臺灣博碩士論文加值系統

由於國內股票市場近幾年的持續低迷，國債收益率也不高，再加上如基金、期貨等投資管道皆不理想，民眾的儲蓄很難轉化為投資，國內資本市場需要更快地建設、更徹底地改革。但投資人如何在目前的投資環境下，仍挑選出能在低的投資風險下來獲取穩定報酬的股票投資組合呢？
為了找出能在低風險下獲取穩定報酬的投資組合，本研究欲利用資料探勘技術以及加投資風險的來建立一個多層次投資組合篩選機制，以便提供投資人作為投資股票市場擇股的考量。本文章主要分為五個步驟進行，第一、先以R*來預測下一投資期間的買賣決策。第二、以MSCI成分股與全額交割股之所有財務比率進入類神經－敏感度分析，以篩選攸關財務體質優劣之財務比率進入決策樹訓練。第三、利用決策樹形成之準則所形成標準化擇股程序以及帶入福斯特．佛萊斯(Foster Friess)投資大師的觀念篩選出我們需要的優良與劣質公司。第四、以灰關聯分析對第三步驟所選出我們需要的優良與劣質公司做擇股優先順序。最後，再利用修正後的歷史模擬法預測由優良或劣質的公司所組成的投資組合，冀望在預期買賣決策下都具有低風險高報酬的預期目標。本研究以台灣所有上市公司(剔除金融相關產業)做為研究對象，實證期間為2001年11月至2005年9月，共操作了11次。研究結果顯示經由本研究所選出的各投資期間的投資組合平均年報酬率為100.5%，且平均年報酬率明顯的優於大盤的平均年報酬率。

Because of continuously recessive stock market performance locally, low fund profit and inferior investment channels for funds and futures, it is quite difficult for people to convert their bank savings to investment. Anyway, our local capital is urgently required for speedier construction and innovation. The question is about how investors can still availably obtain stock investment portfolio combinations with stable remuneration under low investment risks like our current business climate.
To find the investment portfolio combinations with stable remuneration but low risks, this research is aimed to adopt the data mining technique added with investment risks to construct a multi-tier screening mechanism for investment portfolio combinations so that it is available for investors' consideration about stock selection within the stock market. This article is mainly preceded through 5 steps. First, we firstly use R* to prediction the buy-sell decision within an investment period. Second, we adopt the entire financial ratio from MSCI component stock and Full Delivery for neural sensitivity analysis so that we can screen out the financial ratios with superior or inferior physiques for decision tree training. Third, we adopt the standardized stock selection procedures reached by principles to the decision tree formation into the concepts proposed by the investment mater, Foster Friess, so that we can find out the superior or inferior companies. Fourth, we adopt the grey Relational Analysis to rank the stock selection orders between the superior and inferior companies. Finally, we further adopt the modified Historical Simulation (VaR) method to predict the investment portfolio combinations formed by the said inferior and superior companies. It is our expectation all buy-sell decision can be featured with the low risks but high remuneration within the predicted goals. This research makes a research on the experimental subjects from all public companies within the stock market (excluding the banking companies) in the duration from November 2001 to September 2005. We totally executed 11 bouts of operation. Research results indicate the average annual return rate is 100% for various investment portfolio combinations during each period. Also, our annual return rate is significantly better than that of the overall stock market.

Chinese Abstract....................................................................................................................i
English Abstract...................................................................................................................ii
Table of Contents..................................................................................................................v
List Of Tables......................................................................................................................vii
List of Figures....................................................................................................................viii
Chapter 1 Introduction........................................................................................................1
1.1 Background.........................................................................................................1
1.2 Motivation and purposes...................................................................................2
1.2.1 Motivation........................................................................................................2
1.2.2 purposes...........................................................................................................2
1.3 Subjects and Limitations...................................................................................3
1.3.1 Subjects............................................................................................................4
1.3.2 Limitations.......................................................................................................4
1.4 Thesis Structure..................................................................................................5
Chapter 2 Literature Review...............................................................................................7
2.1 Data Mining........................................................................................................7
2.2 Risk Analysis of Historical Simulation.............................................................9
Chapter 3 Research Methodology.....................................................................................11
3.1 Sensitivity analysis by Neural Network..........................................................12
3.2 Decision tree category......................................................................................17
3.3 Grey relational analysis...................................................................................19
3.4 Brief of historical distributions.......................................................................22
Chapter 4 Research Designs.............................................................................................28
4.1 Option study.....................................................................................................28
4.2 Sensitivity analysis...........................................................................................29
4.3 Entering Foster Friess’s concepts into decision tree....................................31
4.4 Putting into Grey relational Ranking............................................................33
4.5 Measurement of risk value.............................................................................34
Chapter 5 Analysis of Empirical Results........................................................................36
5.1 Empirical operation of this study...................................................................36
5.2 Sensitivity Analysis of financial ratios...........................................................38
5.3 Selective portfolios of this study.....................................................................39
Chapter 6 Conclusion and Research Suggestion............................................................43
6.1 Conclusion........................................................................................................43
6.2 Research and Suggestion................................................................................46
References..........................................................................................................................48

1.B.S. Ahna , S.S. Chob , C.Y. Kimc , (2000) , The integrated methodology of rough set theory and artificial neural network for business failure prediction .
2.Berry, M. J. A. and Linoff, G., (1997) , Data Mining Technique For Marketing, Sale
,And Customer Support, Wiley Computer.
3.Beaver, W. H., “Financial Ratios as Predictors of Failures”, Journal of Accounting
Research, Vol.4,No. 1, 1966, pp71-127.
4. Ball, R. and P.Brown, “An Empirical Evaluation of Accounting Income Numbers”, Joural of Accounting Research 6, 1968,pp.159-178.
5. C. E. Lee, M. Pan , and Y. ANGELA LIU (2001) , On Market Efficiency ofAsian Foreign Exchange Rates: Evidence from A Joint Variance RatioTest and Technical Trading Rules .
6. David C. Parkes , Bernardo A. Huberman , (2001) , Multiagent Cooperative Search for Portfolio Selection .
7. Francis E.H. Tay , Lixiang Shen , (2002) , Economic and financial prediction using rough sets model .
8. Frank Witlox , Hans Tindemans , (2004) , The application of rough sets analysis in activity-based modeling .Opportunities and constraints .
9. Hull, John C. (2000) , Options, Futures, & other Derivatives , Prentice-Hall International,inc .
10. Kyung-Shik Shin , Yong-Joo Lee , (2002) , A genetic algorithm application in bankruptcy prediction modeling .
11. Yukiko Oritoa , Hisashi Yamamotob , Genji Yamazakib , (2003) , Index fund selections with genetic algorithms and heuristic classifications .