Testing Grid Market Making Under Different Order Latency Assumptions

Summary

This example studies how order latency affects a grid-based market-making strategy for ETH perpetual futures. The strategy estimates trading intensity from the distance between the midpoint and observed trade arrivals, fits a decay relationship to that intensity, and updates volatility estimates from midpoint changes. It uses those estimates to set a reservation price and quote spread, then maintains buy and sell order grids while skewing quotes according to inventory.

The notebook compares backtests using order latency derived from feed latency, observed live order latency, and amplified feed latency. It shows the configuration and reporting workflow, but the supplied text gives no numerical performance comparisons or conclusions from the plots. Results depend on the specific ETH data, queue model, fees and maker rebate assumptions; the example also simplifies some trade-arrival cases and assumes no partial fills. Latency modeling is therefore central to interpreting any apparent strategy performance.

Key ideas

• The strategy maintains layered limit orders on both sides of the market and adjusts quotes for inventory.
• Trading intensity and volatility estimates drive its spread and reservation price.
• The notebook compares latency data based on feed latency, live order latency, and amplified feed latency.
• Backtest assumptions include a queue model, no partial fills, and a maker rebate.
• The provided text describes the experiment but does not report numerical results or conclusions.

Tags

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