Implements standardized risk management and drawdown protection patterns for algorithmic trading systems.
This skill provides a robust framework for managing financial risk within trading systems by implementing a standardized Drawdown Guardrails pattern. It ensures consistency across backtesting, live execution, and model training by tracking high-water marks and equity peaks rather than just initial capital. By providing sticky halt flags and threshold-based position scaling, it prevents catastrophic losses and trading whipsaws, while allowing existing positions to exit safely via stop-losses. This is an essential pattern for developers building reliable automated trading infrastructure or ML-based financial models where production stability is paramount.
Key Features
01Sticky hard-halt triggers to prevent automatic resumption after high-risk events
02Safe-exit logic allowing stop-losses to execute even when new trades are blocked
03Integrated training penalties to discourage models from learning high-drawdown behaviors
04Multi-tier risk response including warning zones and position size scaling
050 GitHub stars
06Peak-equity high-water mark tracking for accurate drawdown calculation
Use Cases
01Standardizing risk metrics between backtesting engines and production environments
02Protecting live trading accounts from catastrophic losses during market volatility
03Implementing early stopping and fitness penalties in PyTorch-based trading model training
