Applying Avellaneda–Stoikov to Mean-Reverting Futures Pairs

Summary

The document describes a cross-exchange futures pairs strategy involving products whose prices are expected to remain close. The trader enters one leg with a limit order and the other with a market order, aiming for price neutrality. The strategy’s signal is the ratio between the two futures prices, treated as a synthetic market that tends to revert toward its usual level.

The author asks whether the Avellaneda–Stoikov optimal-control framework can determine suitable limit prices in this setting, or whether a different approach is needed. No answer, model specification, performance data, or execution analysis is supplied. The scenario nevertheless highlights that a market-making framework may need to account for a spread signal, two-leg execution, and the risk that only one leg fills; the document leaves open how to model those features and optimize quotes.

Key ideas

• The strategy trades futures pairs across exchanges when their price ratio deviates from its usual level.
• One leg is submitted as a limit order and the opposite leg as a market order.
• The trader views the ratio as a mean-reverting synthetic market and seeks overall price neutrality.
• The document asks whether Avellaneda–Stoikov can optimize quotes for this setup but gives no resolution.
• Two-leg execution and one-sided fill risk are relevant modeling questions raised by the scenario.

Tags

# Optimal control for pairs trading?


# Optimal control for pairs trading?












My trading scenario is essentially pairs trading 'identical' futures pairs from different exchanges, similar to arbitrage but for futures.

One side is entered with limit orders and the opposite side is entered with market orders so that overall I am price neutral.

I make money from the ratio between the prices of the two products, which is very close to 1.0 but deviates over time.

What I am trading is a synthetic market composed of these ratios which is highly mean reverting.

My question is, can I still use the optimal stochastic control work of Avellaneda-Stoikov or is there more work I should be looking at here to make sure the prices I'm posting are optimal for the market?

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

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