Comparative evaluation of technical and fundamental analysis models when predicting stock prices

Abstract

The article considers a comparative analysis of the effectiveness in use of ARIMA, ARCH, GARCH models, а multi-factor forecasting model, and a decision tree model. Model functionality can be evaluated on the practical examples presented in the article. The results of applying the Dickey-Fuller test according to various data to verify the presence of non- stationarity are obtained. Parametric arguments for the models under study are described. The initial data, the order of the study, the results and charts are presented. Using the R programming language, practical studies focused on the functionality of the technical and fundamental analysis models were carried out to obtain the forecast values of PJSC "Sberbank" stock rate. The software modeling process showed the strengths and weaknesses of each of the models considered. The best results were shown by the multi-factor model. The paper gives quantitative indicators of the forecast values. A comparative table of the statistical indicators showing the results of the forecast models is presented and the conclusions are drawn based on the suitability of their modeling. Current study was carried out to identify models of technical and fundamental analysis that give the most accurate forecast of the stock price with the possibility of further implementation in a computer program.