A high-frequency, market-neutral relative-value strategy designed to capture short-horizon statistical dislocations in U.S. equities.

The strategy operates on a dynamic universe of high-liquidity U.S. stocks, applying a multi-stage correlation filter driven by industry structure and recent return dynamics. Within these clusters, cointegration analysis is used to identify statistically stable equity pairs suitable for spread trading.

Each pair is modeled using a stochastic mean-reversion framework, estimating equilibrium levels, volatility regimes, and mean-reversion half-life. Trades are initiated when spreads deviate beyond statistically significant bounds, positioning for convergence back to modeled fair value.

Risk is managed through a dual-constraint system combining:

• • Price-based drawdown limits
• • Time-based stop losses derived from half-life estimates

Positions are exited upon mean reversion within the expected temporal window or automatically closed when risk constraints are violated, maintaining strict exposure control and execution discipline.