Paper Proposes Multi-Level Market Making Using Reinforcement Learning

chaumian · x · 2026-08-20

The paper 'Multi-Level Market Making with Reinforcement Learning' introduces a framework for market making in limit order books. The algorithm maximizes revenue by dynamically submitting orders across multiple price levels while controlling inventory. It models allocations using multivariate logistic-normal distributions and aggregates features with a deep-set encoder. Performance is illustrated in simulated markets with noise, tactical, and strategic traders.

Original post →

More from Research

Research channel →