How ai-yang/alphapilot works
AlphaPilot is a stock-quant research and trading platform. A typical workflow prepares market data, mines and evaluates factors, saves promising strategies, then produces daily signals or deploys them through a controlled live runtime. The Web portal and research API expose shared jobs and results; backtesting, visualization, notifications, and formula-based miners are additional capabilities. The README documents the major product stages and runtime links; source excerpts confirm selected engine, data, mining, and durable-task responsibilities. Other subsystem links are documented rather than inferred from unsampled implementation.
Main entry points: tasks.py Durable Tasks · cli.py Command Line · database.py Factor Library · alpha_mining Factor Mining
Components 25
User Access
- Web Portal
alphapilot/modules/portal/module.py - Research API
alphapilot/research/api.py - Command Line
alphapilot/app/cli.py - Durable Tasks
alphapilot/research/tasks.py - Task Store
alphapilot/research/store.py
Research Workflow
- Kernel Engine
alphapilot/kernel/engine.py - Factor Mining
alphapilot/modules/alpha_mining - Formula Miners
alphapilot/modules/alphaforge - Factor Library
alphapilot/systems/factor/database.py - Backtest Evaluation
alphapilot/systems/backtest/service.py - Strategy Assets
alphapilot/systems/strategy/service.py - Daily Signals
alphapilot/modules/daily_trade/module.py
Market Systems
- Data Preparation
alphapilot/systems/data/prepare_cn.py - Data Adapters
alphapilot/adapters/builtin/data_source
Trading Runtime
- Live Runtime
alphapilot/systems/live/runtime.py - Risk and OMS
alphapilot/systems/live - Broker Gateway
alphapilot/systems/live/gateway.py
Operations
- Backtest Views
alphapilot/modules/backtest_viz - Notifications
alphapilot/systems/notify/service.py
Other components
- Researcher
- MCP Client
- LLM Provider
- Market Data Provider
- Broker Service
- Telegram, Feishu, Email
Connections 26
- Researcher → Web Portal: uses
README.md - Researcher → Command Line: runs commands
README.md - MCP Client → Research API: connects
README.md - Web Portal → Durable Tasks: shares jobs
README.md - Research API → Durable Tasks: submits jobs
README.md - Durable Tasks → Task Store: persists jobs
alphapilot/research/tasks.py - Durable Tasks → Kernel Engine: builds engine
alphapilot/research/tasks.py - Command Line → Data Preparation: prepares data
README.md - Data Preparation → Data Adapters: downloads from
README.md - Market Data Provider → Data Adapters: provides prices
README.md - Command Line → Factor Mining: starts mining
README.md - Command Line → Formula Miners: starts mining
README.md - Factor Mining → LLM Provider: requests generation
- Formula Miners → Factor Library: emits factors
alphapilot/modules/alphaforge/pipeline.py - Factor Mining → Factor Library: adds candidates
- Factor Library → Backtest Evaluation: evaluates factors
README.md - Backtest Evaluation → Backtest Views: visualizes results
README.md - Factor Library → Strategy Assets: selects factors
alphapilot/systems/strategy/service.py - Strategy Assets → Backtest Evaluation: delegates training
alphapilot/systems/strategy/service.py - Strategy Assets → Daily Signals: generates signals
README.md - Daily Signals → Live Runtime: supplies targets
README.md - Live Runtime → Risk and OMS: routes through
README.md - Risk and OMS → Broker Gateway: dispatches orders
README.md - Broker Gateway → Broker Service: sends orders
README.md - Durable Tasks → Notifications: reports completion
- Notifications → Telegram, Feishu, Email: delivers alerts
README.md