A Multi-objective Model for Portfolio Selection with Budget Constraints and a Deep Learning-based for Stock Price Prediction

Kakaei, Serveh

doi:10.22034/jaisis.2023.377702.1059

A Multi-objective Model for Portfolio Selection with Budget Constraints and a Deep Learning-based for Stock Price Prediction

Document Type : Original Article

Author

Serveh Kakaei

Department of Industrial Engineering, University of Kurdistan

10.22034/jaisis.2023.377702.1059

Abstract

The stock market is a key pivot in every growing and thriving economy, and every investment in the market is aimed at maximizing profit and minimizing associated risk. For this purpose, a multi-objective optimization model with the goals of reducing risk, and increasing profit are provided considering the budget constraint. The presented model helps investment companies to make more beneficial stock portfolios. As well as in this paper, a deep learning model Long Short-Term Memory (LSTM) is used to predict all days of the next year’s stock price according to historical data. The dataset consists of the TSE index traded in Tehran Stock Exchange financial market. First, among the 50 active industries in the TSE stock market, 10 highly profitable industries are selected. The stock price of the companies in these industries is predicted according to the data of the last 15 years. The results obtained from 170 stocks and 10 industries show that the automobile, investment, and pharmaceutical materials and products industries were the best industries for investment in 2023 in the Tehran stock exchange.

Keywords

Multi-objective optimization

Long short-term memory neural network

Stock market prediction

Portfolio selection

Subjects