Daily Stock Analysis (DSA) analyses a watchlist of stocks with a large language model on each trading day and pushes a decision dashboard to messaging and email channels. It covers mainland China A-shares, Hong Kong, US, Japanese, Korean, and Taiwanese listings and common exchange-traded funds, and combines market data, news sentiment, technical indicators, and fundamentals into one report per holding plus a market review.
A report carries a core conclusion, a score, a trend read, buy and sell levels, risk alerts, catalysts, and an action checklist. A chat mode lets you ask follow-up questions about a single symbol with 15 built-in strategy views, including moving-average crossover, Chan theory, wave theory, trend, hot themes, event-driven, growth quality, and re-rating; sessions support multi-turn questioning, export, sending to a notification channel, and background execution.
Market data comes from several sources with fallback chains, and the defaults are free sources that run without configuration; token-based sources are recommended for scheduled or batch use. News and sentiment are gathered through search APIs, with an optional social-sentiment source for US listings. Notification channels include WeCom, Feishu, Telegram, Discord, Slack, and email.
Three run paths are documented: a scheduled GitHub Actions workflow on a fork of the repository, which runs on weekdays at 18:00 Beijing time and skips non-trading days; a local Python 3.10+ process with command-line flags for debug, dry-run, explicit symbols, market review, scheduling, and serving; and Docker. A local web workbench adds manual analysis, task progress, history, full Markdown reports, backtests, position management, imports from images or spreadsheets, and light and dark themes.
Features
- Decision reports: conclusion, score, trend, entry and exit levels, risk alerts, catalysts, and an action checklist per symbol
- Multi-market coverage: A-shares, Hong Kong, US, Japanese, Korean, and Taiwanese listings plus common ETFs
- Strategy chat: 15 built-in strategy views with multi-turn follow-ups over the web interface, a bot, or the API
- Data source fallbacks: several market-data providers with a documented priority order and free defaults
- News and sentiment: several search providers, plus an optional social-sentiment source for US listings
- Notification channels: WeCom, Feishu, Telegram, Discord, Slack, and email, with signature checks and Markdown-to-image options
- Scheduled runs: a GitHub Actions workflow with trading-day checks, forced runs, and resume support
- Web workbench: manual analysis, task progress, report history, backtests, positions, and configuration management
- Imports and completion: image, CSV, Excel, and clipboard import with code, name, pinyin, and alias completion
