運用類神經網路、投資標的基本面和技術面分析決定最佳投資組合

Application of neural network techniques integrating fundamental and technical analysis for optimal portfolio

楊伶升(Ling-Sheng Yang)

指導教授 ： 吳泰熙

國立臺北大學/商學院/企業管理學系/碩士(2009年)

摘要

在2008年全球金融風暴和通貨膨脹的壓力下，如何全方面管理存款以外之金融商品投資，已成為一重要議題，亦為本研究所欲探討的方向。本研究應用人工智慧技術結合現代投資組合理論，並考量投資人實際投資心理，以期建構出一套結合實務和理論之投資管理流程。
Markowitz平均數—變異數投資組合模型架構所建立之效率前緣，其將股價的非預期上漲或下跌皆視為風險，此種現象與投資人實際想法不同。再者，因影響股價和證券市場波動之因素相當廣泛，本研究利用類神經預測模型，以期能預測出股價未來走勢，進而決定投資之時機和消弭非預期中產生損失之風險。由於證券投資中包含四個部分：股票擇時、交易策略、股票選擇和投資組合。因此，本研究利用類神經網路之預測能力，及藉由證券市場線和下方風險MV模型導出最佳投資組合，並將公司財務基本面指標，包括股東權益報酬率、毛利率和現金流量成長率作為選股策略，和實際進行股票市場中之技術指標，即RSI、KD和MACD指標所形成之交易策略應用於本研究模型中，以期建立一個完整的投資流程和一個良好之投資組合，其績效指標可適當的反映期望報酬率和風險之間的均衡關係。
本研究所建構之模型實際驗證為2005年1月至2008年12月之歷史資料，其表現與台灣資訊科技指數、科技型基金和台灣加權指數相較下，本研究所建構之投資組合模型各季之投資績效，不論是投資報酬率和夏普指標表現都較為優異。藉由本研究之投資流程，建構出一套可供投資人參考之投資決策系統和完整投資流程。

關鍵字

類神經網路 ； Markowitz投資組合 ； 半變異數 ； 交易策略 ； 基本面分析

並列摘要

In 2008, global financial crisis influenced the whole worlds’ economic in a negative way. The behaviors of investors also changed during that time. Pure saving did not have positive return due to the inflation; on the other hand, financial products need to be chosen carefully during the depression. Thus, how to manage our investment well becomes a subject we want to discuss in this study. This research not only combines together the modern portfolio theorem and the artificial intelligence, but also takes investors’ behavior into account in order to construct a process and system of investment and find out the optimal portfolio as the ultimate goal.
The mean-variance model originally introduced by Markowitz assumes that the total return of a portfolio can be described using the mean return of the assets and the variance of return (risk) between these assets. However, for any portfolio which lies on this efficient frontier, the phenomenon of considering both unexpected rising and declining of stock price as risk does not match the idea of the public. In order to improve the drawback of traditional Markowitz mean-variance model, which does not meet the actual concern of investors, the base of downside risk of the theory of Markowitz portfolio model is applied in this study. In this study, we apply not only the downside risk of Markowitz mean-variance model, but also neural network to find out the optimal portfolio. Our purpose in this study is to construct an investment system and discover the optimal portfolio for the general investors. In order to meet this goal, we apply both fundamental financial analyses and technology indicators in our investment process. Fundamental financial analyses include return on equity, gross margin ratio, and cash flow growth ratio. RSI index, MACD index, and KD index are applied as technical indicators in this study. We expect to use the downside risk of Markowitz mean-variance model, neural network, fundamental analyses, and technology indicators to build an investment system and find out the optimal portfolio which can reflect a balance between return rate and risk.
In this study, we apply a simulation experiment which uses the historical data from January, 2005 to December, 2008. The results of the experiment show that in terms of the return rate or Sharpe ratio, the performance of our portfolios have gained more profit than the Taiwan Information Technology Index, Technology funds, and Taiwan Weighted Stock Index.

並列關鍵字

Neural network ； Markowitz portfolio model ； fundamental and technology analyses

參考文獻

施正宏（2004），結合總體經濟指標及個股財報資料以預測個股漲跌—以台灣電子類股為例，中原大學資訊管理研究所未出版之碩士論文。

黃光廷（2002），技術分析、基本分析與投資組合避險績效之研究，國立成功大學會計學研究所所未出版之碩士論文。

Abarbanell, J. and Bushee, B. (1997), ‘Fundamental Analysis, Future Earnings, and Stock Prices.’ Journal of Accounting Research, 35: 1-24.

Amir, E. and Lev, B. (1996), ‘Value-Relevance of Nonfinancial Information: The Wireless Communications Industry.’ Journal of Accounting and Economics, 22, 3-30.

Bawa, V.S. and Lindenberg, E.B. (1977), ‘Capital Market Equilibrium in a Mean-Lower Partial Moment Framework.’ Journal of Financial Economics, 5, 189-200.