NNArbitrage

In a market flooded with lagging indicators and overfit strategies, this utility offers a fresh edge: real-time statistical arbitrage powered by an adaptive neural network, built entirely in MQL5—no DLLs, no external dependencies.

🚀 Key Advantages

• Smart Spread Modeling
  It dynamically calculates a hedge ratio between two correlated instruments (like EURUSD vs GBPUSD), forming a synthetic spread that reflects true relative value. This isn’t just correlation—it’s cointegration-aware logic.
• Neural Network Intelligence
  A lightweight, in-code neural net learns from recent market behavior to predict short-term mean reversion. It adapts as conditions change, filtering out noise and highlighting high-probability setups.
• Z-Score Precision
  The spread is normalized using a rolling z-score, giving traders a clear view of when the pair is statistically stretched—and when it’s likely to snap back. 
• Session-Aware Filtering
• Trades are only flagged during high-liquidity hours (e.g., London–New York overlap), avoiding false signals in thin markets.
• Execution-Ready Signals
  While it’s an indicator, not an EA, it outputs clear long/short spread setups that can be wired into automated systems or used manually with confidence.
• No Curve-Fitting, No Black Box
  Everything is transparent and tunable. You control the training horizon, z-score thresholds, and risk filters. The neural net trains online—no preloaded weights, no overfitting.
• 📈 Use Cases
• Pairs Traders looking to exploit temporary dislocations between correlated assets.
• Scalpers who want to add a statistical edge to their short-term setups.
• Quant-Inclined Discretionary Traders who want to visualize spread behavior and neural predictions before pulling the trigger.
• EA Developers who want to integrate machine learning logic without relying on external libraries.