证券市场若干问题的实证研究

发布时间：2018-05-12 03:31

本文选题：金融物理 + 市场微观结构　；参考：《华东理工大学》2012年博士论文

【摘要】：自从进入二十世纪五十年代以后,金融市场的理论研究得到了迅猛的发展。同时,随着计算机的发展,使得金融研究人员对金融市场高频交易数据的实证研究成为现实。中国的金融市场是一个新兴的市场,同时也具有其不同于一般西方市场的特色,这使得对中国金融市场的研究成为必要。本论文以中国证券市场为研究对象,运用了金融物理学、计量经济学等方法并结合高频数据对市场的内部微观现象和结构进行了实证的研究。同时,本论文也对当今金融市场新的研究方法做了深入的研究和探讨,并提出了一套基于投资效用的市场投资理论以及基于该理论的投资策略。在对我国金融市场微观现象和结构的研究中,首先我们对我国沪深两市的买卖价差进行了研究。通过Lomb功率谱分析,我们验证了两市股票价差存在以日为周期的周期模式,并且该周期呈现“L”型的日内模式。同时,由该日内模式得到在每个交易日开盘后的一个小时内,两市的买卖价差都呈现幂率递减的现象,且其幂指数分别为βSHSE=0.20±0.067和βSZSE=0.19±0.069。而在对个股的日内模式的研究中也发现了同样的性质,且其幂指数大致服从正态分布。这说明了由累积的信息导致市场的价差增大是一个内生的动力学过程。第二,我们使用深圳证券交易所的超高频交易数据,对深证市场上交易者的撤单行为进行了研究。研究中我们将撤单的时间间隔作为研究的对象,并分别考虑了三种不同撤单的交易者的行为,即买撤单、卖撤单以及所有撤单。我们发现每支股票撤单间隔的分布密度都可以由韦伯分布函数来描述,且市场的撤单间隔表现出一个非常好的标度率的性质。在对其撤单间隔长度的分类讨论中撤单间隔表现出一个较强的记忆相关性。此外,我们在撤单间隔的数据中同样也发现了一个呈“八”字型的日内模式。通过降趋脉动分析法和多重分形降趋脉动分析发现,撤单间隔之间存在长程相关性和多重分形的性质,而撤单的日内模式对此并没有产生任何的影响。这些结果都说明了市场交易者撤单的行为并不是一个泊松过程,这在我们对指令驱动市场的建模中具有重要的参考作用。而在第二部分对基于事件时间的交易者的撤单行为的研究中,我们却得到了与连续时间下撤单行为不同的结论。首先,撤单的分布不再具有标度性,同时,交易者撤单的分布很好的服从了一个截尾的泊松分布,这些结论给基于事件研究的模型提供了较好的实证基础。第三,我们对市场交易者的交易积极性进行了研究。通过对深证指令簿订单数据的研究,我们发现：(1)交易指令簿的深度(包括已方指令簿的深度和对手方指令簿的深度)对交易双方的积极性都产生积极的影响,即当市场指令簿深度增加时,市场交易的积极性就会增加。但是由该因素所产生的影响非常有限,在模型中其影响几乎可以忽略不计；(2)大的买卖价差能够有效的降低市场交易者的下单积极性,这说明深市的交易者大多是成本厌恶型的投资者,当面临较大的交易成本时,他们更多的是选择承担风险；(3)指令簿中高频率的买单委托指令能够有效的激发买方交易者提交积极的买单指令,而低频率的卖单委托则能迫使卖方交易者提交积极的卖单指令；(4)不稳定的价格波动使得更多的交易者在更利于自己的价位上等待自己的交易时机,这说明了深市的交易者多为成熟的交易者,在出现股价剧烈波动的时候不轻易追涨杀跌。而在对当今金融市场新的研究方法的探讨中,我们首先引入了可视图的研究方法。通过产生分形布朗运动和多重分形随机游走的模拟实验,我们发现其对应的可视图的度分布呈现出幂率下降的行为,同时该幂函数的的幂指数和原序列的Hurst指数之间呈现出一个较好的线性关系。而在多重分形的性质并没有对结果产生太大的影响。在对中国沪深两市和香港股市的实证研究中,我们再一次验证了该关系。第二,我们在对金融市场资产收益的估值上引入了Levy-Roll的估值方法。对于他们的理论,我们相应提出了一套更为完善和细致的对该方法的鲁棒性的检验方法。在该方法通过我们的监测后,我们则使用该方法对整个市场的进行了实证检验,而该方法表现出了较之传统估计更为稳定性的收益率；最后,在对一组资产组合的实际应用中,该方法再一次展现了相比于使用样本参数等常规方法的优越性。从而我们认为,使用Levy-Roll的收益估值方法能够获得的资产的内在投资价值,同时在如今对CAPM一片的质疑声中,我们再一次肯定了CAPM在现代金融学中的核心地位。最后,我们提出了一套基于投资效用的惯序投资理论。在该模型下,我们指出在证券市场中,交易者的交易效用由市场的上升趋势β和投资者所使用的交易策略的准确率α所共同决定,且仅当两参数的总和大于1时,交易者才能在市场上获得正的预期投资收益效用。此外,我们还指出了在该模型下,当投资者对两个以上的资产进行惯序分析投资时,他能够得到大于对单个资产进行分析投资而带来的预期投资效用,且该效用随着投入资产数量的增加而增加,但其上限为两倍于对单支资产分析投资时所获得的投资效用。随后,我们在该交易理论的基础上提出了一个轮转投资策略,并提出了一套用来判断策略投资优劣的评价体系。在对沪深两市的实证研究中,使用轮转策略的投资收益能够轻松地跑赢所选用的两支股票,并且在随机策略的模拟测试中能够取得较高的轮转正确率和较低的平均惩罚收益。同时,通过计算4753对股票的投资结果,我们认为较少的轮转的能得到较低的平均惩罚收益。
[Abstract]:Since the 1950s, the theoretical research of the financial market has developed rapidly. At the same time, with the development of the computer, the empirical research on the high frequency transaction data of the financial market has become a reality with the development of the computer. The financial market in China is a new market, and it is also different from the general west. The characteristics of the market make it necessary to study the financial market in China. This paper takes the Chinese securities market as the research object, uses the financial physics, econometrics and other methods, and carries out an empirical study on the internal micro phenomena and structure of the market combined with high frequency data. At the same time, this paper also has a new research on the current financial market. The method has been thoroughly studied and discussed, and a set of market investment theory based on investment utility and investment strategy based on this theory have been put forward.
In the study of the microcosmic phenomenon and structure of China's financial market, we first studied the price difference between the two cities and Shanghai cities in China. Through the Lomb power spectrum analysis, we verified the periodic pattern of the two stock price difference in the daily cycle, and the period showed the "L" type in the day mode. In an hour after the opening of each trading day, the sale spreads of the two cities show a decline in the power rate, and the power exponents are respectively beta SHSE=0.20 + 0.067 and beta SZSE=0.19 + 0.069., respectively, and the same properties are found in the study of the intra day mode of the stock, and the power exponents are highly subordinate to the normal distribution. This shows the accumulated information. The increase of market spreads is an endogenous dynamic process.
Second, we use the ultra high frequency transaction data of the Shenzhen stock exchange to study the behavior of the traders' withdrawal in the Shenzhen stock market. In the study, we take the time interval of the withdrawal as the object of the study, and consider the behavior of the traders with three different kinds of withdrawal, namely, the purchase of the bill, the sale of the withdrawal and all the withdrawal. The distribution density of the stock withdrawal interval can be described by the Webb distribution function, and the withdrawal interval of the market shows a very good scale rate. In the classification discussion of the length of the withdrawal interval, the withdrawal interval shows a strong memory correlation. In addition, we also find that in the data of the withdrawal interval, we also find that An intraday model with a "eight" type is found. Through the analysis of the decline in the pulsation and the multifractal degradation, it is found that there is a long range correlation and a multifractal nature between the intervals of the withdrawal, and the internal model of the withdrawal has no effect on it. These results show that the behavior of the market trader is not the behavior of the withdrawal. A Poisson process plays an important role in the modeling of the instruction driven market. In the second part of the study on the behavior of the trader based on the event time, we get the conclusion that the withdrawal behavior is different from that in the continuous time. First, the distribution of the withdrawal is no longer in scale, while the trader withdraws. The distribution of the single well obeys a censored Poisson distribution. These conclusions provide a good empirical basis for the model based on event study.
Third, we have studied the trading enthusiasm of market traders. Through the study of the order data of the Shenzhen stock order book, we found that: (1) the depth of the transaction instruction book (including the depth of the late instruction book and the depth of the opponent's instruction book) has a positive impact on the enthusiasm of both parties, that is, when the market instruction book increases in depth. In addition, the enthusiasm of market trading will increase. However, the influence produced by this factor is very limited, and its influence in the model can almost be ignored. (2) the big sale price difference can effectively reduce the initiative of the market traders, which shows that most of the traders in the deep market are cost aversion investors, when they are faced with larger ones. When trading costs, they are more likely to take risks; (3) the high frequency purchase order instruction in the instruction book can effectively motivate the buyer to submit an active purchase order, while the low frequency sell orders can force the seller to submit a positive selling order; (4) the unstable price fluctuation makes more traders. Waiting for the time to deal with its own price shows that the traders in the deep market are mostly mature traders who are not easy to chase and fall when the stock price fluctuates.
In the discussion of new research methods in the current financial market, we first introduce a view based method. Through the simulation experiments that produce fractal Brown movement and multifractal random walk, we find that the corresponding degree distribution of the corresponding view shows a line of power rate decline, and the power exponent and the original sequence of the power function. There is a good linear relationship between the Hurst index, and the nature of multifractal does not have much effect on the results. In the empirical study of China's Shanghai and Shenzhen two cities and Hongkong stock markets, we once again verified the relationship.
Second, we introduce the Levy-Roll valuation method to the valuation of financial market asset returns. For their theory, we propose a more robust and detailed test method for the robustness of the method. After our monitoring, we use this method to empirically examine the whole market. The method shows a more stable yield than the traditional estimate; finally, in the actual application of a group of portfolios, the method once again shows the superiority of the conventional methods, such as the use of the sample parameters, so we believe that the intrinsic investment of the assets that can be obtained by using the Levy-Roll's income valuation method is considered. Value, at the same time, in today's questioning of CAPM, we once again affirmed the core position of CAPM in modern finance.
Finally, we propose a set of inertial sequencing investment theory based on investment utility. Under this model, we point out that in the stock market, the trading utility of traders is determined by the rising trend beta of the market and the accuracy of the trading strategy used by investors, and only when the sum of two references is more than 1 can the trader get the market. In addition, we also point out that under this model, when investors invest in more than two assets, he can get the expected investment effectiveness greater than the investment in the analysis of individual assets, and the utility increases with the increase in the number of invested assets, but the upper limit is two. Then, we put forward a rotation investment strategy on the basis of the transaction theory, and put forward a set of evaluation system to judge the advantages and disadvantages of the strategy investment. In the empirical study on the two cities and Shanghai cities, the investment income of the rotation strategy can easily win and win the choice. The two stocks used, and in the simulation test of random strategies, can achieve higher rotation accuracy and lower average penalty returns. At the same time, by calculating the investment results of 4753 stocks, we believe that less rotation can get lower average penalty returns.

【学位授予单位】：华东理工大学
【学位级别】：博士
【学位授予年份】：2012
【分类号】：F832.51;F224

【参考文献】