Real-Time Financial Advisory System: A GenAI-Driven Conversational and Explainable Framework

N. Saranya Reddy; K. Siva Ganesh; K. Sravani; K. Jaya Ram; K. Komali

doi:10.62643/ijerst.2026.v22.n1(2).pp202-208

Authors

N. Saranya Reddy Author
K. Siva Ganesh Author
K. Sravani Author
K. Jaya Ram Author
K. Komali Author

DOI:

https://doi.org/10.62643/ijerst.2026.v22.n1(2).pp202-208

Keywords:

GenAI; financial advisory; NLP-SQL; LangGraph; sentiment analysis; portfolio optimization; explainable AI; risk assessment; conversational interfaces.

Abstract

Retail investors today face information overload from fragmented financial data without accessible, explainable decision support systems. This paper presents a Real-Time Financial Advisory System that leverages Generative AI to deliver conversational, explainable financial analysis through natural language interfaces. The system processes a comprehensive 51-stock universe comprising top S&P 500 constituents plus INFY, analyzing 1.89M historical price records and 845MB of financial news data spanning 2022–2024. A PostgreSQL database with 11 interconnected tables supports sophisticated financial modeling, while FinBERT-based sentiment analysis achieves 94% accuracy on 6,000+ financial headlines. The core innovation employs a LangGraph-orchestrated NLP-SQL pipeline with multi-strategy reasoning capabilities, converting natural language queries into optimized database operations. Portfolio optimization using Markowitz Modern Portfolio Theory yields Sharpe ratios of 0.424, validated through Monte Carlo stress testing with 10,000 simulation paths. A Streamlit web interface delivers real-time conversational interaction with transparent AI insights, risk assessment, and investment recommendations.