【作者】 杨浩；

【导师】 詹文杰；

【作者基本信息】 华中科技大学 ， 管理科学与工程， 2012， 硕士

【摘要】 在双向拍卖市场中，由于信息的不对称，买卖双方都面临着十分复杂的交易决策过程。即使在这样的环境下，双向拍卖市场通常可以产生90%以上的市场效率，但是这种高效的资源配置特性并不总是存在，在某些供需曲线下双向拍卖的市场效率会非常低。本文将交易策略中标高和时间两个维度作为切入点，探究如何消除供需曲线形状对市场资源配置影响。本文首先详细分析了传统双向拍卖实验、在线双向拍卖实验和双向拍卖仿真实验各自特点，说明了选择计算机仿真的必要性。在深入分析经典交易策略模型的基础上，用C语言设计并实现相关算法，仿真速度快，可以在短时间内完成大量仿真实验。为了充分挖掘每个交易策略的特性，本文从标高维度和时间维度进行相应的优化和设计：在标高维度上，考虑到智能交易策略需要经过一段时间的学习才能达到最优状态，因此构建了连续的实验回合环境；在时间维度上，设计出一个报价时机模型，该模型可描述不同交易策略的报价时机，借助这一模型可以在仿真算法中加入报价时机机制。通过仿真实验发现，GD交易策略可以消除供需曲线形状对市场资源配置影响，无报价时机的GD交易策略也可以消除供需曲线形状对市场资源配置影响，有报价时机的ZIP交易策略不能完全消除供需曲线形状对市场资源配置影响。最后从市场微观角度进行分析，给出了标高消除供需曲线影响的定性解释，同时发现了报价时机对报价次数有显著影响。更多还原

【Abstract】 In the market of double auction, due to the asymmetry of information, both the buyersand sellers are faced with a very complex offering and decision-making process. Even insuch an environment, a double auction market will generally produce more than90%of themarket efficiency, but the characteristics of efficient allocation of resources are not alwaysexisting. In some supply and demand curves, the efficiency of a double auction market willbe very low.This paper puts the markup dimension and time dimension of trading strategies as thebreakthrough point, and explores how to eliminate the influence of the shape of supply anddemand curves on the market allocation of resources. This paper firstly analyzes thecharacteristics of the traditional double auction experiment, online double auctionexperiment and double auction simulation experiment in detail and explains the necessity ofchoosing a computer simulation. On the basis of deeply analyzing the model of classicquotation strategy, C language is used to design and implement related algorithms, and thespeed of simulation is fast so that a large number of simulation experiments can be finishedin a short time. In order to fully dig the characteristics of each trading strategies, this paperconducts a corresponding optimization and design from the aspect of markup and timedimension. In the aspect of markup dimension, considering the intelligent trading strategiesneed a period of learning to achieve the optimal state, thus the continuous experimentperiod environment is constructed. In the aspect of time dimension, a model of quotationtime is designed, which can be used to describe the quotation time of different tradingstrategies, and with the aid of this model, we can add the mechanism of quotation time intothe simulation algorithm.The results of experiment shows that GD strategy can eliminate the influence of theshape of supply and demand curves on the market allocation of resources, and so do the GD strategy with no quotation time. But ZIP strategy with quotation time cannot completelyeliminate the influence of the shape of supply and demand curves on the market allocationof resources. Finally from microscopic angle of the market, the qualitative explanation ofhow the markup plays a key role is given, and at the same time, the significant effect ofquotation time on the quotation numbers is also found.更多还原