MODELS FOR FORECASTING FINANCIAL RISKS IN A TRANSFORMED GLOBAL ENVIRONMENT

Abstract

The article examines the scientific and methodological foundations of financial risk forecasting in a transformed global environment. The relevance of the topic is determined by the growing instability of international financial markets, the persistence of high public debt, the expansion of non-bank financial intermediation, the development of private credit, the acceleration of digital transformation and the increasing use of artificial intelligence in financial institutions. These processes change the nature of risk transmission and reduce the reliability of forecasting approaches that rely only on historical averages, normal distribution assumptions or isolated analysis of individual institutions. The purpose of the article is to systematize the main groups of financial risk forecasting models and to substantiate an integrated architecture suitable for conditions of structural breaks, geopolitical shocks and data uncertainty. The study uses a systemic approach, comparative analysis, content analysis of reports of international financial institutions, generalization of econometric and financial literature, structural-functional modelling and tabular systematization. The article shows that traditional models, including discriminant analysis, volatility models, Value-at-Risk, stress testing and credit scoring, remain important elements of risk management, but they should be complemented by scenario analysis, systemic risk indicators, network models and machine learning tools. The main scientific result is the proposed hybrid architecture of forecasting, which combines market, credit, liquidity, sovereign, network, operational, technological and geopolitical risk blocks. The practical value of the results lies in the possibility of using this approach by financial institutions, regulators and public finance authorities for early warning systems, macroprudential monitoring, stress testing, capital planning and risk-based decision-making. The article emphasizes that the effectiveness of forecasting depends not only on model complexity, but also on data quality, interpretability, backtesting, governance procedures and the ability to update models when the global environment changes.