This project is a full-stack AI-powered insider trading detection platform. It uses FinBERT for sentiment analysis and SentenceTransformer models for semantic correlation to detect suspicious trading activity in real-time.
Feel free to check out the live demo here!
-
Python FastAPI Backend: Provides NLP inference services using FinBERT for sentiment analysis and SentenceTransformers for semantic correlation.
-
Java Spring Boot Orchestrator: Manages real-time trade generation, coordinates AI analysis requests, and broadcasts results via WebSocket/STOMP for live streaming.
-
Next.js + TypeScript Frontend: A responsive dashboard for visualizing trades, risk scores, and correlation networks in real-time. Includes D3.js-powered network graphs.
-
SEC EDGAR Integration: Fetches real SEC filings (8-K, 10-K, Form 4) to correlate insider trades with material corporate events.
-
Real-time WebSocket Streaming: Live trade feed and high-risk alerts pushed to the dashboard instantly via STOMP over SockJS.