本文關(guān)鍵詞：股票組合投資的穩(wěn)健統(tǒng)計分析方法研究　出處：《暨南大學(xué)》2016年博士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： 組合投資 穩(wěn)健統(tǒng)計 均值-方差模型 夏普指數(shù)模型 絕對離差 離群值

【摘要】：獲取較高的收益一直是證券投資的最根本目的,但是在投資活動中,收益與風(fēng)險總是相伴而行,收益越高,則風(fēng)險越大,收益越低,則風(fēng)險越小。構(gòu)建投資組合的主要目的就是分散風(fēng)險,即在風(fēng)險和收益之間到一個平衡點,在風(fēng)險一定的情況下,以期獲得最大收益。本文主要以投資組合模型為研究對象,其最傳統(tǒng)且最經(jīng)典的理論是由Markowitz在1952年提出的均值-方差模型,該模型以證券歷史收益率的方差作為投資組合的風(fēng)險度量,被廣大投資者和研究者所使用,該模型在理論和實際的應(yīng)用中均具有非常重要的意義。隨著研究的深入,不少研究者發(fā)現(xiàn)使用方差來度量風(fēng)險可能存在不可回避的缺陷。為了克服現(xiàn)有理論的不足以及更好地運用投資組合模型獲取較高收益,很多專家和學(xué)者進行了廣泛而深入的探討。本文從穩(wěn)健的角度對投資組合模型進行了研究,并且對幾個模型進行了穩(wěn)健改進,最后對改進后的模型進行了模擬和實證分析。本文的研究成果主要有以下幾個方面:一、穩(wěn)健統(tǒng)計方法的構(gòu)建。由于證券市場中存在離群值,在構(gòu)建穩(wěn)健的投資組合方法之前我們對常用的一種統(tǒng)計方法與穩(wěn)健思想結(jié)合,構(gòu)建出穩(wěn)健的統(tǒng)計方法,而傳統(tǒng)的多元分析方法與許多傳統(tǒng)方法一樣,容易受到離群值的影響,導(dǎo)致計算結(jié)果與實際情況產(chǎn)生差異,因此本文構(gòu)建了穩(wěn)健因子分析方法,并使用證券數(shù)據(jù)對穩(wěn)健因子分析和傳統(tǒng)因子分析進行模擬研究和實證分析,從模擬和實證結(jié)果可以看出,我們構(gòu)建的穩(wěn)健方法比傳統(tǒng)方法更能有效抵抗離群值。二、均值-方差模型的改進。對于均值-方差模型來說,其正態(tài)性是極其重要的一個假設(shè)。在收益與風(fēng)險權(quán)衡中,當(dāng)證券收益數(shù)據(jù)服從正態(tài)分布時,其方差和均值才是度量風(fēng)險、收益最好的統(tǒng)計量。傳統(tǒng)的投資組合方法在構(gòu)造統(tǒng)計量時沒有考慮統(tǒng)計量的穩(wěn)健性,對離群值非常敏感,本文結(jié)合穩(wěn)健統(tǒng)計的思想,進一步對均值-方差投資組合模型進行改進,使其能夠更能滿足我們?nèi)粘Ｉ钪械玫降慕鹑陬悢?shù)據(jù),并且對離群值具有更高的抵抗作用。而且從實證結(jié)果中可以看出,我們構(gòu)建的穩(wěn)健組合投資方法的確比傳統(tǒng)方法更優(yōu)。三、均值-絕對離差模型的改進。相對于其他投資組合模型來說,均值-絕對離差模型中的絕對離差是一個比較穩(wěn)健的統(tǒng)計量,但是計量期望收益率的是均值,而均值不穩(wěn)健,所以本文對這個模型中的均值使用穩(wěn)健均值進行改進,得到穩(wěn)健均值-絕對離差模型,并且使用中國證券數(shù)據(jù)進行比較分析。由分析結(jié)果可知,我們構(gòu)建的穩(wěn)健均值-絕對離差模型比傳統(tǒng)的方法更能抵抗離群值的影響。四、夏普指數(shù)模型的改進。傳統(tǒng)的回歸方法中對每個樣本數(shù)據(jù)均賦予相等的權(quán)重,從而使得離群值對整個模型的影響增強,因此本文我們結(jié)合穩(wěn)健統(tǒng)計的思想對夏普指數(shù)模型進行改進,即在回歸分析中我們使用了穩(wěn)健回歸的理念,對樣本數(shù)據(jù)賦予不同的權(quán)重,殘差越大,權(quán)重越小,殘差越小,權(quán)重越大,這樣能夠有效降低離群值對整個模型計算結(jié)果的影響,結(jié)合夏普指數(shù)模型,構(gòu)建了穩(wěn)健夏普指數(shù)模型,使得組合投資更加趨向于它真正的投資價值。由實證分析結(jié)果可以看出,改進的夏普指數(shù)模型有較好的抗差性。五、穩(wěn)健組合投資系統(tǒng)的原型建立。本文將統(tǒng)計方法與智能化信息系統(tǒng)結(jié)合起來,在開源系統(tǒng)R語言的基礎(chǔ)之上,建立了一套“穩(wěn)健組合投資分析系統(tǒng)”,來實現(xiàn)系統(tǒng)的算法與系統(tǒng)評價中圖形的繪制,從而最終設(shè)計出具有一定實用價值的穩(wěn)健組合投資系統(tǒng)。
[Abstract]:To obtain higher returns has been the most fundamental purpose of securities investment, but in the process of investment, profits and risks are always accompanied by the line, the higher the income, the greater the risk, the income is low, the risk is smaller. The main purpose is to construct the investment portfolio risk diversification, to a balance between risks and the income, risk in certain circumstances, to obtain the maximum benefit. This paper takes the portfolio model as the research object, the most traditional and most classic theory is put forward by Markowitz in 1952 mean - variance model, the variance of the model by the securities history profit rate as the portfolio risk measure, the use of the majority of investors and researchers, and this model has very important significance in theory and practical application. With further research, many researchers found that the use of variance to measure the risk can not be avoided In order to overcome the existing defects. The shortage of theory and better use the portfolio model to obtain a higher income, many experts and scholars have carried out extensive and in-depth study. This paper from the perspective of robust portfolio model is studied, and several models were robust improved, finally the improved model is analyzed and simulated empirical research. The research results of this paper are mainly the following aspects: first, constructing robust statistical methods. Due to the presence of outliers in the stock market before the investment portfolio construction method for a robust statistical method commonly used in combination with robust ideas, construct robust statistical method, multivariate analysis method and traditional and many the traditional method, is easily affected by the outliers, leading to the differences of calculation results and the actual situation, this paper constructs a robust factor analysis method, and The use of securities data on robust factor analysis and traditional factor analysis and empirical simulation study, from simulation and experimental results show that the robust method we constructed can effectively resist outliers more than the traditional method. Two, the improved mean variance model. The mean variance model, the normality is a that is extremely important. In the tradeoff between risk, when the stock returns are normally distributed, the mean and variance is a measure of risk, return the best statistics. The traditional method of investment portfolio in the structure of the statistic when did not consider the robustness of statistics, is very sensitive to outliers, combined with robust statistical methods further, the mean variance portfolio model is improved, so that it can better meet the financial data obtained in our daily life, and has higher resistance to outliers . and from the empirical results can be seen that the robust portfolio approach we construct is superior to traditional methods. Three, the mean absolute difference from the improved model. Compared with other portfolio model, mean absolute deviation model of absolute deviation is a statistic relatively robust, but the measurement of expectations rate of return is mean, while the mean is not robust, so the mean in this model using robust mean was modified to obtain robust mean absolute deviation model, and use the Chinese stock data were compared and analyzed. According to the analysis results, we construct the robust mean absolute deviation model can resist the influence of outliers more than the traditional method. Four, improve the SHARP index model. The traditional regression method for each sample data are given equal weight, so that outliers impact on the entire model In this paper we combine enhanced robust statistical methods to improve the SHARP index model, namely in the regression analysis we used robust regression theory, and gives different weights to the sample data, the greater the residual weight, smaller residual is smaller, the greater the weight, which can affect the results of the outlier model the decrease, combined with SHARP index model, constructs a robust SHARP index model, the portfolio investment tend to its real investment value. By the empirical analysis results show that SHARP index improved model has better robustness. Five, robust portfolio system prototype is established. In this paper, with intelligent statistical method information systems are combined on the basis of the open-source R language system, established a robust portfolio analysis system ", to realize the algorithm and system evaluation in Figure Painting In the end, a robust portfolio investment system with a certain practical value is designed.

【學(xué)位授予單位】：暨南大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2016
【分類號】：F832.51;F224
，

本文編號：1414221