METHODOLOGY FOR ASSESSING THE EFFECTIVENESS OF MODELS FOR PREDICTING THE BEHAVIOR OF FINANCIAL MARKETS

DOI: 10.31673/2409-7292.2025.011908

Abstract

Forecasting financial markets is of key importance due to its impact on risk management and investment decisionmaking. Accurate forecasts can help avoid significant financial losses and maximize profits. Traditional methods for forecasting
financial markets include time series models, such as the autoregressive integrated mean model, the mean model, and the
autoregressive mean model, which are based on statistical principles and use historical data to predict future values. However,
traditional models have their limitations, especially in cases of high volatility or economic crises, when the market is subject to
significant changes. The paper proposes a new method for analyzing models that takes into account risks and adaptability of
forecasting models using recurrent neural networks. It has the ability to be flexibly configured due to the use of weighting factors
in calculations, which allows you to adapt not only to changes in the market, but also to various analytical tools. The architecture
of the developed system assumes a modular approach, where each component of the method is implemented as a separate
module with clearly defined interaction interfaces. This allowed us to test the methodology on real historical data and confirm
its effectiveness compared to traditional approaches.
Keywords: neural networks, information model, financial analytics, decision making.

References
1. Rocca R. Interpreting R²: a Narrative Guide for the Perplexed [Електронний ресурс]. Towards Data
Science. – 2024. – Режим доступу до ресурсу: https://towardsdatascience.com/interpreting-r%C2%B2-a-narrativeguide-for-the-perplexed-086a9a69c1ec.
2. Демчик Я.М., Розен В.П. Оцінки похибки прогнозних моделей та прогнозів спожитої електричної
енергії на об’єктах енергетичного ринку. Енергетика: економіка, технології, екологія. 2019. № 4. — С. 69-78.
3. Kozyrkov C. Why is Mean Squared Error (MSE) So Popular? [Електронний ресурс]. Towards Data
Science. – 2022. – Режим доступу до ресурсу: https://towardsdatascience.com/why-is-mean-squared-error-mse-sopopular-4320d5f003e5.
4. Robert H. Shumway, Devid S. Stoffer. Time Series Analysis and Its Applications: With R Examples.
Springer. Springer Texts in Statistics. 2017. 562 p.
5. Бідюк П. І., Гуць Є. В., Гавриленко В. В., Рудоман Н. В. Прогнозування цін акцій з використанням
рекурентної нейронної мережі LSTM. Системи управління, навігації та зв'язку, 2021, 3(65). С. 64-68.
https://doi.org/ 10.26906/SUNZ.2021.3.064
6. Замрій І.В., Федоренко М. Л. Аналіз використання алгоритмів штучного інтелекту для глибокого
аналізу фінансових даних. Сучасний захист інформації, 3(59), 2024. С. 55–62. https://doi.org/10.31673/2409-
7292.2024.030005
7. Xinhui Li. Application of Neural Networks in Financial Time Series Forecasting Models. Journal of Function
Spaces 2022(3):1-9. https://doi.org/10.1155/2022/7817264
8. Kady Sako, Berthine Nyunga Mpinda, Paulo Canas Rodrigues. Neural Networks for Financial Time Series
Forecasting. Entropy 2022, 24(5), 657; https://doi.org/10.3390/e24050657
9. Пархоменко Б. М., Акименко А. М. Використання інформаційних моделей для прогнозування
поведінки фінансових показників. Технічні науки та технології. 2024. № 2(36). С. 173-180.
https://doi.org/10.25140/2411-5363-2024-2(36)-173-180
10. Олександра Манзій О., Сеник Ю., Пелех В., Сеник А., Андрейчук С. Використання нейронних мереж
для задач інвестиційного аналізу. Галицький економічний вісник, № 2 (87) 2024. С. 164-174.
11. Christopher Krauss, Xuan Anh Do, Nicolas Huck. Deep neural networks, gradient-boosted trees, random
forests: Statistical arbitrage on the S&P 500. European Journal of Operational Research. Volume 259, Issue 2, 2017, рр.
689-702. https://doi.org/10.1016/j.ejor.2016.10.031
12. David M. Q. Nelson; Adriano C. M. Pereira; Renato A. de Oliveira. Stock market's price movement
prediction with LSTM neural networks. 2017 International Joint Conference on Neural Networks (IJCNN). Anchorage,
AK, USA, 2017, pp. 1419-1426, https://doi.org/10.1109/IJCNN.2017.7966019.
13. Fischer T. Deep learning with long short-term memory networks for financial market predictions. European
Journal of Operational Research. Volume 270, Issue 2, 16 October 2018, рр. 654-669.
https://doi.org/10.1016/j.ejor.2017.11.054
14. Kelum Gajamannage, Yonggi Park, Dilhani I. Jayathilake. Real-time forecasting of time series in financial
markets using sequentially trained dual-LSTMs. Expert Systems with Applications. 2023, Volume 223, 119879,
https://doi.org/10.1016/j.eswa.2023.119879.