How likitha-1504/ai-trading works
AI Micro-Trading Agent is a local-first Indian-equities trading system with a React terminal, FastAPI control plane, deterministic trading controls, and an independent exit watchdog. The principal workflow runs from market ticks through features, strategy ranking, advisory AI, deterministic risk and cost sizing, broker execution, and position monitoring. Research, simulation, operations, and dashboard capabilities are shown as product branches. Workflow links follow the README; sampled code confirms selected internal responsibilities, but many subsystem integrations were not inspected.
Main entry points: feature_engine.py Feature Engine · execution_engine.py Execution Engine · position_watcher.py Position Watcher · adapters Broker Adapter
Components 24
Market Signals
- Tick Normalizer
backend/app/adapters/normalizer.py - Feature Engine
backend/app/engine/feature_engine.py - Strategies
backend/app/strategies - Candidate Ranker
backend/app/strategies/candidate_ranker.py - Market Intelligence
backend/app/market_intelligence/intelligence_engine.py
Trade Control
- AI Council
backend/app/ai - Deterministic Trade Gate
backend/app/gate/trade_gate.py - Risk & Cost Sizing
backend/app/capital/risk_engine.py
Execution Safety
- Execution Engine
backend/app/execution/execution_engine.py - Broker Adapter
backend/app/adapters - Position Watcher
backend/app/watcher/position_watcher.py - SQLite WAL
backend/app/db
Research & Operations
- Post-Trade Learning
backend/app/learning - Backtesting & Paper
backend/app/backtesting - Strategy Research
backend/app/research/research_engine.py - Research Sandbox
backend/app/sandbox/sandbox_engine.py - Capital Scaling
backend/app/capital_scaling/scaling_manager.py - Recovery & Operations
backend/app/disaster_recovery
Trading Terminal
- REST API
backend/app/api/v1 - React Terminal
frontend/src/App.tsx - API Client
frontend/src/services/api.ts
Other components
- Trader / Operator
- Broker Market Feed
- Exchange / Simulator
Connections 25
- Broker Market Feed → Tick Normalizer: streams ticks
- Tick Normalizer → Feature Engine: normalizes events
- Feature Engine → Strategies: provides features
- Strategies → Candidate Ranker: produces candidates
- Candidate Ranker → AI Council: requests review
- AI Council → Deterministic Trade Gate: returns advice
- Deterministic Trade Gate → Risk & Cost Sizing: permits candidate
- Risk & Cost Sizing → Execution Engine: sends sized order
- Execution Engine → Broker Adapter: dispatches order
- Broker Adapter → Exchange / Simulator: submits order
- Broker Market Feed → Position Watcher: supplies ticks
- Position Watcher → Broker Adapter: dispatches exits
- Execution Engine → SQLite WAL: records trades
- Position Watcher → SQLite WAL: records exits
- SQLite WAL → Post-Trade Learning: supplies trade history
- Trader / Operator → React Terminal: uses terminal
- React Terminal → API Client: requests data
- API Client → REST API: calls REST
- Trader / Operator → REST API: controls trading
- Feature Engine → Market Intelligence: analyzes market context
- Trader / Operator → Backtesting & Paper: runs simulations
- Trader / Operator → Strategy Research: evaluates strategies
- Strategy Research → Research Sandbox: tests in isolation
- Trader / Operator → Capital Scaling: authorizes stage changes
- Trader / Operator → Recovery & Operations: manages recovery