptimization with center of mass technique (PSOCoM) is used to develop anefficient forecasting model for various stock price indices. This technique appliedto the task of training an adaptive linear combiner to form a new stockmarket prediction method. The results of the experiments showed that theproposed techniques are better than that it the other PSO based models accordingto the prediction accuracy using a number of scalable and multi modalbenchmark test functions. The presented technique PSOCoMovercame theother compared versions of PSO algorithm in aspects of convergence rate,and scalability. The suggested prediction model based on PSOCoMtechnique was compared with other models such as standard PSO, Geneticalgorithm, Bacterial foraging optimization, and adaptive bacterial foragingoptimization. The experimental results show that the proposed algorithmis the best among other algorithms in terms of MSE and the accuracy ofprediction for some stock price indices (DJIA, S&P500 and(NASDAQ100).Whereas, the proposed forecasting model gives accurateprediction for short- and long-term prediction. In appendix A, the C# code of particle swarm optimization is proposed.