Work

A Review of Scalable Agent-Based Simulation Frameworks for Liquidity Collapse in Limit Order Book Markets

Year: 2026
Type: article
Abstract: Liquidity collapse in electronic limit order book markets can emerge when order-flow imbalances, declining market depth, rising price impact, and the withdrawal of liquidity providers reinforce one another. This paper reviews research relevant to scalable simulation of these dynamics across five connected areas: agent-based financial markets, limit order book modelling, liquidity fragility, parallel and distributed computation, and reinforcement learning for adaptive trading. The reviewed literature demonstrates substantial progress within each area, but also exposes a persistent integration challenge. Behaviourally rich agents and realistic market microstructure increase computational and synchronization costs, whereas many accelerated financial workloads do not reproduce the continuous interaction of heterogeneous traders through a shared order book. A structured comparison of the reviewed approaches is used to identify this gap and derive requirements for a future simulation architecture. The proposed research direction combines heterogeneous trader populations, an event-driven limit order book, adaptive learning agents, and GPU or distributed execution. Its effectiveness would need to be assessed through computational scaling experiments and validation against empirical market characteristics, rather than trading profitability alone.
Language: English
Open Access status: diamond
OpenAlex ID: W7234821054
FWCI:
Cites:
Cited by:
Related to: 10