Analysis of Medium-Term Forecasting Methods for Processes with Structural Shifts in Financial and Commodity Markets

Abstract. Medium-term price forecasting methods for financial and commodity markets are considered. The forecasted processes are nonstationary and nonlinear; they contain structural shifts arising due to systemic changes in the market structure and extreme events affecting the market. The probability of structural shifts grows with increasing the forecasting horizon, so the forecasting problem should be solved considering possible changes in the process of interest on the forecasting horizon. To forecast future changes in the process, it is necessary to expand the information field of the forecast, i.e., include expert judgments and the results of qualitative analysis of the processes, e.g., using the methods of fundamental analysis, cognitive analysis, and their implementation algorithms. Medium-term price forecasting in financial and commodity markets is a necessary element in the planning and management of socio-economic and production systems and investment management. This survey addresses the peculiarities of the forecasted processes determining the requirements for medium-term forecasting methods, their implementation, and the selection of necessary information included in the forecast to detect future changes in the process and their causal factors. Models and methods of statistical forecasting, artificial intelligence, and fractal analysis are considered, in addition to those using information from various sources in the forecasting algorithm: expert judgments, news about extreme events, and search engine data. The results of this survey are summarized in the context of medium-term forecasting. Finally, some promising lines of research in this area are outlined.

Keywords: commodity and financial markets, time series, structural shifts, medium-term forecasting.

Acknowledgments. This work was supported in part by the Russian Science Foundation, project no. 23-21-00455; https://rscf.ru/project/23-21-00455/.