主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2025, Vol. 33 ›› Issue (12): 1-12.doi: 10.16381/j.cnki.issn1003-207x.2022.0524cstr: 32146.14.j.cnki.issn1003-207x.2022.0524

• •    下一篇

计算实验与订单簿市场建模发展动态综述

崔毅安1, 韦立坚2, 熊熊3,4()   

  1. 1.深圳大学管理学院,广东 深圳 518055
    2.中山大学管理学院,广东 广州 510275
    3.天津大学管理与经济学部,天津 300072
    4.中国社会计算研究中心,天津 300072
  • 收稿日期:2022-03-18 修回日期:2022-05-07 出版日期:2025-12-25 发布日期:2025-12-25
  • 通讯作者: 熊熊 E-mail:xxpeter@tju.edu.cn
  • 基金资助:
    国家自然科学基金项目(72001149);国家自然科学基金项目(72141304);国家自然科学基金项目(72171239);国家自然科学基金项目(72531006);科技部重大专项课题(2022YFC3303304);广东省自然科学杰出青年基金项目(2021B1515020073);深圳大学青年教师科研启动基金项目(868-000001033404);深圳大学改革开放研究专题项目(868-000003022116)

Review of the Development of Agent based Modeling and Order Book Market Modeling

Yian Cui1, Lijian Wei2, Xiong Xiong3,4()   

  1. 1.College of Management,Shenzhen University,Shenzhen 518055,China
    2.School of Business,Sun Yat-sen University,Guangzhou 510275,China
    3.College of Management and Economics,Tianjin University,Tianjin 300072,China
    4.China Center for Social Computing and Analytics,Tianjin 300072,China
  • Received:2022-03-18 Revised:2022-05-07 Online:2025-12-25 Published:2025-12-25
  • Contact: Xiong Xiong E-mail:xxpeter@tju.edu.cn

摘要:

计算实验金融是人工智能在金融领域应用的重要方向,也是金融监管科技的关键手段,正逐渐发展为监管科技创新的先进工具,近年来受到学术界与业界的广泛关注。在计算实验研究范式的不断完善过程中,逐渐形成了包括个体行为建模、市场机制建模、模型校准三大研究方向。个体行为建模正逐渐沿着由简到繁(如何复现更多格式化特征),再到化繁为简(复杂模型是否可以解释更多市场动态)的思路展开探索;市场机制建模相对更易理解,其建模本身非常重要,研究者至今仍在持续致力于揭示在复杂市场涌现现象中个体行为与机制之间的交互机理;模型校准已逐渐成为计算实验研究的标配步骤,反映出计算实验研究范式正在不断走向规范与成熟。订单簿模型作为最贴近真实市场运作的交易机制方式,近年来已成为计算实验金融建模的研究热点。目前,基于计算实验的订单簿模型已能够复现实证研究中所观察到的部分格式化特征,这为构建与A股市场高度一致的订单簿机制模型提供了重要的可信性保证。

关键词: 计算实验, 订单簿市场, 行为建模, 机制设计, 模型校准

Abstract:

Agent based modeling is an important application of artificial intelligence in the financial sector and a key tool in financial regulatory technology. It is increasingly evolving into an advanced instrument for regulatory technological innovation and has garnered widespread attention from both academia and industry in recent years. As the agent based modeling research paradigm continues to mature, three major research directions have been formed: individual behavior modeling, market mechanism modeling, and model calibration. The research approach to individual behavior modeling has progressively evolved from simplicity to complexity (how to replicate more stylized facts) and then from complexity back to simplicity (whether complex models can explain more market dynamics). Market mechanism modeling is relatively easier to understand, and the modeling process itself is highly important. Researchers continue to dedicate efforts to uncovering the interactive mechanisms between individual behavior and market structures in complex emergent market phenomena. Model calibration has gradually become a standard step in agent based modeling research, reflecting the increasing standardization and maturity of the research paradigm. The order book model, as the trading mechanism most closely aligned with real market operations, has become a focal point of research in agent-based modeling in recent years. Currently, the agent-based order book models are already capable of replicating some of the stylized facts observed in empirical studies, providing important credibility assurance for constructing order book mechanism models that are highly consistent with the A-share market.

Key words: agent based model, order book market, behavioral modeling, market mechanism design, model calibration

中图分类号: