x-pai/AlphaBot

AlphaBot – 智能股票分析助手

What it is – AlphaBot is an open‑source application that bundles a large‑language‑model (LLM)‑driven chat assistant with market‑data pipelines, visualisation, and simple quantitative‑trading helpers. It is aimed at retail investors who want AI‑generated analysis instead of relying only on gut feeling.

Key capabilities

• AlphaBot Agent – a conversational finance bot that can answer questions, run web‑searches for the latest news, and generate charts or reports from natural‑language prompts.
• Multi‑source market data – pulls real‑time quotes from Alpha Vantage (global markets) and from Tushare/AKShare for Chinese A‑shares.
• AI‑enhanced analysis – technical‑analysis patterns, fundamental‑metric comparisons, and a pre‑trained machine‑learning model that forecasts price movements. The LLM (via OpenAI API) adds narrative explanations.
• Trading aides – stock search, historical price view, common technical indicators, risk‑assessment scores, and suggestions of quant strategies.
• Web UI – a React front‑end (served on localhost:3000) with a Swagger‑documented FastAPI back‑end (localhost:8000/api/v1/docs).

How to run it

1. Docker (recommended) – clone the repo, then execute ./deploy.sh (or ./deploy.ps1 on Windows). Docker‑Compose builds the back‑end, front‑end, and a database container. The services become reachable at the URLs above.
2. Manual install –
   • Back‑end: cd backend && pip install -r requirements.txt && uvicorn app.main:app --reload (optionally run python train_model.py to retrain the ML predictor).
   • Front‑end: cd frontend && npm install && npm run dev.
   • Default admin login is admin/admin123.

Technology stack

• Back‑end – Python, FastAPI, Uvicorn, SQLAlchemy (or similar) for the API; uses OpenAI API for LLM calls and a custom ML model for price prediction.
• Front‑end – React (JavaScript/TypeScript) with a UI that embeds the chat agent and chart components.
• Data ingestion – Alpha Vantage, Tushare, AKShare APIs.
• Containerisation – Docker + Docker‑Compose for one‑click deployment.

Intended audience – Retail traders and hobbyist developers who want a ready‑made AI‑powered stock‑analysis platform that can be self‑hosted and extended.

Caveats – The README stresses a strong disclaimer: the tool does not guarantee profits, the AI can make mistakes, and data quality depends on third‑party providers. Use it for insight, not as an automated trading bot.

Community & support – The project provides a WeChat public account, a WeChat group, a QQ group, and a Bilibili tutorial video. Sponsorship is encouraged via WeChat QR codes.

Contribution – See CONTRIBUTING.md for branch policies, PR templates, and issue reporting guidelines.