How an LLM‑Powered Open‑Source Tool Automates Multi‑Market Stock Analysis

1. A “Zero‑Cost” Full‑Stack Architecture

The project follows a straightforward design philosophy: zero cost, low barrier, full automation. It can be deployed with a single click using GitHub Actions, requiring no dedicated server—just fork the repository, configure a few API keys, and it runs.

Technically, it uses Python 3.10+ and aggregates market data from TickFlow, AkShare, Tushare, and YFinance. News and sentiment are fetched via SerpAPI, Tavily, and Brave. The AI layer supports Gemini, OpenAI, DeepSeek, Tongyi Qianwen, Claude, and local Ollama models. Results are pushed through enterprise WeChat, Feishu, Telegram, Discord, Slack, email, and other channels.

The system also includes a “strategy system” with eleven built‑in models such as Chan‑Lun, Elliott Wave, moving‑average crossovers, and sentiment cycles, offering diverse technical and fundamental perspectives rarely found together in open‑source tools.

2. What the AI‑Generated Decision Dashboard Looks Like

The core output is a structured Markdown report containing a one‑sentence summary, composite score, buy/sell signals, risk alerts, and a checklist. For each stock, the AI synthesizes technical indicators (MA, MACD, RSI), capital flow, news sentiment, and fundamentals (PE, ROE, revenue growth) into a 0‑100 rating with specific support and resistance levels and stop‑loss advice.

Example: for an A‑share, the system evaluates technical patterns, chip distribution, recent hot‑topic news, and financial metrics, then provides a concise actionable recommendation.

"For retail investors, the biggest problem is information overload, not lack of data. This project completes the final step from information to decision."

The tool also offers AI‑backtesting, allowing users to validate historical reports, view directional accuracy, and simulate returns to refine stock selections and strategy parameters.

3. 5‑Minute Setup: Zero‑Cost Deployment for Beginners

Deployment is reduced to three steps when using GitHub Actions:

Fork the repository, star it, then go to Settings → Secrets and variables → Actions to add the API key for the chosen LLM service (the guide recommends AIHubMix, which can call Gemini, GPT, Claude, DeepSeek, etc., without a VPN).

Configure your watchlist. The system supports code, name, pinyin, alias auto‑completion and bulk import from images, CSV/Excel, or clipboard.

Set up a scheduled run. GitHub Actions includes a cron scheduler; configure it to run each morning before market open, and the results are pushed directly to your device.

4. Final Thoughts

The rapid accumulation of 32,000 stars indicates that the "AI + investment" direction meets a strong demand among developers and retail investors. The tool does not aim to replace personal judgment; instead, it automates data collection and preliminary analysis so users can focus on final decisions.

While no AI tool can guarantee profit, for anyone spending more than 30 minutes daily on data gathering, a five‑minute deployment of this zero‑cost solution can save considerable time.