graduate student
The article explores the possibilities of using Bayesian neural networks to model risks in financial markets. Financial markets are characterized by a high degree of uncertainty and complexity, in which traditional methods of risk analysis become ineffective. The author suggests using Bayesian neural networks, combining the flexibility of deep neural networks with the principles of Bayesian statistics, for more accurate risk analysis and informed decision-making in financial markets. The research materials and methods are based on data from publications of the Bank of Russia and the work of Russian scientists in the field of measurements based on Bayesian intelligent technologies. The results of the study show that Bayesian neural networks make it possible to effectively manage uncertainty in financial markets, given the complexity of the financial system, the lack of initial information and the competence of specialists. The uncertainty analysis process includes the stages of conceptualization, model development, information collection, quantification of uncertainty, and pooling of uncertainties. The authors emphasize that Bayesian neural networks have a number of advantages: constant updating of knowledge, the use of probability distributions and the assessment of the probability of scenarios. In conclusion, the author notes that Bayesian neural networks allow modeling risks in financial markets and assessing the level of uncertainty, being the basis for developing recommendations for making managerial decisions. The purpose of the study is to consider the possibility of using Bayesian neural networks to model risks in financial markets.
uncertainty analysis, Bayesian neural networks, risk modeling, forecasting, financial markets
1. Bobkov I.A., Burdina A.A., Nehrest-Bobkova A.A. Modelirovanie neopredelennosti pri pomoschi neyronnyh setey // Ars Administrandi (Iskusstvo upravleniya), 2023. – T. 15, № 1. – S. 45-59.
2. Smirnova L.N. Ocenka riskov innovacionnogo proekta s primeneniem teorii nechetkih mnozhestv i bayesovskih setey // Torgovlya, servis, industriya pitaniya, 2023. – № 3 (2). – S. 184-193.
3. Domanov A.O. Osnovy bayesovskogo podhoda k kolichestvennomu analizu (na primere evroskepticizma) // Polit. Nauka, 2021. – № 1. URL:https://cyberleninka.ru/article/n/osnovy-bayesovskogo-podhoda-k-kolichestvennomu-analizu-na-primere-evroskeptitsizma (data obrascheniya: 13.09.2024).
4. Prokopchina S.V. Svertochnyy podhod k integracii metodov iskusstvennogo intellekta i teorii izmereniy na osnove bayesovskih intellektual'nyh tehnologiy. Koncepciya bayesovskoy izmeritel'noy neyroseti. Koncepciya IIIoT – intellektual'nogo IIoT // Mezhdunarodnaya konferenciya po myagkim vychisleniyam i izmereniyam, 2021. – T. 1. – S. 3-8.
5. Korneev D.S. Ispol'zovanie apparatnyh neyronnyh setey dlya postroeniya modeley ocenok i upravleniya riskami na predpriyatii // UBS, 2007. – № 17. URL:https://cyberleninka.ru/article/n/ispolzovanie-apparata-neyronnyh-setey-dlya-sozdaniya-modeli-otsenki-i-upravleniya-riskami-predpriyatiya (data obrascheniya: 13.09.2024).
6. Toropova A.V. Podhody k diagnostike soglasovannosti dannyh v bayesovskih setyah doveriya // Tr. SPIIRAN, 2015. – vypusk 43. – S. 156-178.
7. Seleznev S., Habibullin R. Bystraya ocenka bayesovskih modeley prostranstva sostoyaniy s ispol'zovaniem simulyaciy // Seriya preprintov ob ekonomicheskih issledovaniyah, dekabr' 2022. – № 104. – S. 42.
