透過您的圖書館登入
IP:3.139.233.43
  • 學位論文

應用級比檢驗建立擇股策略整合決策模型

An Application of Class Ratio in Integrated Stock Selection Model

指導教授 : 周宗南
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


近年來國民所得與理財知識的提升,使得台灣證券交易市場交易活絡,投資金融商品以創造財富,儼然已成為全民運動,其中又以電子產業最吸引投資人目光;對未來股價的預測,亦成為投資人興趣的焦點。但股價的詭譎多變受到許多因素的影響,一般投資人又無充份的資訊及技術去分析股價,作出正確的買賣決策;為了找出獲取穩定報酬的投資組合,供投資人作為投資股票市場擇股的考量,本研究希望透過客觀的財務比率建構出一套有效且具有科學依據的擇股模型,幫助投資散戶在眾多家公司中選擇體質較好的公司,獲得最佳獲利保障與風險控管。 本研究以台灣證券交易所電子工業類股之上市公司為研究對象,研究期間為2001年6月至2007年9月,資料型態以季資料為週期,資料來源為臺灣經濟新報資料庫(TEJ),包含一般財務比率資料與股價資料。運用類神經網路、灰色預測及灰色決策,並以級比檢驗作為模型整合之依據建立整合模型進行綜合比較,其結果如下: 1.不論運用何種模型,所得之平均報酬率均高於台灣加權股價指數、電子類股價指數甚至五大銀行之三個月存款利率,顯示無論選擇任何模型,均能帶給投資人的投資報酬率遠高於定期存款利率,提高財富真實價值。 2.經級比檢驗後之整合模型的投資組合,報酬績效均較原基本模型之投資組合大幅提升,由此可知灰色預測模型中利用級比檢驗降低誤差,對報酬率之提升確實有極大的幫助,同時結合灰色局勢決策所建構的擇股模型預測可發揮最大之綜效。 3.各模型中以模型B於每季所選出的前十家公司做次數統計時,發現台積電(2330)被挑選之建議次數高達22次為最高。 4.若以各模型報酬率超過全體樣本次數方式評估比較,模型B為所有模型中績效最好的模型,但其投資報酬率卻比模型A低。結果說明較多次數勝率不一定會獲得較高之報酬率。

並列摘要


In recent years, the increase of personal income comes up with the dissemination of financial knowledge have brought prosperity for the securities trading market in Taiwan. The prediction of stock price for electronic industry companies in particular attracts the attention of most personal investors. However, the rapid changing environment of stock market tends to be complex for a personal investor to understand and follow. For the purpose of assisting the personal investors to construct profitable portfolios, this study aims to employ financial ratios of companies and computational approaches to construct efficient stock selection model for the investors to acquire maximum returns. This research work collected both financial ratios and stock price data of the electronic industry companies listed in Taiwan Stock Exchange during the period between June 2001 and September 2007 for the portfolio evaluation. Three basic models including Artificial Neural Network, Grey Prediction and Grey Decision combined with two novel integrated models based on the cross ratio criteria were applied to select promising stocks for portfolios. The experimental results show that all models performed marked average returns and exceeded the defined benchmarks. Meanwhile, the outcomes of the integrated models surpassed the basic models and demonstrated the effectiveness of the cross ratio strategy. The Taiwan Semiconductor Manufacturing Company (TSMC) reached 22 times been selected as promising stocks in portfolios among all companies of the list. The performance of models also showed higher frequency of outperformance does not determine maximum investment return.

參考文獻


2.朱國珍(2004),「信評式債券型基金之評價-通用迴歸類神經網路模式之應用」,臺中健康暨管理學院經營管理研究所碩士論文。
15.黃光廷(2001),「技術分析、基本分析與投資組合避險績效之研究」,國立成功大學會計系研究所碩士論文。
1.Amilon, H. (2001) “A Neural Network Versus Black-Scholes : A Comparison of Pricing and Hedging Performance,” Working Paper, Dept. of Economics, Lund University
4.Markowitz ,H.M,(1952), Portfolio selection, ” Journal of Finance, pp.77-91,Mach。
5.Markowitz ,H.M,(1959), “Portfolio selection : Efficient Diversification of Investments ” , New York : John Wiley & Sons。

被引用紀錄


葉美均(2016)。應用基因演算法整合五大構面選股策略〔碩士論文,朝陽科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0078-1108201714023954

延伸閱讀