[论文解读] Econophysics: Empirical facts and agent-based models

Anirban Chakraborti, Ioane Muni Toke+2 more|arXiv (Cornell University)|Sep 10, 2009

Complex Systems and Time Series Analysis参考文献 144被引用 49

一句话总结

本文综述了经济物理学中的实证事实与基于主体的模型，聚焦于金融时间序列、订单簿动态及财富分配。文中提出的基于主体的模型能够再现市场现象，如肥尾分布与波动率聚集，表明随机性学习策略在大规模系统中比确定性策略更具效率（f̄ ≈ 0.8）。

ABSTRACT

This article aims at reviewing recent empirical and theoretical developments usually grouped under the term Econophysics. Since its name was coined in 1995 by merging the words Economics and Physics, this new interdisciplinary field has grown in various directions: theoretical macroeconomics (wealth distributions), microstructure of financial markets (order book modelling), econometrics of financial bubbles and crashes, etc. In the first part of the review, we discuss on the emergence of Econophysics. Then we present empirical studies revealing statistical properties of financial time series. We begin the presentation with the widely acknowledged stylized facts which describe the returns of financial assets- fat tails, volatility clustering, autocorrelation, etc.- and recall that some of these properties are directly linked to the way time is taken into account. We continue with the statistical properties observed on order books in financial markets. For the sake of illustrating this review, (nearly) all the stated facts are reproduced using our own high-frequency financial database. Finally, contributions to the study of correlations of assets such as random matrix theory and graph theory are presented. In the second part of the review, we deal with models in Econophysics through the point of view of agent-based modelling. Amongst a large number of multi-agent-based models, we have identified three representative areas. First, using previous work originally presented in the fields of behavioural finance and market microstructure theory, econophysicists have developed agent-based models of order-driven markets that are extensively presented here. Second, kinetic theory models designed to explain some empirical facts on wealth distribution are reviewed. Third, we briefly summarize game theory models by reviewing the now classic minority game and related problems.

研究动机与目标

• 整合经济物理学中的实证发现与基于主体的建模方法，弥合物理学与经济学之间的鸿沟。
• 探究金融市场的统计特性（如肥尾分布与波动率聚集）如何从主体互动中涌现。
• 评估基于主体的模型在再现真实市场行为方面的解释力与校准能力。
• 探讨学习、策略选择与随机决策在实现高效市场结果中的作用。
• 识别在基于主体框架下建模多维订单簿与系统性风险方面的开放性挑战。

提出的方法

• 对高频金融数据进行实证分析，以重现如肥尾收益与波动率聚集等典型事实。
• 应用随机矩阵理论与图论研究金融市场中跨资产的相关性。
• 基于动理学理论与反应-扩散过程原理，开发用于订单驱动市场的基于主体模型。
• 实施少数者博弈框架，以模拟战略行为与资源配置的效率。
• 使用随机学习算法，其中主体根据过往结果（如 n(t)）更新策略，实现在 N⁰ 或 ln N 时间内的收敛。
• 在大规模 N 极限下，比较确定性策略与随机策略在效率（f̄）与收敛速度方面的表现。

实验结果

研究问题

• RQ1 典型事实如肥尾分布与波动率聚集如何从金融市场的微观层面主体互动中产生？
• RQ2 在大规模基于主体系统中，随机学习策略在实现高效结果方面发挥何种作用？
• RQ3 订单簿的基于主体模型在现实性、校准与解释力之间如何取得平衡？
• RQ4 为何在少数者博弈类设置中，某些‘更简单’的策略会优于更复杂的策略？
• RQ5 当前基于主体模型在捕捉多维市场动态与系统性风险方面存在哪些局限性？

主要发现

• 利用高频金融数据，可稳健地再现如肥尾收益与波动率聚集等典型事实。
• 随机并行学习策略在 O(1) 或 O(ln N) 时间内收敛，显著优于需 O(N) 时间的确定性策略。
• 表现最佳的随机策略实现了 f̄ ≈ 0.8 的效率指标，而确定性策略的效率较低。
• 在系统效率方面，天真或较不复杂的策略常优于复杂且避群策略。
• 当前订单簿的基于主体模型在处理多资产多维交易方面仍存在局限，是关键的开放挑战。
• 2007–2008 年金融危机加剧了人们对基于主体模型作为理解系统性风险与市场不稳定性工具的兴趣。