daily_stock_analysis: LLM Stock Analysis System
/daily_stock_analysis
daily_stock_analysis is an LLM-powered multi-market stock analysis system that runs as a GitHub Actions cron job and pushes a daily buy/hold/sell dashboard to Telegram, Slack, Discord, or email. Reach for it if you track a short watchlist across A-shares, HK, US, Japan, Korea, or Taiwan stocks and want a repeatable AI read on sentiment and risk. Skip it if you need guaranteed-reliable market data — the README says its free defaults aren't.
What is the LLM-powered Stock Analysis System?
daily_stock_analysis is an LLM-powered multi-market stock analysis system that tracks a configurable watchlist across A-share, HK, US, Japan, Korea, and Taiwan stocks and ETFs. Each run pulls market data, technical indicators, news, and fundamentals from configurable sources like AkShare and SerpAPI, then has an LLM turn that into a scored decision dashboard with buy/sell points, risk alerts, and catalysts, delivered to chat apps or email.
Key Features for Market Intelligence
- ✓A daily AI decision report per stock, covering a core takeaway, a score, trend read, buy/sell points, risk alerts, catalysts, and an action checklist.
- ✓Multi-market data aggregation across A-share, HK, US, Japan, Korea, and Taiwan stocks and ETFs — quotes, candlesticks, technical indicators, news, announcements, and fundamentals.
- ✓A web and desktop workbench for manual analysis, task progress, historical reports, full Markdown output, backtesting, position tracking, and config management, in light or dark theme.
- ✓An Agent chat layer for multi-turn strategy Q&A across 15 built-in strategies — moving averages, Chan theory, wave theory, trend, hot topics, events, growth, and expectation, among others — over Web, bot, or API.
- ✓Smart import and autocomplete: pull a watchlist from an image, CSV/Excel, or the clipboard, with ticker/name/pinyin/alias completion.
- ✓Automation and push across GitHub Actions, Docker, local scheduled jobs, and a FastAPI service, with delivery to WeChat Work, Feishu, Telegram, Discord, Slack, or email.
- ✓Swappable AI models and data providers: Anspire, AIHubMix, Gemini, OpenAI-compatible endpoints, DeepSeek, Qwen, Claude, or local Ollama for the LLM; AkShare, Tushare, Baostock, YFinance, Longbridge, or TickFlow for market data.
- ✓An optional US-only social sentiment feed pulling Reddit, X, and Polymarket signals through the Stock Sentiment API.
Applying the System for Investment Insights
- •Running a free, server-free daily scan of a personal watchlist through GitHub Actions and getting the dashboard in Telegram or Slack each morning.
- •Deploying locally or in Docker with Tushare or Longbridge tokens configured for steadier data on longer-running or higher-volume tracking.
- •Asking the Agent chat follow-up questions against a specific strategy, like a moving-average crossover or Chan theory read, instead of only reading the fixed report.
- •Browsing historical reports and running backtests in the local web workbench before trusting a strategy.
- •Importing a watchlist from a screenshot, CSV, or the clipboard instead of typing tickers by hand.
Deployment Options: GitHub Actions, Local, Docker
daily_stock_analysis documents two deployment paths. GitHub Actions (recommended): fork the repo, then add repository secrets for at least one AI model key (ANSPIRE_API_KEYS or AIHUBMIX_KEY are the README's recommended options, with GEMINI_API_KEY, ANTHROPIC_API_KEY, or OPENAI_API_KEY as alternatives), at least one notification secret (WECHAT_WEBHOOK_URL, FEISHU_WEBHOOK_URL, TELEGRAM_BOT_TOKEN plus TELEGRAM_CHAT_ID, DISCORD_WEBHOOK_URL, SLACK_BOT_TOKEN plus SLACK_CHANNEL_ID, or EMAIL_SENDER plus EMAIL_PASSWORD), and the required STOCK_LIST secret. Enable Actions, then run the daily-analysis workflow manually to test; by default it runs on trading-day weekdays at 18:00 Beijing time. For local or Docker use: git clone the repo, run pip install -r requirements.txt, copy .env.example to .env and edit it, then run python main.py; Docker-specific steps live in the README's separate full deployment guide rather than the main README.
Strengths
- ✓Pushes to six notification surfaces out of the box — WeChat Work, Feishu, Telegram, Discord, Slack, and email — so the daily dashboard lands wherever the team already reads.
- ✓GitHub Actions deployment is genuinely server-free: fork, add secrets, enable Actions, and the workflow runs itself on a schedule.
- ✓Covers six markets (A-share, HK, US, Japan, Korea, Taiwan) plus ETFs in one watchlist instead of one region at a time.
- ✓The Agent chat layer ships 15 built-in strategies (moving averages, Chan theory, wave theory, trend, hot topics, events, growth, expectation, and more) for follow-up questions beyond the fixed daily report.
- ✓Model and data-source choice isn't locked in — swap between Gemini, OpenAI-compatible endpoints, DeepSeek, Qwen, Claude, or a local Ollama model, and between free and paid market-data providers.
Considerations for Data Sources and Stability
- △Default free market-data sources (AkShare, Baostock, YFinance) come with a stability disclaimer in the README itself — rate limits, interface changes, and network fluctuations aren't ruled out, and steadier data means adding a token-based source like Tushare, Longbridge, or TickFlow.
- △Multi-agent orchestration and custom strategy files for the Agent chat are explicitly labeled experimental in the README, not a finished feature.
- △The project's own disclaimer states it's for learning and research only and does not constitute investment advice — treat every score and buy/sell call as a starting point, not a signal to act on.
- △The primary README is written in Chinese first, with English and Traditional Chinese versions maintained as separate docs (docs/README_EN.md, docs/README_CHT.md) rather than being the default view.
Exploring Related Quantitative Finance Projects
Common Questions About the System
daily_stock_analysis covers A-share, Hong Kong, US, Japan, Korea, and Taiwan stocks, plus ETFs, in one watchlist; the README documents separate data-source coverage and capability limits per market in its market-support guide.
daily_stock_analysis works with Anspire, AIHubMix, Gemini, OpenAI-compatible endpoints (including DeepSeek and Qwen), Claude, and local Ollama models — the README recommends cloud APIs for GitHub Actions and Ollama for local or Docker deployments.
daily_stock_analysis is MIT-licensed and free to self-host, and GitHub Actions deployment itself costs nothing. You still need your own AI model API key; the README notes Anspire gives new users about 30 yuan of free credit, and the default free market-data sources work at no cost but aren't stability-guaranteed.
daily_stock_analysis pushes its decision dashboard to WeChat Work, Feishu, Telegram, Discord, Slack, or email, and also stores full Markdown reports viewable in its local web workbench at http://127.0.0.1:8000.
daily_stock_analysis's recommended path runs entirely on GitHub Actions, which the README describes as a 5-minute, zero-cost setup with no server required; it can also run locally or in Docker instead of GitHub Actions.
daily_stock_analysis's own README disclaimer says it's for learning and research only, does not constitute investment advice, and that the author isn't responsible for losses from using it.
The problem it solves
Retail investors juggling tickers across A-shares, HK, US, Japan, Korea, and Taiwan markets have to manually pull quotes, news, and announcements from separate sources every day and then judge sentiment and risk themselves — a slow, repeatable chore that most people skip until something moves. daily_stock_analysis automates that read: it aggregates the multi-market data, news, and fundamentals for a fixed watchlist and has an LLM generate a structured decision dashboard on a schedule, so the daily scan happens automatically instead of being done, or skipped, by hand.
Who should try it — and who should skip
Try daily_stock_analysis if you already keep a personal watchlist across A-share, HK, US, Japan, Korea, or Taiwan stocks and want that list re-analyzed and pushed to chat every trading day without pulling the data yourself. It also suits anyone comfortable configuring GitHub Actions secrets and picking their own AI/data/news providers, since nothing runs with zero configuration beyond the free defaults. Skip it if you want investment advice or automated trade execution — the README is explicit that this is a research and learning tool, not an advisor or a broker integration, and you'd have to build execution yourself.
Related repositories
Still deciding about daily_stock_analysis?
One click hands the question to an AI along with this page — see what it says about daily_stock_analysis.