Investment portfolio analysis by using neural networks

Abstract

Purpose – the purpose of the article is to compare the formation of portfolios and to make predictions about how it will change. Research methodology – for analysis, optimization and predictions use the neural network models that are created using a neural recurrent long short-term memory cell architecture network and Markowitz’s modern portfolio theory Findings – this article compares the portfolios of IT field with different instruments and level of optimization. Research limitations – the main limit of the article is that only historical data is used. The real-time investment would check the performance of the portfolio creation methodology under uncertain conditions. Practical implications – the results of the article give opportunities for investors and speculators in the finance market by using neural networks for forming investment portfolios, as well as analysing and predicting their changes. Originality/Value – the growing high-tech use in financial markets changes our habits and our understanding of the surrounding world. The financial sphere has also had several changes, and it has undergone major changes that will change the approach to producing financial forecasts and analysis. Including Artificial Intelligence in these processes brings new innovative opportunities.