How jesse-ai/jesse works
Jesse is a self-hosted crypto strategy framework: users define Python strategies, then run backtests and research, optimize parameters, and deploy in paper/live trading. The main runtime spans API controllers, strategy and simulation modes, exchange/order handling, and session results; candles feed both simulation and live market views. Separate branches provide candle import/export, statistical significance and Monte Carlo analysis, ML, charts, and MCP access. The graph uses README capabilities and the supplied tree; most subsystem wiring beyond sampled controller, service, and repository excerpts is intentionally left unspecified.
Main entry points: controllers HTTP controllers · backtest_mode.py Backtest mode · live_controller.py Live mode · strategy_service.py Strategy runtime
Components 27
API and access
- HTTP controllers
jesse/controllers - Authentication
jesse/services/auth.py
Strategy and trading
- Strategy runtime
jesse/services/strategy_service.py - Trading routes
jesse/routes - Backtest mode
jesse/modes/backtest_mode.py - Live mode
jesse/controllers/live_controller.py - Exchange adapters
jesse/exchanges - Order handling
jesse/services/order_service.py - Technical indicators
jesse/indicators - Metrics and reports
jesse/services/metrics.py - Interactive charts
jesse/services/charts.py
Research workflows
- Parameter optimization
jesse/modes/optimize_mode - Monte Carlo
jesse/modes/monte_carlo_mode - Rule significance
jesse/research/rule_significance_testing - Machine learning
jesse/research/ml.py
Market data and state
- Candle service
jesse/services/candle_service.py - Candle import/export
jesse/controllers/candles_controller.py - Candle repository
jesse/repositories/candle_repository.py - Session records
jesse/models/BacktestSession.py - Order records
jesse/repositories/order_repository.py
User integrations
- MCP server and tools
jesse/mcp/server.py - Notifications
jesse/services/notifier.py - WebSocket updates
jesse/services/ws_manager.py - Redis pub/sub
jesse/services/redis.py - Research API
jesse/research
Other components
- Trader / researcher
- MCP assistant
Connections 36
- Trader / researcher → HTTP controllers: submits requests
- HTTP controllers → Authentication: checks access
- HTTP controllers → Backtest mode: starts backtests
- HTTP controllers → Live mode: starts live sessions
- HTTP controllers → Parameter optimization: starts optimization
- HTTP controllers → Monte Carlo: starts simulations
- HTTP controllers → Rule significance: requests tests
- HTTP controllers → Candle import/export: manages candles
- Backtest mode → Strategy runtime: runs strategy
- Live mode → Strategy runtime: runs strategy
- Strategy runtime → Trading routes: evaluates routes
- Strategy runtime → Technical indicators: uses indicators
- Backtest mode → Candle service: reads candles
- Live mode → Candle service: reads candles
- Candle service → Candle repository: loads candles
- Candle import/export → Candle repository: imports and exports
- Strategy runtime → Order handling: creates orders
- Order handling → Exchange adapters: submits orders
- Order handling → Order records: persists orders
- Backtest mode → Session records: records sessions
- Parameter optimization → Backtest mode: evaluates strategies
- Parameter optimization → Session records: records results
- Monte Carlo → Session records: records results
- Rule significance → Candle service: uses history
- Backtest mode → Metrics and reports: produces metrics
- Backtest mode → Interactive charts: produces charts
- Live mode → Interactive charts: provides chart data
- MCP assistant → MCP server and tools: invokes tools
- MCP server and tools → HTTP controllers: calls Jesse API
- Research API → Backtest mode: runs backtests
- Research API → Parameter optimization: runs optimization
- Research API → Rule significance: runs tests
- Research API → Monte Carlo: runs simulations
- Machine learning → Backtest mode: uses backtest data
- Live mode → Notifications: sends alerts
- WebSocket updates → Redis pub/sub: subscribes events