Integration of alternative data into interest rate forecasting models

The article considered the features of integrating alternative data into interest rate forecasting models. As a result of the conducted research, it was possible to identify the main challenges: the quality of alternative sources, the risk of overfitting with a high dimension of the feature space, the difficulty of reconciling the different frequency of data updates, and the requirements for the explainability of complex models. The approach proposed in this paper, based on the analysis of other studies, demonstrates the potential for expanding the information field of yield curve models through non-traditional sources and AI techniques, and also defines the directions for further research in the field of transparency and reliability of forecasting systems in macro–financial analysis. The information reflected in the work will be of interest to other academic researchers in the field of econometrics and financial mathematics who are developing and testing new methodologies for forecasting interest rates using alternative sources of information (socio-economic indicators, transactional data, signals). In addition, the results obtained will be in demand by central bank specialists and institutional investors seeking to improve the accuracy of risk management and portfolio investment strategies through the introduction of integrated models capable of taking into account high-frequency and "unofficial" market signals.