Open Source
Explore the latest AI open-source projects from GitHub and HuggingFace.
Explore the latest AI open-source projects from GitHub and HuggingFace.
## Introduction Daily Stock Analysis is an open-source, LLM-powered stock analysis platform that generates automated daily trading intelligence reports for A-shares, Hong Kong stocks, and US equities. With over 20,000 GitHub stars and explosive growth of 3,000+ stars in a single week, it has become one of the most popular AI-driven financial analysis tools in the open-source ecosystem. The project addresses a real pain point for retail investors: the overwhelming volume of market data, news, and technical signals that need to be synthesized into actionable decisions. Rather than building yet another charting tool, Daily Stock Analysis leverages multiple LLM providers to generate concise decision dashboards with specific entry points, risk warnings, and bullish catalysts. ## Architecture and Design The system combines multiple data sources with LLM reasoning to produce comprehensive analysis reports. Data flows through a pipeline that collects market data, news sentiment, and technical indicators, then passes the synthesized information to an LLM for natural language analysis. | Layer | Components | |-------|------------| | Market Data | AkShare, Tushare, YFinance, Pytdx, Baostock | | News Intelligence | Tavily, SerpAPI, Bocha, Brave, MiniMax | | LLM Providers | Gemini, OpenAI, Claude, DeepSeek, Qwen via LiteLLM | | Delivery Channels | WeChat Work, Feishu, Telegram, Email, Pushover | | Infrastructure | GitHub Actions, FastAPI, Docker | The architecture is deliberately modular. Market data providers, news sources, and LLM backends can all be swapped independently. The unified LiteLLM interface means switching from Gemini to Claude requires changing a single environment variable. ## Key Capabilities **Decision Dashboards**: Each analysis generates a structured report containing a one-sentence core conclusion, precise buy/sell entry points, an operation checklist with risk warnings, and bullish catalysts. This format is designed for quick scanning during trading hours. **11 Built-in Trading Strategies**: The agent strategy chat mode provides interactive analysis using moving average crossovers, Elliott Wave theory, market regime detection, support/resistance levels, and seven other quantitative strategies. Users can engage in multi-turn dialogue to explore different analytical perspectives. **Smart Import**: Vision AI extracts stock codes from screenshots, eliminating manual data entry. CSV and Excel uploads are also supported for batch analysis. **Zero-Cost Operation**: The recommended deployment uses GitHub Actions with free-tier LLM APIs from Gemini or DeepSeek, requiring no server infrastructure and no ongoing costs. **Multi-Market Coverage**: Full support for Shanghai and Shenzhen A-shares, Hong Kong stocks, and US equities including major indices. Each market uses appropriate data sources and regulatory context in its analysis. **Automated Scheduling**: GitHub Actions workflows run analysis on configurable schedules, delivering reports to preferred channels before market open. ## Developer Integration The fastest deployment path uses GitHub Actions: ```bash # Fork the repository # Configure secrets: LLM_API_KEY, TELEGRAM_BOT_TOKEN, etc. # Enable the scheduled workflow ``` For local development: ```bash git clone https://github.com/ZhuLinsen/daily_stock_analysis cd daily_stock_analysis pip install -r requirements.txt python main.py --stock AAPL --provider gemini ``` The built-in FastAPI dashboard provides a web interface for settings management, analysis history, and manual triggers. Docker deployment is available for teams preferring containerized infrastructure. ## Limitations This is an analysis tool, not a trading bot. It does not execute trades or connect to brokerage APIs. LLM-generated analysis can contain hallucinated data points or incorrect technical interpretations, and should never be the sole basis for trading decisions. Free-tier LLM APIs have rate limits that constrain the number of stocks that can be analyzed per day. The system's analysis quality depends heavily on the underlying LLM's financial knowledge, which varies significantly between providers. Real-time data feeds have inherent delays, making the tool better suited for daily swing analysis than intraday scalping. The project's documentation is primarily in Chinese, which may present a barrier for international users. ## Who Should Use This Daily Stock Analysis is ideal for retail investors who want automated morning briefings synthesizing overnight developments, technical signals, and news sentiment. Developers interested in building LLM-powered financial applications can use it as a reference architecture for combining market data with language model reasoning. Quantitative researchers exploring the intersection of LLMs and traditional technical analysis will find the 11 built-in strategies a useful starting point. Teams already using GitHub Actions for automation can add stock analysis to their workflow with minimal additional infrastructure.