Development and Application of an Algorithm for Assessing Bond Similarity on the Moscow Exchange Based on G-spread Changes

This paper presents the development and validation of an innovative algorithm designed to identify bond analogs based on their static characteristics. The algorithm was developed to refine the assessment of the historical Value at Risk (VaR) on a 1-day horizon, using a two-year historical window. Utilizing a database of over 1000 corporate bonds denominated in rubles, each supported by at least 60 quotes over the past two years, the algorithm demonstrates the ability to identify bonds with comparable risk profiles. This opens up new possibilities for risk assessment, especially for new or infrequently traded bonds where traditional valuation methods prove ineffective. The study highlights the potential of the algorithm to radically transform risk assessment approaches in the bond market, emphasizing its importance for investors and financial analysts facing uncertainties with new or illiquid bonds.