The most time-consuming part of daily stock trading isn’t watching the ticker—it’s synthesizing fragmented technical indicators, announcements, news sentiment, and capital flows into an actionable judgment. An open-source project with 33k stars has now automated this entire pipeline.
daily_stock_analysis, maintained by ZhuLinsen, is an LLM-powered smart analysis system for A-shares, Hong Kong stocks, and US equities. It automatically fetches multi-source data daily, calls large language models to generate a “decision dashboard,” and pushes it to WeCom, Feishu, Telegram, Discord, Slack, or email. The entire pipeline runs on GitHub Actions at zero server cost.
The Problem It Solves
Retail investors face several information gaps when making decisions:
- Market data and news are scattered across multiple platforms
- Technical analysis requires manual chart reading—K-lines, moving averages, MACD
- Fundamental data requires digging through financial reports and announcements
- Different markets have different trading mechanics (A-share limit-up rules, HK T+0, US pre/after-market)
This project stitches these gaps into an automated pipeline: data collection → multi-dimensional analysis → LLM-generated decision dashboard → push notification.
Core Capabilities
The system comprises several modules, each serving a distinct purpose:
| Module | Function |
|---|---|
| Decision Dashboard | One-line conclusion + score + buy/sell levels + risk alerts + action checklist |
| Multi-dimensional Analysis | Technicals, real-time quotes, chip distribution, news sentiment, announcements, capital flow, fundamentals |
| Market Strategy System | Built-in A-share review, US Regime analysis, moving average, Chan theory, Elliott wave, sentiment cycle |
| Market Review | Daily market overview, index performance, advance/decline statistics, sector strength |
| Agent Q&A | Multi-turn strategy Q&A with 11 built-in strategies |
| AI Backtesting | Post-hoc verification of historical analysis with directional accuracy and simulated returns |
Example push notification:
🎯 2026-02-08 Decision Dashboard
Analyzed 3 stocks | 🟢Buy:0 🟡Hold:2 🔴Sell:1
⚪ Zhongwu High-Tech (000657): Hold | Score 65 | Bullish
📰 Sentiment: Market关注its AI attributes and strong earnings growth
🚨 Risk: Major capital net sell-off of 363M CNY on Feb 5
✨ Catalyst: Positioned as core AI server HDI supplier
Tech Stack
The project supports a wide range of AI models and data sources without locking into any single provider:
AI Models: AIHubMix, Gemini, OpenAI-compatible, DeepSeek, Qwen, Claude, Ollama local models. Configure at least one to run.
Market Data: TickFlow, AkShare, Tushare, Pytdx, Baostock, YFinance, Longbridge—covering A/H/US markets.
News Search: Anspire, SerpAPI, Tavily, Bocha, Brave, MiniMax, SearXNG—significantly impacts sentiment analysis quality.
Social Sentiment: Stock Sentiment API (Reddit / X / Polymarket, US stocks only, optional).
Comparison with Alternatives
| Dimension | daily_stock_analysis | Traditional Trading Software | Paid Research Reports |
|---|---|---|---|
| Cost | Zero server cost (GitHub Actions) | Free or monthly fee | Hundreds to thousands/year |
| Customization | Watchlist + multi-channel push + custom strategies | Fixed features | Fixed coverage |
| Automation | Daily scheduled push | Manual checking required | Periodic publication |
| AI Analysis | LLM synthesizes multi-source data | Technical indicators only | Human analysis |
| Market Coverage | A/H/US + ETF | Usually single market | Limited |
The trade-off: LLM analysis accuracy depends on prompt quality and data source reliability—not suitable as a direct trading signal. GitHub Actions free tier has monthly limits; high-frequency analysis requires a self-hosted server.
Quick Start
The fastest way is via GitHub Actions, deployable in 5 minutes:
- Fork the repository and configure Secrets
- Configure AI model (at least one):
AIHUBMIX_KEY=yourKey # or GEMINI_API_KEY / ANTHROPIC_API_KEY / OPENAI_API_KEY - Configure notification channel (at least one):
WECHAT_WEBHOOK_URL=your_webhook_url # or FEISHU_WEBHOOK_URL / TELEGRAM_BOT_TOKEN + TELEGRAM_CHAT_ID - Configure watchlist (required):
STOCK_LIST=600519,hk00700,AAPL,TSLA - Enable Actions and trigger a manual test
By default, it runs automatically at 18:00 Beijing time on each trading day, skipping non-trading days.
Local setup is equally simple:
git clone https://github.com/ZhuLinsen/daily_stock_analysis.git
cd daily_stock_analysis
pip install -r requirements.txt
cp .env.example .env && vim .env
python main.py
Docker deployment and FastAPI server mode (python main.py --serve-only) are also supported.
Worth Watching
- 558 commits, 120 tags, highly active—latest commit was 13 hours ago
- Supports new models like Kimi K2.6; the team continuously expands compatibility
- Web workbench provides manual analysis, config management, backtesting, and portfolio management
- Agent Q&A supports multi-turn follow-up and custom strategy files
Limitations: LLM analysis is reference information, not investment advice; free models may be unstable during peak hours; A-share data source real-time quality is limited by free API constraints.
Sources
- GitHub repository: https://github.com/ZhuLinsen/daily_stock_analysis
- Full deployment guide: docs directory in the repository