纯统计配对交易目标与股票价差统计特征关系分析-analysis on the relationship between pure statistical matching trading target and statistical characteristics of stock price difference.docx

纯统计配对交易目标与股票价差统计特征关系分析-analysis on the relationship between pure statistical matching trading target and statistical characteristics of stock price difference.docx

纯统计配对交易目标与股票价差统计特征关系分析-analysis on the relationship between pure statistical matching trading target and statistical characteristics of stock price difference

哈尔滨工业大学管理学博士学位论文哈尔滨工业大学管理学博士学位论文AbAbstract-- PAGE IV III -对交易结果有明显提高,但二级行业配对存在人为排除优良股票对的可能性。在既定的 GGR 交易策略下,股票对选取标准对交易结果是有影响的,但最高相 关系数、最小方差、最小 SSD 三个股票对选取标准的等效性较高,选出的股票 对多为二级行业配对的股票对,其中银行业股票对在这三个选取标准中表现最 为显著。所以行业配对、市值配对等基本面配对方法所得到的股票对,最终会 在价差统计特征上表现出独特的性质,故通过纯统计方法能有效而迅速地选出 基本面配对方法的股票对。另外,通过纯统计配对方法寻找最优股票对和改进 交易策略将是未来研究的重点。为研究建仓策略对配对交易年收益率影响,选用并确定下文纯统计配对交 易目标与价差统计特征关系研究的建仓策略,验证和比较沪深港三个市场的有 效性,本文构建了 WM-FTBD 折回首日建仓改进策略,结合 GGR 和 Herlemont 策略,在沪深港证券市场交易,运用三种检验方法,从数理和实证上得出,FTBD 策略成功率和年收益率都更高,同时发现深市年收益率呈离散、尖峰和正偏性, 经三种交易策略检验,深圳市场的年收益率最高,上海次之,香港最低,且差 异明显。结果表明,有效建仓策略总体可改进收益,但也承担更多风险;价差 动量效应和均值回复效应有助于解释价差变化和收益率差异;配对交易在成熟 有效市场不一定适合,但在发展中国家将有广阔发展空间。为比较回归方法和重要模型,提出纯统计配对交易目标与价差统计特征关 系的多元非线性回归模型,讨论并指明模型各变量及其系数的解释意义,本文 探讨了配对交易年收益率主要决定因素,构造平仓次数、成败标记、价格序号 距离三个指标,结合八个现有指标,执行 FTBD 交易策略,在沪深港市场运用 逐步回归和限定自变量个数最大 R 方回归法寻找并确立了年收益率 WM 六因素 多元非线性回归模型,同时发现了不同市场影响收益率的主要因素及个数有区 别。WM 模型整体理想,除峰度外其他自变量系数符号稳定性都较好。结果表 明,进行股票对选取时,应尽量选择持仓时间长、平仓次数多、成功率高且方 差大、相关系数又不是很大的股票对,结合两种回归方法将更有利于最优模型 的构建和检验。关键词:配对交易;统计套利;对冲;统计特征;融资融券AbstractThe world capital market is expanding and evolving rapidly. The regulations of the capital markets of the countries and the world are adjusting and declining. The stock markets’ trends in every country are still difficult to grasp. Hedge will be one of the most important parts in investment. Pairs trading is one of the easy and important way in hedge, developing very fast in developed and developing countries. Shorting and margin financing in China will become reality soon. Pure statistical pairs trading is the new kind of pairs trading that mainly used the statistical methods to build and choose the stock pairs. It have flexible modes and needs not to collect auxiliary information. It will not be infected by the auxiliary information, such as industry, equity ownership structure, market capitalization. It emphasizes the main analysis object of stock price and makes the base of the research more easy and directly.To fully understand the research situation on the pairs trading at

您可能关注的文档

文档评论(0)

1亿VIP精品文档

相关文档