摘要
传统的金融学研究方法为“由上及下”,此类方法倚重于总体把握,而忽视个体行为,特别是缺乏个体以及个体与环境之间的互动。基于Agent的计算金融是一种新的研究金融市场行为的工具。首先阐述了金融市场本身的复杂性和传统金融学存在的一些困惑。在这基础上提出了基于Agent的计算金融方法。其次,概要介绍了目前基于Agent的人工金融市场的主要设计方法,并且提出了在设计过程中要注意的一些问题。最后,分析了这种方法相对于传统金融学研究方法的优点和存在的问题,并进一步提出了该领域的未来研究方向。
Compared with the bottom up approach used in agent - based computational finance, traditional finance emphasizes particularly on the equilibrium of the whole markets, while ignoring the action of individuals, especially the interaction between the individual and environment. Agent - based computational finance is a new method for simulating actual financial market. Firstly, this paper concentrates on the character of complexity in finance market and analyses the disadvantages of traditional finance compared with this simulation approach. By surveying research on agent - based models in finance, this paper illuminates the main aspects of designing agent - based artificial financial market as well as some noticeable issues in the process of designing it. Finally, based on the conclusions of the advantages of agent - based computational finance as well as its drawbacks, this paper gives the prospect in this research field.
出处
《计算机仿真》
CSCD
2007年第2期262-265,共4页
Computer Simulation
关键词
计算金融
人工金融市场
复杂性
机器学习
Computational finance
Artificial financial markets
Complexity
Machine learning
