Multi-Agent AI Stock Analyst

Overview

An AI-powered stock analysis platform that transforms raw market data into transparent BUY, HOLD or SELL recommendations. The final investment score is calculated by a deterministic Python scoring engine, while the language model explains the reasoning in plain English — keeping recommendations transparent, consistent and explainable.

The problem

Stock research usually requires separate tools for financial statements, technical charts, market news, valuation and portfolio analysis. Switching between platforms is slow and makes investment decisions difficult to validate.

How it works

A user enters a ticker and seven specialised agents analyse it independently before a final recommendation is produced:

• Supervisor Agent — validates the stock symbol and coordinates the workflow.
• Fundamentals Agent — revenue, earnings, debt, valuation ratios and overall financial health.
• Technical Analysis Agent — RSI, MACD, moving averages, support and resistance, using rule-based algorithms.
• News & Sentiment Agent — analyses recent financial news and evaluates market sentiment.
• Forecasting Agent — generates Bull, Base and Bear scenarios from historical market behaviour.
• Portfolio Analysis Agent — diversification, concentration risk and portfolio exposure.
• Portfolio Manager Agent — combines all results through a weighted scoring framework into the final recommendation, confidence score, SWOT and rationale.

Key engineering decisions

Scoring is deterministic rather than model-generated, so a recommendation can be audited and reproduced; the model's role is explanation, not arithmetic. Validation, retry and fallback mechanisms support reliable execution, and the full analysis is exported as a professional PDF investment report.