Agentic AI Stock Trading Automation

Please create an agentic AI system that can decide when to buy or sell equities by fusing four live data streams at once: social-media sentiment, breaking news, broad market trends, and classic price/volume patterns. The model’s logic, feature engineering, and orchestration are up to you so long as the final engine can run continuously, update its signals in real time, and learn from new information without manual prompts. Execution, not just signalling, is essential. Every recommendation must translate immediately into a trade on a paper-trading account (I will supply the API key). Each order should include the symbol, side, size, price, timestamp, and a short AI-generated justification so that later performance analysis is straightforward. I would like a lightweight interface—CLI, web dashboard, or notebook view—showing current holdings, open orders, realised/unrealised P&L, and a rolling log of the AI’s most recent reasoning. Feel free to leverage Python, LangChain, Hugging Face, sentiment APIs, or other libraries you prefer, provided the stack is reproducible and well-documented. Deliverables • Fully commented source code and requirements.txt/poetry.lock • Deployment guide that lets me start the bot on my own VPS in one command • Live demo on the supplied paper account, running for at least one market session • Back-test report that covers the last six months so I can benchmark live performance I’ll review the build once the demo session ends and sign off after the back-test and live log match the described behaviour.