How the Open Grid engine decides what to trade, when, and how much.
The Open Grid framework trades regimes, not predictions. It adapts position sizing, signal weights, and stop logic to whether the market is trending, mean-reverting, or neutral.
Core principles
Regime first. ADX and Hurst exponent classify each day into trend, mean-reversion, or neutral. Signals and sizing change accordingly.
Volatility normalization. Momentum scores are divided by realized volatility so the entry threshold works across calm and volatile periods.
Risk before return. Every trade is sized by ATR risk units. A single position cannot exceed 15% of capital. Gross exposure is capped at 150%.
Confirmation, not noise. A separate confirmation score (volume, breadth, or external model) is blended with momentum. No trade fires on price alone.
Architecture
Data Feed -> Feature Engineering -> Regime Classifier -> Signal Engine -> Risk Manager -> Execution -> Reporting
Time horizon
Primary holding period is 2 to 10 trading days. The engine uses a 10-day time stop to prevent positions from drifting without conviction.