ZhuLinsen/daily_stock_analysis
A configurable agent pipeline combines market data, news, LLM analysis, dashboards, and scheduled delivery. Useful as a reference architecture, but its free data sources carry reliability limits.
The project analyzes watchlists across **six stock markets plus ETFs**, combining price data, indicators, news, filings, fundamentals, and reports. It offers **15 built-in strategies**, a web dashboard, backtesting, portfolio tracking, and several notification channels.
Builders can study it as an end-to-end agent application: interchangeable model and search providers, prioritized data fallbacks, Web/Bot/API surfaces, and deployment through **GitHub Actions, Docker, local scheduling, or FastAPI**. The default workflow runs weekdays at **18:00 Beijing time**.
The project analyzes watchlists across **six stock markets plus ETFs**, combining price data, indicators, news, filings, fundamentals, and reports. It offers **15 built-in strategies**, a web dashboard, backtesting, portfolio tracking, and several notification channels. Builders can study it as an end-to-end agent application: interchangeable model and search providers, prioritized data fallbacks, Web/Bot/API surfaces, and deployment through **GitHub Actions, Docker, local scheduling, or FastAPI**. The default workflow runs weekdays at **18:00 Beijing time**. Free market sources may be rate-limited, change interfaces, or suffer network failures; steadier batch operation requires token-based providers. News quality depends on a configured search service, and the generated analysis is explicitly not investment advice.
This turns general agent-workflow ideas into a deployable finance research application with provider interchangeability, data fallbacks, multiple interfaces, scheduling, and backtesting. It also makes operational limits concrete: dependable output depends on external market and search providers, while broad feature coverage does not establish investment validity or source stability.