How quill3h/agent-ai-t-ng-h-a-vi-c-trading works
Crypto Paper Agent is a cryptocurrency market-data, strategy research, and simulated-trading system; the dependencies shown are paper-only, not live order execution. The principal workflow is historical market data into feature-enriched backtesting, strategy decisions, simulated broker execution, and performance reporting. Separate research tooling supports funding-arbitrage and reinforcement-learning experiments, while a web preview presents offline evidence. The README is empty, and excerpts are partial, so some subsystem wiring—especially data orchestration and report invocation—is not asserted.
Main entry points: engine.py Backtest Engine · paper_broker.py Paper Broker · App.jsx Evidence Preview · fetcher.py Market Fetcher
Components 24
Market Data
- Market Fetcher
crypto-paper-agent/src/data_layer/fetcher.py - Vision Downloader
crypto-paper-agent/src/data_layer/binance_vision_downloader.py - Parquet Cache
crypto-paper-agent/src/data_layer/cache_manager.py - Feature Pipeline
crypto-paper-agent/src/features/indicators.py - News Calendar
crypto-paper-agent/src/features/news_calendar.py
Trading Simulation
- Backtest Engine
crypto-paper-agent/src/backtest/engine.py - Trading Strategies
crypto-paper-agent/src/strategies - Paper Broker
crypto-paper-agent/src/execution/paper_broker.py - Order Models
crypto-paper-agent/src/execution/order_models.py - Position Sizing
crypto-paper-agent/src/risk/position_sizing.py - Order Invariants
crypto-paper-agent/src/risk/invariant_checks.py - Circuit Breakers
crypto-paper-agent/src/risk/circuit_breakers.py
Research
- Funding Arbitrage
crypto-paper-agent/src/strategies/funding_arbitrage.py - RL Environment
crypto-paper-agent/src/research/rl_env.py - Research Workflow
crypto-paper-agent/src/research/workflow.py - Research Validation
crypto-paper-agent/src/research/validation.py
Reporting
- Report Generator
crypto-paper-agent/src/report/generator.py - Performance Metrics
crypto-paper-agent/src/report/metrics.py - Trade Logger
crypto-paper-agent/src/logging/trade_logger.py
Web Preview
- Evidence Preview
crypto-paper-agent/web-preview/src/App.jsx - Offline Replay
crypto-paper-agent/web-preview/src/lib/replay.js - Evidence Data
crypto-paper-agent/web-preview/src/data/evidence.js
Other components
- Researcher
- Binance public API
Connections 14
- Researcher → Backtest Engine: runs simulation
- Market Fetcher → Binance public API: fetches data
- Vision Downloader → Binance public API: downloads archives
- Backtest Engine → Feature Pipeline: computes features
- Backtest Engine → Trading Strategies: evaluates signals
- Backtest Engine → Paper Broker: submits orders
- Backtest Engine → Paper Broker: finalizes run
- Paper Broker → Position Sizing: sizes orders
- Funding Arbitrage → Circuit Breakers: records outcomes
- RL Environment → Paper Broker: simulates actions
- Research Workflow → Research Validation: validates research
- Report Generator → Trade Logger: writes trade records
- Evidence Preview → Offline Replay: builds replay
- Evidence Preview → Evidence Data: reads evidence