A cross-platform (Windows & macOS) desktop dashboard and trading terminal for a Kraken spot account: an adjustable card workspace with live positions and true cost basis, portfolio-level risk (volatility, BTC beta, concentration, risk contribution, tail metrics), Bloomberg-style analytics, real-time charts and market microstructure, and a full order suite with hard safety rails. A Claude-powered analyst panel narrates the numbers; the math engine computes them.
Personal-use software, MIT-licensed. It talks to Kraken's official APIs and nothing else: no third-party price feeds, no telemetry, and your API keys never leave your machine (they live in your OS keychain — Windows Credential Manager or macOS Keychain — not on disk).
v2.0.0 — the core desk is feature-complete: adjustable workspace, live
trading with rails, streaming market data, the analytics suite, and the AI
analyst. Runs on Windows and macOS; a per-user Windows installer is built from
packaging/windows/ and a macOS .app/.dmg from
packaging/. The canonical spec lives in
docs/ (PRD, app flow, tech stack, frontend guidelines, backend
structure, implementation plan); session state is tracked in
progress.txt.
- Adjustable workspace — every panel is a card you can float, tab, split, and resize; three saved perspectives (Trading / Analysis / Monitor) and a reset, with your layout persisted between sessions
- Positions — live holdings, mark price, USD value, average cost, break-even, unrealized PnL (cost basis labeled per quote currency)
- Risk — annualized and EWMA volatility, BTC beta, HHI, effective bets, per-asset risk contribution, VaR/CVaR, with every assumption footnoted
- Analytics — ratios (Sharpe, Sortino, Calmar, rolling), exposure (correlation heatmap, concentration, sector map), attribution (per-asset realized PnL, monthly-returns heatmap, BTC benchmark), and scenario stress (drawdown, BTC shocks, Monte Carlo NAV fan)
- Charts & microstructure — candlestick chart with volume and selectable timeframes, a cumulative-depth order book, a live Time & Sales tape, and a watchlist; one active-symbol bus keeps them in sync
- Trading — market/limit/stop/take-profit/trailing orders with a confirmation dialog, a paper-mode default, a max-order-value cap, and an append-only local audit log; open-orders management; live order state via WebSocket
- Performance — equity curve, realized/unrealized PnL history, drawdown, per-position stats, trade expectancy
- Alerts — price/PnL/risk rules with native OS notifications (Windows toasts, macOS Notification Center)
- Analyst — rules-based narration for free; optional Claude analysis (your
own Anthropic key,
claude-opus-4-8) with portfolio commentary, trade review, and free-text Q&A, streamed and priced per call. It narrates engine output and never invents numbers
The workspace ships three saved perspectives (all shown here on demo data).
Trading — watchlist, live chart, analytics, and the AI analyst:
Analysis — full ratio panel, order ticket, and alerts:
- Paper mode is the default; going live requires an explicit, typed confirmation
- Every order passes validation, a size cap, and a confirmation dialog
- Every order attempt is audit-logged locally
- Kraken API keys need only query + trade permissions. Never enable Withdraw Funds — the app has no withdrawal code path and never will
- Windows — download
crypto-quant-desk-2.0.0-setup.exeor build it from source. Per-user install (no admin), Start-menu shortcut, clean uninstall that preserves your data and keys. - macOS — download the
.dmgor build the.appfrom source, then drag it to Applications. It is unsigned, so approve it once on first launch (right-click → Open).
Build steps for both are in packaging/README.md.
First launch opens in demo mode (synthetic portfolio, real market data). Connect your own account via File > Settings with a Kraken API key created at https://www.kraken.com/u/security/api (permissions: Query Funds, Query Open/ Closed Orders & Trades, Query Ledger Entries, Create & Modify Orders — never Withdraw). Optionally add an Anthropic key there to enable the AI analyst.
Requires Python 3.11+. On macOS/Linux use python3 and source .venv/bin/activate.
git clone https://github.com/ccharafeddine/crypto-quant-desk.git
cd crypto-quant-desk
python -m venv .venv
.venv\Scripts\activate # Windows; source .venv/bin/activate on macOS
pip install -e ".[dev]"
python -m cqdsrc/cqd/
├── engine/ # pure math: risk, metrics, cost basis, performance (no I/O)
├── data/ # Kraken REST + WebSocket clients, normalizer, credentials
├── trading/ # order service, paper broker, limits, audit log
├── alerts/ # rule engine + native OS notifications (Windows/macOS)
├── analyst/ # rules narration + optional Claude integration (grounded, priced)
└── ui/ # PySide6 QtAds card workspace, dockable panels, token-driven themes
Design rules: the engine is pure functions (fully tested, no I/O/Qt/network); all market and portfolio data comes from Kraken's official APIs; every order flows through one service with non-bypassable rails; no database — settings, JSON state, and the OS credential vault.
pytest -q # tests
ruff check src tests # lint
ruff format src tests # formatMIT. See LICENSE.
Experimental software that can place real orders on your live Kraken account when you enable live mode with trade-permissioned keys. Use at your own risk. The authors accept no liability for financial losses.


