【作者】 李成刚；

【导师】 田益祥；

【作者基本信息】 电子科技大学 ， 金融工程， 2012， 博士

【摘要】 Markowitz和Tobin提出均值-方差模型，标志着现代证券投资组合理论发展的开端。从此，各国学者对均值-方差模型进行了深入的研究，以均值-方差模型为基础建立了不同的投资组合，提出了不同的投资策略。另一方面，金融市场微观结构理论的发展，为证券投资组合理论的发展提供了新的发展方向。作为市场微观结构理论的一个核心变量——定单流，不仅具有明确、直观的涵义，而且能够刻画资金的流向，具有丰富的信息含量，反映了投资者的投资组合再平衡行为。因此，本文利用定单流刻画资金的流向，选择投资股票和板块，构建基于定单流的证券投资策略，分析投资策略的风险，为证券投资者的投资策略提供参考，指导投资者的投资决策。首先，本文引入定单流刻画资金的净流入和净流出，根据资金的流向选择股票和板块，提出了基于定单流的股票和板块选择方法。采用事件研究法分析了证券分析师推荐股票和板块的总体特征。根据这些特征，采用朴素贝叶斯分类（NaiveBayes Classifier，简称NBC）方法分别将深证成指指数股票和所有板块分为符合这些特征的股票和板块以及不符合这些特征的股票和板块，筛选出符合这些特征的股票和板块，计算其收益率，并与证券分析师推荐的股票和板块收益率以及深证成指指数收益率进行比较分析。实证结果显示，基于定单流的股票和板块选择方法获得的收益率比分析师推荐的股票和板块收益率更高，也比指数的收益率更高，并能获得显著的超额收益率。其次，本文将定单流引入证券投资策略的构建中，提出了基于定单流的静态投资策略。从投资者期望效用最大化角度，将定单流引入投资组合模型，根据定单流指标确定组合权重，构建基于定单流的证券投资组合模型。在只含有风险资产以及同时含有风险资产和无风险资产两种情况下，通过数学推导，得到基于定单流的证券投资组合模型的最优投资权重。在此基础上，提出基于定单流的静态投资策略，根据定单流指标确定投资策略的最优权重。选取深市A股前30只股票进行实证分析，结果表明根据定单流指标确定投资权重，能取得比均值-方差模型和市场指数更高的投资收益。再次，考虑多个投资时期，从投资者期望效用最大化角度，引入定单流指标建立含有交易成本的多期动态投资组合模型。在各个投资时期，根据定单流冲击系数动态调整组合权重。采用数学推导求解动态组合投资模型的最优权重。然后，提出基于定单流的动态投资策略，根据定单冲击系数动态调整投资策略的投资权重。选取深市A股前30只股票进行实证分析，结果表明基于定单流的动态投资策略不仅能跑赢市场指数，而且能获得比均值-方差投资组合更好的投资收益。最后，本文分析了基于定单流的证券投资策略的风险。基于定单流，提出定单冲击系数，构建定单冲击系数与收益率的二元GARCH模型，检验定单冲击系数与收益率之间的波动溢出效应。在此基础上，根据定单冲击系数的波动分析收益率的波动，分析基于定单流的证券投资策略的风险。将深证综指指数股票分为高定单冲击系数组合与低定单冲击系数组合，比较不同定单冲击系数组合的风险，并进行评价。结果表明，收益率与定单冲击系数之间存在双向波动溢出效应；高定单冲击系数组合的收益率比低定单冲击系数组合更高，但是其风险也更大，说明本文构建的基于定单流的证券投资策略在获得高收益的同时，也承担着高风险。评价结果显示，与低定单冲击系数投资组合相比，高定单冲击系数投资组合能获得更高的风险溢价。更多还原

【Abstract】 Markowitz and Tobin proposed the Mean-variance model, marking the beginningof the development of modern portfolio theory. Since then, scholars from variouscountries have deeply studied the Mean-variance model, constructed differentportfolios based on the Mean-variance model and proposed different investmentstrategies. On the other hand, the development of financial market microstructuretheory provides a new direction for the development of portfolio theory. As a corevariable of the market microstructure theory, order flow not only has a clear andintuitive meaning, but also can characterize the flow of funds, has a wealth ofinformation content, and reflects investor’s portfolio rebalancing. Therefore, this paperuses order flow to depict the flow of funds, choose stocks and plate to invest, build onthe portfolio investment strategies based on order flow, and analyze the risk ofinvestment strategies. This can provide a reference for the equity investors and guideinvestors’ investment decisions.Firstly, this paper introduces order flow to depict the net inflow and net outflow offunds, and selects stocks and plates according to the flow of funds, proposes the stockand plate selection method based on order flow. This paper uses event study to analyzethe general characteristics of stocks and plates that securities analysts recommend.Based on these characteristics, this paper uses the Naive Bayes Classifier (NBC) toclassify Shenzhen Component Index stock and all the plates into sections in line withthese characteristics or not. We screen stocks and plates that meet these characteristics,calculate their return, and compare it with the return of stocks and plates that analystsrecommend and the Shenzhen Component index return. The empirical results showthat the return obtained by the stock and plate selection method based on order flow ishigher than the stocks and plates analysts recommend. Also, it is higher than the indexreturn, and can gain significant excessive return.Secondly, this paper introduces order flow into the establishment of securitiesinvestment strategy, and puts forward the static investment strategy based on orderflow. From the perspective of investors’ expected utility maximization, this paper introduces order flow into the portfolio model, determine the portfolio weightsaccording to order flow indicator, and build the portfolio model based on order flow.Through mathematical derivation, we obtain optimal weights of the securityinvestment portfolio under two conditions of only contained risky assets and containedboth risk assets and risk-free asset. On this basis, the paper proposes static investmentstrategy based on order flow, and determines the optimal weight of the staticinvestment strategy according order flow indicator. Selecting30stocksfrom the Shenzhen A shares to empirically analyze, the results show that determiningthe investment weights according to order flow indicator, we can obtain higherinvestment return than the Mean-variance model and the market index.Thirdly, considering various investment period, from the perspective of investors’expected utility maximization, this paper introduces order flow indicator, buildsmulti-period dynamic portfolio model with transaction costs. In each Investmentperiod, we dynamically adjust portfolio weights according to the order flow impactcoefficient. We use mathematical derivation to solve the optimal weights of thedynamic portfolio model. Then we propose the dynamic investment strategy based onorder flow, and dynamically adjust the investment weights according to order flowimpact coefficient. Selecting30stocks from the Shenzhen A shares for empiricalanalysis, the results show that the dynamic investment strategy based on order flow notonly can beat the market index, but also can get higher return than the Mean-varianceportfolio.Finally, this paper analyzes the risk of the securities investment strategies basedon order flow. Based on order flow, we propose the order impact coefficient, andconstruct the binary GARCH model of order impact coefficient and return to testvolatility spillover between order impact coefficient and return. On this basis, weanalyze the volatility of return according to the volatility of order flow impactcoefficient, then analyze the risk of the securities investment strategies based on orderflow. The Shenzhen Composite Index stocks are divided into high and low orderimpact coefficient stock portfolios. This paper compares the risk of different orderimpact coefficient stock portfolios, and does an evaluation. The results show that thereis bi-dictional volatility spillover effect between the return and order impact coefficient.The return of high order impact coefficient stock portfolio is higher, but its risk is higher. This indicates that the securities investment strategies based on order flowconstructed in this paper can gain higher return, it also bears a higher risk meantime.Evaluation results show that compared with low order impact coefficient stockportfolio, high order impact coefficient stock portfolio can gain higher risk premium.更多还原