QuantDinger Documentation
August 24, 2026 · View on GitHub
QuantDinger is a self-hosted AI trading platform that connects market research, Strategy API V2 development, backtesting, paper/live execution, operations, and AI-agent access. This documentation describes the current v5 release.
中文文档 · Official website · Web app · GitHub
Live trading can submit real orders. Start in paper mode, grant exchange keys the minimum permissions required, and confirm the legal and operational requirements for your jurisdiction. QuantDinger does not provide investment advice.
Choose your goal
| Goal | Start here | Continue with |
|---|---|---|
| Install QuantDinger | Cloud deployment | Installation troubleshooting |
| Prepare production | Production hardening | Observability |
| Develop a strategy | Strategy API V2 guide | Indicator guide |
| Connect an AI agent | MCP setup | Agent Gateway quickstart |
| Integrate over HTTP | Human OpenAPI | API conventions |
| Extend the backend | Architecture overview | Extension guide |
Five-minute installation
Prerequisites: Docker with Compose v2. The installer asks for an administrator account, generates the required secrets, writes the deployment environment, and starts the prebuilt GHCR stack.
Linux or macOS:
curl -fsSL https://raw.githubusercontent.com/OpenByteInc/QuantDinger/main/install.sh | bash
Windows PowerShell:
irm https://raw.githubusercontent.com/OpenByteInc/QuantDinger/main/install.ps1 | iex
After startup:
| Surface | Default local address |
|---|---|
| Desktop web app | http://127.0.0.1:8888 |
| Mobile web app | http://127.0.0.1:8889 |
| Backend health check | http://127.0.0.1:5000/api/health |
For source builds, reverse proxies, HTTPS, upgrades, and proxy settings, use the cloud deployment guide.
System map
The v5 runtime separates long-lived ownership from request handling:
| Process | Responsibility |
|---|---|
migration | Applies database schema changes before dependent services start. |
backend | HTTP authentication, validation, queries, and durable command submission. |
trading-worker | Strategy runtimes, broker sessions, orders, leases, and reconciliation. |
scheduler-worker | Portfolio, deployment, payment, and notification schedules. |
celery-worker | Finite retryable work such as AI jobs, backtests, reports, and maintenance. |
celery-beat | Periodically enqueues Celery work. |
PostgreSQL is the system of record. redis is an evictable cache;
redis-jobs is the durable Celery broker/result tier. See the
architecture overview before changing process
ownership or shared state.
Documentation map
Deployment and operations
- Cloud deployment
- Installation troubleshooting
- Production hardening
- Observability
- Multi-user operation
- OAuth configuration
- Administrator and settings troubleshooting
- Notifications: Email, SMS, and Telegram
Trading and research
- Strategy API V2 development
- Chart indicator development
- Interactive Brokers
- Runnable examples in
examples/
APIs and agents
- Human Web API and OpenAPI
- Agent documentation
- MCP setup
- Agent Gateway quickstart
- Machine-readable contracts:
api/openapi.yamlandagent/agent-openapi.json
Language and maintenance policy
Each human-readable page uses one primary language. English and Chinese pages are paired when both are maintained; the website language switch links directly to the paired page. Machine-readable schemas, identifiers, environment variable names, and code remain in their canonical English form.
The repository docs/ directory is the source of truth. Update the relevant
index and counterpart language page when a contract or workflow changes. Do not
publish planning notes, completed roadmaps, or temporary audits as product
documentation.