Speaker
Гильермо Себастьян Тикона Пералес
Description
This work investigates a hybrid approach to financial forecasting that integrates sentiment analysis of investor-generated content with traditional technical indicators. Experimental results showed that incorporating sentiment data significantly improves predictive accuracy when combined with technical features. Two neural network models: RuBERT and FISHQA—are compared for their effectiveness in classifying sentiment within Russian-language financial texts. The study confirms that domain-specific sentiment analysis can enhance model interpretability and forecasting performance in financial analytics.
