Spot and Delivery Futures Spread Arbitrage with Stationarity and Z-Scores

Summary

This prototype describes a market-neutral approach to trading the spread between spot assets and dated futures. It proposes checking spread stationarity with an Augmented Dickey–Fuller test, then using a Z-score to identify unusually wide or narrow spreads. The stated directions pair long spot with short futures when the spread is high, and short spot with long futures when it is low; positions are intended to close as the spread returns toward a target range. The document also describes order-book midpoint checks, limit orders, simultaneous exits, and rollback steps if one leg fails.

Risk controls include loss limits, delivery-date safeguards, holding-time limits, and cooldowns. The source is incomplete, and no backtest results or evidence of profitability are provided. The described execution workflow cannot establish that either leg will fill at the assumed prices. In addition, the implemented stationarity test approximates its p-value using a normal distribution, while standard ADF inference uses nonstandard critical values; its stationarity gate therefore needs independent validation. Thresholds and operational details are examples, not demonstrated optimal settings.

Key ideas

• The strategy screens spot–futures spreads for stationarity before trading them.
• A Z-score is used to identify large spread deviations and guide paired position direction.
• The proposed exit waits for the spread to move back toward a target range.
• Order-book checks and rollback logic aim to limit execution risk when one leg fails.
• The excerpt gives no performance results, and its ADF p-value calculation needs validation.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.