Self-hosted · Durable workers · AI agent ready — v5.0.1 released

Your private AI quant operating system.

Research markets, generate Python strategies, backtest ideas, and run live trading workflows — on infrastructure you fully control. One stack for AI analysis, charting, strategy code, backtests, quick trade, and live operations.

Now on mobile Use the mobile web app, or install the native Android build.
ai.quantdinger.com / dashboard
Dashboard
Trading Bots
AI Analysis
Backtest
Portfolio
Settings
Equity $ 48,231.05 + 3.42%
24h PnL + $ 1,605.22 + 3.44%
Win rate 64.8% 30d
Active bots 12 + 2 today
StrategyMarketModePnLStatus
Grid ETH/USDT — Neutral Crypto · SwapLive + 842.10 ● running
DCA BTC/USDT — Accum. Crypto · SpotSignal + 312.66 ● running
Martingale SOL/USDT Crypto · SwapLive − 58.40 ● running
Trend AAPL US Stocks · IBKRSignal + 206.15 ● running
15+
Crypto · stocks · forex & more
6
LLM providers
4
Built-in bot strategies
100%
Self-hosted, your data
Platform

One stack. Every edge you need.

Research to execution in one product — AI analysis, charting, strategy code, backtests, quick trade, and live operations connected from day one.

Fast AI Market Analysis

Structured, low-latency analysis with memory, ensemble voting, and confidence calibration. Compare views across OpenAI, Claude, Gemini, DeepSeek, Grok and more.

Strategy API V2

Write one Python strategy contract for backtests and live execution. initialize(context) owns the universe, subscriptions, warmup and benchmark; lifecycle callbacks emit explicit, auditable order intents.

Deterministic Backtesting

Server-side backtests share the Strategy API V2 manifest, time semantics, sizing, risk controls and order intent model used by live deployments. Every run retains its code and configuration snapshot.

Autonomous Trading Bots

Grid, Martingale, Trend Following and DCA out of the box. Execution-aware, restart-resilient, with signal or fully-live execution modes.

Self-Hosted by Design

One Docker Compose stack brings up separate API, trading, scheduler, Celery and migration roles. Your keys, strategies, AI history and trades remain on infrastructure you control.

Operator-Ready

PostgreSQL commands, renewable leases, heartbeats and fencing tokens protect durable execution. Optional Prometheus, Grafana and Alertmanager add production visibility.

Who it's for

Built for everyone running real strategies.

Whether you trade solo, run a quant team, build a product, or operate a fund — QuantDinger is shaped for the way you actually work.

AI-assisted individual traders

Solo traders and quant enthusiasts who want AI, charting, backtests and bots in one workspace — instead of juggling 5 tools.

Quant research teams

Teams that want a unified toolchain — data ingestion, strategy code, backtests, live execution, and a shared history of experiments.

Python strategy developers

Developers who want a fast path from Python idea → backtest → live execution, with a documented SDK and clear extension points.

Privacy-first teams

Research desks and operators that cannot send keys, alpha, or trading history to a hosted SaaS. Everything stays inside your infra.

Fintech & AI-trading startups

Founders who need users, billing, credits, OAuth and payments in the stack from day one — not months of re-building plumbing.

Asset management firms

Internal desks wanting one integrated platform for research, strategy code, execution, alerts, and operator-side admin controls.

Use cases

One platform, many operating models.

The same deployable stack adapts to an internal research lab, a live strategy incubator, a customer-facing product, or a classroom.

01

Internal research & backtest lab

Stand up a private quant research and backtesting platform for your team — reproducible runs, shared strategy library, saved snapshots.

02

AI trading assistant products

Build consumer or pro-grade AI trading assistants on top of ensemble analysis, calibration, and memory — with a working UI out of the box.

03

Strategy incubation & live validation

Run a strategy incubator end-to-end: ideation, Python code, backtest, paper trade, then graduate to live with small size.

04

Paid signals & strategy service

Offer member-based trading signals or strategy subscriptions with built-in memberships, credits, and USDT TRC20 payments.

05

Quant education & training

Ship a hands-on learning environment for quant finance classes and bootcamps — real Python strategies, real charts, safe paper trading.

06

Internal ops for asset managers

Run a private operations console for a fund or family office: portfolio monitors, alerts, multi-user roles, and audit-friendly history.

MCP · v5.0.1 For AI agents

Plug Cursor, Claude Code & Codex straight into your quant stack.

QuantDinger ships a first-class Agent Gateway and quantdinger-mcp 0.4.0. Its 25 tenant-scoped tools cover market data, Strategy API V2, chart indicators, asynchronous backtests, runtime visibility and explicit quick orders without exposing broker credentials or administrator JWTs.

One-line MCP install

Published on PyPI as quantdinger-mcp 0.4.0 with 25 curated tools. Use stdio locally or sse / streamable-http for network clients.

uvx quantdinger-mcp
# PyPI: quantdinger-mcp==0.4.0

Four capability scopes, least privilege

Issue a token with only Read, Write, Backtest or Trade access. Market and instrument allowlists, expiry, rate limits, paper-only restrictions, idempotency and audit logs remain server-enforced.

Two paths: hosted SaaS or self-host

Path A · Hosted — sign up at ai.quantdinger.com, paper-only by design, T-scope rejected at issuance. Try in 30 seconds. Path B · Self-host — clone, docker compose up, you own scopes, exchange keys, and the live-trading kill switch. Same MCP wiring; only QUANTDINGER_BASE_URL changes.

Live SSE progress Agents stream snapshot → progress → result on long backtests — no blind polling.
Safe by default All agent trading is paper-only until the operator flips AGENT_LIVE_TRADING_ENABLED; hosted SaaS pins it shut.
Immutable audit Every agent request is logged with token, scope, route, status & latency — viewable from the admin panel.
Coverage

Every venue worth trading.

Live crypto execution across Binance, OKX, Bitget, Bybit, Gate and HTX, plus IBKR and Alpaca workflows — unified under one operating model.

Crypto ExchangesBinance · OKX · Bybit · Bitget · Gate · HTX …
$
US StocksIBKR · Yahoo · Finnhub
A
US Equities APIAlpaca · paper & live workflows
FuturesCME · COMEX · CBOT
¥
CN / HK StocksSSE · SZSE · HKEX
Factor libraryBuilt-in & custom indicators
Spot & DerivativesSwap · Perps · Options
Pluggable feedsCustom REST / WS adapters
Mobile app

Your quant stack, now in your pocket.

Open the mobile web app in your browser, or install the native Android build. iOS TestFlight coming soon.

Mobile Web

Same full-feature app, optimised for phones. No install — just open and use.

Open m.quantdinger.com
m.quantdinger.com

Android APK

Native install, splash screen, safe-area & push-ready. Android 7+.

Download APK
www.quantdinger.com/download/app.apk

On Android, you may need to allow "Install from unknown sources" for your browser before installing.

iOS

TestFlight build coming soon — in the meantime, use the mobile web app.

Coming soon
Workflow

From idea to live capital — in minutes.

A straight line from market view to autonomous execution, with a reproducible backtest safety net in between.

01

Research & analyze

Pull multi-timeframe data, run ensemble AI analysis, inspect indicator signals and comparables — all in one workspace.

02

Build or generate

Configure a ready-made bot with the visual wizard, or write full Python indicators / strategies using our documented SDK.

03

Backtest & iterate

Deterministic runs with realistic commissions, slippage, and leverage. Freeze every experiment by code hash and config snapshot.

04

Deploy & operate

Promote a verified strategy through an explicit deployment flow. Monitor PnL, equity, positions, alerts, worker ownership and idempotent order retries.

Technical architecture

Durable process ownership from API to execution.

v5 separates HTTP, long-lived trading, schedules and finite jobs into independent process roles. PostgreSQL commands, leases and fencing keep execution recoverable across restarts and multi-process deployments.

Client Who talks to the platform
Web Browser Mobile (Capacitor) AI Agents · MCP Human API Automation
Edge Delivery & static assets
Nginx Prebuilt Vue 3 dist HTTPS / CDN Same-origin API proxy
Process roles Explicit runtime ownership
HTTP API Auth, validation, durable commands
Trading Worker Strategy runtimes, orders, broker sessions
Scheduler Portfolio, deployments, payments, alerts
Celery Jobs AI, backtests, reports, maintenance
Agent Gateway R/W/B/T scopes · audit · job streams
One backend image · API + trading + scheduler + Celery worker/beat + migration
State Persistence & runtime coordination
PostgreSQL 18 · commands & leases Redis cache · evictable Redis jobs · AOF + noeviction Heartbeats · fencing tokens
Integrations & operations Adapters, metrics and alerts
Crypto exchanges · IBKR · Alpaca Market data · fundamentals · macro · news OpenRouter · OpenAI-compatible · Google Telegram · Email · SMS · Discord Prometheus · Grafana · Alertmanager JSON logs · request IDs · health endpoints MCP · Cursor · Claude Code · Codex
Frontend
Vue 3 · Ant Design Vue · KLineCharts · ECharts · Capacitor
Backend
Flask · Gunicorn · Strategy API V2 · Celery · CCXT
Data
PostgreSQL 18 · Redis cache · Redis jobs · encrypted credentials
Deploy
Docker Compose · production hardening · optional observability overlay
AI stack

The best of every model. At your command.

We don't pick a winner — we orchestrate them. Switch providers per query, or let the ensemble vote with confidence calibration.

OpenRouter
OpenAI GPT
Anthropic Claude
Google Gemini
DeepSeek
xAI Grok
Moonshot Kimi
Qwen
Local LLMs
FAQ

Questions, answered.

Is QuantDinger really self-hosted?

Yes. The default deployment is your own Docker Compose stack with PostgreSQL, separate cache and job Redis services, isolated backend process roles, credentials, and environment under your control.

Do you store my exchange API keys?

Broker credentials and MFA secrets are encrypted with a stable, operator-controlled CREDENTIAL_ENCRYPTION_KEY. They are decrypted only in memory for authorized operations and should never appear in logs.

Can I write my own strategies and indicators?

Absolutely. Executable strategies use Strategy API V2: initialize(context) declares the universe and subscriptions, while handle_data, on_rebalance, or scheduled callbacks emit auditable order intents. The same manifest drives backtests and live deployments.

Which venues are supported for live trading?

The canonical v5 live-crypto matrix covers Binance, OKX, Bitget, Bybit, Gate and HTX for spot and perpetual workflows. Traditional-broker workflows include IBKR and Alpaca.

Is it free?

The backend is Apache 2.0. The prebuilt frontend is source-available under a separate license that allows personal and qualified non-commercial use for free; commercial deployment requires a license — apply here. See TRADEMARKS.md and LICENSE in the repo.

What languages is the app available in?

The app itself is fully localized across 10 languages — English, Simplified & Traditional Chinese, Japanese, Korean, Vietnamese, Thai, French, German, and Arabic.

Can I plug Cursor / Claude Code / Codex directly into QuantDinger?

Yes — v5.0.1 ships quantdinger-mcp 0.4.0 on PyPI with 25 tenant-scoped tools for market data, Strategy API V2, chart indicators, asynchronous backtests, runtime state and explicit quick orders. Install with uvx quantdinger-mcp and configure your MCP client.

Hosted (ai.quantdinger.com) or self-host — which one fits me?

Hosted is best when you want to try QuantDinger from an AI agent in 30 seconds, don't have private exchange keys to plug in, and are fine with paper-only execution — the multi-tenant instance pins paper_only=true and rejects T-scope at issuance, so no agent on the hosted endpoint can ever route real-money orders. Self-host is the right call when you have your own exchange API keys, private strategies/data, run behind a VPN, or eventually want live execution. Both paths use identical MCP wiring — only QUANTDINGER_BASE_URL changes.

Commercial license

Apply for a commercial deployment license.

The backend is Apache 2.0. The prebuilt Vue frontend is source-available — personal and qualified non-commercial use is free; commercial deployment, white-label, or SaaS resale requires a license from us.

  • Self-hosted enterprise / fund deployments
  • Customer-facing SaaS or white-label products
  • OEM embedding of the QuantDinger UI stack

We typically respond within 2–3 business days. For general questions, email support@quantdinger.com.

Ready to run your own quant stack?

Try the hosted app, or deploy the complete v5 stack on infrastructure you control.