Analysis of the logistics models predictive ability within the stock market
Автор: Marshalov D.P., Zaborovskaya O.V., Sharafanova E.E., Konnikov E.A.
Журнал: Известия Санкт-Петербургского государственного экономического университета @izvestia-spgeu
Рубрика: Финансовый сектор экономики
Статья в выпуске: 6-1 (144), 2023 года.
Бесплатный доступ
In this study, three logistic models are used to predict the profitability of stocks for the next day based on text information, the number of requests in social networks, historical data on profitability and the key rate with a delay of up to 2 days. The article presents the results of each model, emphasizing the influence of hyperparameters, regularization, on the accuracy of forecasting, and also draws conclusions about the relationship between current profitability and historical data. In addition, the results of this article are more consistent with previous studies in the scientific literature, confirming the possibility of forecasting the current profitability with past data. The integration of predictive modeling and a comprehensive literature review increases the depth and reliability of the conclusions drawn, providing investors andfinancial analysts to make informed decisions in difficult trading conditions on the stock market.
Forecasting, classification models, regularization, text mining, logistic regression, decision tree, random forests
Короткий адрес: https://sciup.org/148327776
IDR: 148327776