Zero-shot adaptation to order book dynamics

2026-05-20

AI Summary

This work addresses the limitations of traditional market-making strategies in adapting to evolving market mechanisms and shifting trading objectives. Building upon the Avellaneda–Stoikov framework, the authors introduce a successor-measure-based adaptive mechanism that decouples market dynamics from trading goals, enabling real-time generation of optimal quotes through low-dimensional parameters and a target vector. This approach yields the first market-making architecture capable of zero-shot adaptation to dynamic order book changes without requiring retraining. Crucially, it preserves the analytical tractability of the Hamilton–Jacobi–Bellman equation while substantially enhancing both the robustness and computational efficiency of the resulting strategy.

Abstract

We describe an adaptive market-making architecture that preserves the analytical structure of the Avellaneda--Stoikov framework while introducing a successor measure-style adaptation mechanism. In our paper we keep Avellaneda--Stoikov fast Hamilton--Jacobi--Bellman structure and make it adaptive to changing market regimes and trading objectives. The central idea is to separate market dynamics from the trading objective. The market state determines a low-dimensional set of Avellaneda--Stoikov parameters, while recent realized rewards determine a low-dimensional objective vector. The HJB forward map then converts this objective into optimal bid and ask quotes through a scalarization of future reward features.

Problem

• zero-shot adaptation
• order book dynamics
• market making
• adaptive trading
• market regimes

Innovation

• zero-shot adaptation
• market making
• Hamilton-Jacobi-Bellman
• successor measure
• order book dynamics