ENTRYX

A professional, MetaTrader-style trading terminal with a deeply integrated local-AI layer.

EntryX is built as a serious software product — not a demo. It combines professional trading-terminal functionality, real-time market data, advanced charting, order/position management, strategy development, backtesting, and a local-first AI Copilot that understands the market structure behind the chart.

Core principles

• Local-first AI. All AI features run on free/open-source or self-hosted models (Ollama, llama.cpp, Hugging Face). No paid API is required, and none is ever called silently. If a local model is unavailable, EntryX reports "AI unavailable".
• Modular architecture. Frontend, backend, market-data engine, trading engine, broker adapters, chart engine, strategy/backtest engines, indicator engine, AI engine, risk engine, auth, database, notifications, logging, configuration are separated by clear interfaces.
• Provider abstraction. Market data, brokers, and AI models are all adapter-based. EntryX never hard-codes a single provider or broker.
• AI never controls real trading by default. AI analysis and execution are separated. Execution always passes through the Strategy → Risk → Execution → Broker chain with explicit user opt-in, validation, and a kill switch.

Repo layout

entryx/
├── docs/           # Architecture, roadmap, API/event/database contracts, testing strategy
├── backend/        # Python/FastAPI service (auth, DB, market data, trading, risk, strategy, AI)
│   ├── app/
│   │   ├── core/       # config, security, logging, exceptions
│   │   ├── db/         # SQLAlchemy models + session
│   │   ├── api/v1/     # REST routers
│   │   ├── schemas/    # Pydantic contracts
│   │   ├── services/   # business logic
│   │   ├── providers/  # market_data/, broker/, ai/ adapters
│   │   └── ws/         # WebSocket hub
│   └── tests/
├── frontend/       # Flutter desktop app (macOS/Windows/Linux)
│   └── lib/
│       ├── app/         # app shell, routing, theme
│       ├── core/        # config, API client, WS client, state stores
│       ├── features/    # auth, shell, market_watch, chart, ai, trading
└── .github/        # CI

Stack

• Backend: Python 3.14, FastAPI, SQLAlchemy 2, Alembic, WebSockets
• Database: MariaDB (production), SQLite (dev/tests)
• Frontend: Flutter 3.41 (Dart 3.11) — desktop-first trading terminal UI
• AI: Ollama / llama.cpp / Hugging Face open models, via a pluggable AIProvider interface
• Data/state: normalized internal market-data format; provider adapters for any source

Development status

Phase 6 (market structure + Smart Money Concepts) is the current focus; the deterministic structure detectors (swings, HH/HL/LH/LL, BOS, CHoCH, trend/range, breakout/retest) are complete. See docs/ROADMAP.md for the phased build order and status.