Data Based Stock Portfolio Construction Using Computational Intelligence

摘要

The stock market is everywhere in our lives and stocks are sold and bought daily. Many people believe that investing in stocks is one of the most profitable and easiest ways to make money. The lure of easy profit can be proven erroneous when starting to invest in stocks, as stock portfolio construction and management processes are laborious. Constructing and managing a portfolio is multi stage and multi criteria problem and many of the models proposed are based on supporting only one stage. Moreover, available online data may be confusing as there is no clear evidence of how to use and clarify it. Therefore, in this paper, we propose a full-scale model that will exploit open data and will support portfolio management during all stages using Computational Intelligence. Available fundamental data will be used to evaluate stocks using Genetic Algorithms. Open past data of stock prices will be used for stock forecasting using a Multi Layer Perceptron. Eventually, using all the results of precedent stages a portfolio optimization will be implemented using Genetic Algorithms.