基于Agent的多股票连续双向拍卖市场仿真实验研究

发布时间：2018-04-23 04:02

本文选题：Agent + 人工金融市场　；参考：《华中科技大学》2012年硕士论文

【摘要】：基于Agent的行为金融学与投资行为的研究是当前经济学中比较热门的一个方向，通过对国内外相关研究现状的学习，发现目前大多数的建模都是单一风险资产且投资者Agent大多数采用期望效用理论。鉴于此，本文引入前景理论，建立一个基于Agent的多股票连续双向拍卖人工金融市场（人工市场）。研究市场由下至上出现的涌现现象、交易者策略、单股票和多股票市场差异。本文集中使用前景理论、Markowitz投资组合理论、单一心理账户行为资产组合理论建立了人工市场。该人工市场中前景理论交易者采用前景理论来刻画对财富变化的心理效用，使用行为资产组合理论在多股票市场进行投资组合；基础交易者采用期望效用理论来刻画财富的效用，使用Markowitz投资组合理论在多股票市场进行投资组合；随机交易者的价格预测和订单量确定都是随机的，在多股票市场中也没具体的投资组合。清算机制采用订单簿的形式来形成市场交易价格。使用Anylogic软件实现了人工市场仿真平台，通过仿真实验表明：本文的人工市场是有效的，宏观上市场价格具有波动性，收益率具有非正态性，以及各类交易者整体财富的变化情况符合人的主观感受。在此基础上进行了多股票人工市场交易者交易策略比较仿真实验，发现基础交易者和前景理论交易者都很“聪明”，能够不断的增加绩优股的持有量，，同时减少亏损股的持有量。而随机交易者则不能做到这一点。另外前景理论交易者比基础交易者能够做的更好，日均收益率更高，方差更小。最后比较单股票市场和多股票市场的差异，发现投资者在单股票市场上有可能比多股票市场收益率均值高，但要承担更大的风险。
[Abstract]:The research on behavioral finance and investment behavior based on Agent is a hot topic in current economics. It is found that most of the current models are single risk assets and most of the investor Agent uses the expected utility theory. In view of this, this paper introduces the prospect theory to establish a multi-stock continuous two-way auction artificial financial market (artificial market) based on Agent. To study the emergence of market from the bottom up, traders' strategies, single stock and long stock market differences. This paper focuses on the use of Markowitz portfolio theory and the single psychological account behavioral portfolio theory to establish the labor market. In this artificial market, the prospect theory is used to describe the psychological utility of the change of wealth, and the behavioral portfolio theory is used to carry out the portfolio investment in the multi-stock market. Base traders use expected utility theory to describe the utility of wealth, use Markowitz portfolio theory to carry out portfolio investment in multi-stock market, and stochastic traders' price forecasting and order quantity determination are both random. There is also no specific portfolio in the long stock market. The liquidation mechanism adopts the form of order book to form the market price. The artificial market simulation platform is realized by using Anylogic software. The simulation results show that the artificial market is effective, the market price is fluctuating and the yield is non-normal. And all kinds of traders overall wealth changes in line with the subjective feelings. On this basis, a comparative simulation experiment of trading strategies of multi-stock artificial market traders is carried out. It is found that both basic traders and foreground theory traders are "smart" and can continuously increase their holdings of high-grade shares. At the same time reduce the holdings of loss-making stocks. Random traders can't do that. In addition, the foreground theory trader can do better than the basic trader, the daily average yield is higher, the variance is smaller. Finally, comparing the differences between the single stock market and the multi-stock market, it is found that the average yield of investors in the single stock market may be higher than that in the multi-stock market, but they have to bear more risks.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F713.359;F830.91;TP391.9

【参考文献】