TopGit
GitHub Repo Review

PanWatch: AI Stock Monitoring

TNT-Likely/PanWatch
PTopGit review image for TNT-Likely/PanWatch
Review by Topgit.dev for TNT-Likely/PanWatch, with GitHub repository stats and README context.
Quick verdict

PanWatch is a useful self-hosted AI stock monitoring application for investors who want portfolio tracking, scheduled research, and trading alerts in one browser workspace. Its multi-agent TradingAgents workflow is more involved than a basic quote dashboard, but the Docker setup and practical paper-trading tools make it worth trying for people who want a private deployment.

Stars
★ 2.5k
Forks
⑂ 456
Contributors
👥 6
Language
Python
License
MIT
Topic
—
Updated
Nov 2026
Homepage
GitHub

Self-hosted AI stock monitoring

PanWatch is a self-hosted application for portfolio insights, AI stock analysis, scheduled monitoring, and price alerts across A-share, Hong Kong, and US markets. Its TradingAgents workflow assigns technical, sentiment, news, and fundamental research to analyst roles before they hold a bull/bear debate, perform a risk review, and reach a portfolio-manager decision. Users can run models through an OpenAI-compatible provider or local Ollama models.

Portfolios, agents, alerts, and paper trading

  • ✓Track multiple brokerage accounts, holdings, profit and loss, watchlists, quotes, candlesticks, and declared trading styles through a shared portfolio interface.
  • ✓Launch a TradingAgents deep analysis from a holding, with four analyst roles feeding a bull/bear discussion, risk review, final portfolio-manager memo, and inspectable reasoning trail.
  • ✓Schedule pre-market outlooks, intraday monitoring, and daily closing workflows, with exchange holidays and trading sessions filtering automatic runs.
  • ✓Build stock trading alerts from price, percentage, turnover, volume-ratio, and other conditions joined with AND/OR logic, cooldowns, daily limits, expiration, and selected delivery channels.
  • ✓Review ranked opportunities with proposed entry levels, targets, and risk context, then simulate signal-driven entries and exits to inspect equity and performance curves.
  • ✓Deliver reports and alerts via Telegram, WeCom, DingTalk, Feishu, Bark, or webhooks, and install the web interface on a phone as an installable PWA.
How this repository's GitHub stars have grown over time. Source: star-history.com.View the star history ↗

Monitoring stocks across three markets

  • •Maintain a self-hosted portfolio analysis app for holdings spread across mainland China, Hong Kong, and US brokerage accounts.
  • •Perform A-share stock monitoring with technical, sentiment, news, and fundamental inputs assembled into one decision memo.
  • •Automate pre-market, intraday, and end-of-day monitoring on eligible exchange trading days.
  • •Evaluate stock trading alerts and paper-trading signals without placing orders through a live brokerage account.

Run PanWatch with Docker

Run PanWatch with Docker by publishing the image with port 8000 mapped to the host and a persistent volume for `/app/data`. Set `TZ` explicitly because the container does not detect the host timezone. The documented command is `docker run -d --name panwatch --restart unless-stopped -e TZ=Asia/Shanghai -p 8000:8000 -v panwatch_data:/app/data sunxiao0721/panwatch:latest`. Open `http://localhost:8000`, create login credentials, and follow the guided setup. A Docker Compose file is also provided, while local development uses Python 3.11+, Node.js 24.14.0, and pnpm 9.15.9.

Why PanWatch is useful

  • ✓Covers the full monitoring loop in one application: portfolios, market data, AI research, alerts, reports, and simulated execution.
  • ✓Supports A-share, Hong Kong, and US markets while accounting for local calendars, settlement differences, holidays, and daylight-saving changes in paper execution.
  • ✓Shows the reasoning trail behind agent decisions instead of presenting an unexplained buy or sell conclusion.
  • ✓Offers several practical notification destinations, including Chinese workplace tools, Telegram, Bark, and generic webhooks.
  • ✓The Docker image and persistent volume make private deployment straightforward, while the PWA avoids a separate native mobile build.

Data, calendar, and simulation gaps

  • △Structured ETF holdings, NAV, fees, and tracking error are not supplied, although ETF deep analysis does use fund-specific context.
  • △Market-calendar data is bundled for published 2026 closures and half-days; an unpublished weekday remains pending and blocks automatic execution until the calendar is updated.
  • △The paper simulator does not model intraday clearing batches, withdrawal cutoffs, margin, foreign exchange, broker-specific facilities, or ETF-specific T+0 exceptions and fee schedules.
  • △Historical paper trades are not retroactively locked when settlement rules change, and existing balances are treated as settled opening funds.
  • △ETF research does not substitute company financial statements for fund-level disclosures, so investors should not treat the two forms of analysis as equivalent.
View on GitHub ↗Homepage ↗

Manual workflows and simpler monitoring tools

TradingView — charting, watchlists, and price alerts for people who do not need PanWatch’s portfolio-centered TradingAgents research and paper execution.Composer Trade — a browser-based strategy and trading-ideas platform for users focused on published trade plans rather than PanWatch’s multi-agent review workflow.Trade-Ideas — AI-generated stock ideas and alerts for investors who prefer a dedicated idea feed over PanWatch’s locally managed portfolio workspace.

Frequently asked questions

Is PanWatch free and open source?

PanWatch is free and open source under the MIT license. You can run the self-hosted application yourself and connect it to an OpenAI-compatible service or local models through Ollama, though model usage may incur provider quota or cost.

Which stock markets does PanWatch support?

PanWatch supports A-shares, Hong Kong, and US stocks. Search by code or name works across mainland China, Hong Kong, and US instruments, including listed ETFs, while exchange calendars and paper-trading rules follow each market’s local conditions.

Can PanWatch use local AI models through Ollama?

PanWatch can use local models through Ollama, as well as preferred OpenAI-compatible providers. Configure the service under `Settings → AI Services`, choose a default model, and test the connection before generating an analysis because that test consumes model quota.

How do I install and run PanWatch with Docker?

Run the documented `docker run` command with port 8000 mapped, a persistent `/app/data` volume, and an explicit `TZ`. Then open `http://localhost:8000`, create credentials, and complete the guided AI setup. Docker Compose configuration is also included.

Can I install PanWatch on my phone?

PanWatch can be installed on a phone as a PWA from its mobile browser, allowing it to appear on the home screen. It remains a browser-based installation rather than a separately distributed native application, and uses the same workspace as the desktop deployment.

Does PanWatch provide structured ETF holdings and NAV data?

PanWatch does not currently supply structured ETF holdings, NAV, fees, or tracking error. ETF deep analysis does use fund-specific context, but those missing structured fields are separate from company financial statements and should be checked against the fund provider.

The problem it solves

PanWatch addresses the gap between passive brokerage dashboards and an investor’s own research routine. It combines holdings, technical signals, news and sentiment analysis, scheduled market reports, conditional alerts, and simulated trades without requiring the data and workflow to live in separate services.

Who should try it — and who should skip

Try PanWatch if you self-host tools, track holdings across more than one exchange, and want AI research, conditional alerts, scheduled reports, and paper execution in the same interface. Skip it if you need complete ETF disclosure data, an exchange-grade execution simulator, or a lightweight monitoring setup without model configuration and agent workflows.

Related repositories

Source & attribution

Based on the TNT-Likely/PanWatch GitHub repository and its README at https://github.com/TNT-Likely/PanWatch.

GitHub data · last synced Nov 4, 2026Reviewed by Henry
← Back to TopGit

Want a second opinion on PanWatch?

Ask an AI that can read this page — one click and you get its take on PanWatch.

GitHub