Development of Sentiment Analysis Model for Fixed Income Markets in Russia

The study aims to enhance decision-making processes on fixed income markets in Russia by applying machine learning techniques to analyze market sentiment. This task presents significant challenges, primarily due to two factors: the linguistic complexity of the Russian language, compounded by its domain-specific financial lexicon, and the scarcity of labeled datasets required for training supervised models. To address these challenges, the study employs fine-tuned BERT model, leveraging its ability to process complex linguistic structures and domain-specific terminology. A series of experiments are conducted to optimize the model's performance, resulting in a tailored solution capable of capturing the unique characteristics of the Russian financial ecosystem. The proposed model demonstrates robust results in sentiment analysis, providing a pretty reliable tool to support decision-making in Russia's fixed income markets.