How itaygenkin/tradingagency works
AI Trading Agent runs a morning market-analysis workflow and an evening accuracy audit. The morning process checks market availability, gathers pre-market prices and news, generates a Gemini-backed report and stores predictions; after close, pending predictions are compared with observed performance and persisted. Celery distributes per-ticker audit work through Redis. The graph preserves documented integrations and the major domain stages; details of the read-but-unsampled day and night pipeline implementations are limited to the README and SOURCE INDEX references.
Main entry points: day_analysis.py Day Pipeline · celery_tasks.py Audit Tasks · market_provider.py Market Provider · night_audit.py Night Audit
Components 18
Morning Analysis
- Day Pipeline
src/trading_agency/application/day_analysis.py
Evening Audit
- Night Audit
src/trading_agency/application/night_audit.py - Audit Tasks
src/trading_agency/application/celery_tasks.py - Performance Validator
src/trading_agency/application/audit_service.py - Celery App
src/trading_agency/celery_app.py
Data and AI
- Market Provider
src/trading_agency/infrastructure/market_provider.py - Market Analyst
src/trading_agency/infrastructure/llm_engine.py - LLM Factory
src/trading_agency/infrastructure/llm_factory.py
Domain and Storage
- Report Files
src/trading_agency/utils/utils.py - Trading Models
src/trading_agency/domain/models.py - Market Repository
src/trading_agency/infrastructure/repository.py - Result Types
src/trading_agency/domain/result.py
Other components
- Operator
- Gemini LLM
- Market Data
- News Search
- PostgreSQL
- Redis Broker
Connections 23
- Operator → Day Pipeline: starts
README.md - Operator → Night Audit: starts
README.md - Day Pipeline → Market Provider: gets market data
src/trading_agency/application/day_analysis.py - Day Pipeline → Market Analyst: requests analysis
src/trading_agency/application/day_analysis.py - Day Pipeline → Report Files: saves report
src/trading_agency/application/day_analysis.py - Day Pipeline → Market Repository: stores predictions
src/trading_agency/application/day_analysis.py - Market Provider → Market Data: fetches prices
README.md - Market Provider → News Search: searches news
src/trading_agency/infrastructure/market_provider.py - Market Analyst → LLM Factory: gets model
src/trading_agency/infrastructure/llm_engine.py - LLM Factory → Gemini LLM: provides model
README.md - Market Analyst → Gemini LLM: invokes model
src/trading_agency/infrastructure/llm_engine.py - Day Pipeline → Trading Models: uses snapshots
src/trading_agency/application/day_analysis.py - Market Repository → PostgreSQL: reads and writes
src/trading_agency/infrastructure/repository.py - Night Audit → Market Repository: loads predictions
src/trading_agency/application/night_audit.py - Night Audit → Market Provider: gets performance
src/trading_agency/application/night_audit.py - Night Audit → Audit Tasks: dispatches audits
src/trading_agency/application/night_audit.py - Audit Tasks → Market Provider: gets performance
src/trading_agency/application/celery_tasks.py - Audit Tasks → Performance Validator: evaluates prediction
src/trading_agency/application/celery_tasks.py - Audit Tasks → Market Repository: updates results
src/trading_agency/application/celery_tasks.py - Performance Validator → Market Analyst: invokes LLM
src/trading_agency/application/audit_service.py - Performance Validator → Result Types: returns evaluation
src/trading_agency/application/audit_service.py - Audit Tasks → Celery App: registers task
src/trading_agency/application/celery_tasks.py - Celery App → Redis Broker: uses broker
README.md