Traditional and Machine Learning-Based Methods for Financial Instrument Price Forecasting: A Theoretical Approach
MetadataShow full item record
Financial markets are one of the main components of the economy, and their growth and development is a crucial and significant factor in the world. Meanwhile, artificial intelligence is an exponentially developing field. The use of artificial intelligence in financial markets is a new and intensely developing phenomenon, requiring extensive research. The aim of this paper is to present a methodology of machine learning-based method effective application in financial instrument price forecasting in comparison with the traditional method, namely ARIMA. Consequently, the scientific literature on financial markets, traditional and machine learning-based methods were analyzed. Finally, a theoretical model for stock prices forecasting is presented. The main results of the analysis show the most extensively techniques applied in the stock markets and cash markets are methods of time series analysis, econometrics and machine learning. After analysis of methods of machine learning, it can be found most popular supervised learning algorithms are linear regression, decision trees/regression tree, random forest classification/ regression, and support vector machines. As a result of the research, a theoretical model for financial instrument price forecasting is presented and discussed. Based on the theoretical approach proposed several experiments will be performed using ARIMA and Support Vector Regression (SVR) methods.