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

Integrating Macroeconomic Indicators and Financial Reports to Predict Stock Price ReturnTaiwan Stock Market’s Electronic Sector as Example

施正宏(Jackie Shih)

指導教授 ： 李維平

中原大學/商學院/資訊管理研究所/碩士(2004年)

https://doi.org/10.6840/cycu200400520

摘要

股票投資是相當普遍與重要的一種金融活動，投資人作股票投資皆希望能夠提高報酬並掌控風險，因此，過去已有許多的學術研究，期望能找出準確預測股價漲跌的方法，以確實降低投資風險並提高報酬。
股價的預測一般有兩大類的研究方向，分別為基本分析與技術分析兩種，在基本分析學派認為個別公司的股價表現應該反映該公司的營運狀況，因此，過去的研究常常以公司的財務角度切入，試著找出個股財報資料與股價的關聯性。但是，我們覺得光以個股財報資料來做預測是不夠的，因股價的表現除了反映公司的營運以外，也跟大環境的經濟狀況息息相關；因此，我們認為應同時考量外在的總體經濟榮枯(總體經濟指標)與內在的公司營運狀況(個股財報資料)，才能更準確的預測個別公司的股價走勢。
此外，在本研究問題所選用的預測技術方面，過去的學者大多選用類神經網路技術來作股價預測的研究。雖然類神經網路有不錯的預測能力，但是類神經網路所產生出來的預測結果不容易解讀，易流於「黑箱作業」，使得投資人對這樣的技術使用有所忌憚。因此，本研究嘗試利用資料探勘裡的決策樹技術作為研究工具，因為決策樹模型所產生的投資規則，容易被解讀與檢測，更能增加投資人對此技術使用的信心。
我們以國內的電子類股資料，作了六個循環的實驗，依據總經與財務指標的預測所擬定的投資策略，可以獲得平均8.025%的季報酬率，高於單純利用個股財報資料的預測結果所得的平均2.285%的季報酬率，更遠高於同時期大盤指數6.684%平均季漲幅與電子指數4.954%的平均季漲幅。研究結果顯示加入總體經濟指標除了比單純利用財報資料作預測的準確度更高外，也獲得更高的投資報酬率。

關鍵字

股價預測 ； 總經指標 ； 基本分析 ； 決策樹 ； 財報資料

並列摘要

Stock Investment is a very common and important financial activity nowaday. The main objectives for stock investors are to reduce risk while increasing return at the same time. Hence, numerous researches have been done in attempt to detect the systematic patterns in stock prices and help investors to accomplish these tasks.
Stock prediction methods can be categorized into two major categories, fundamental and technical analysis. Fundamental stock analyst thought stock price should reflect its company’s profitability and s/he took a close examination of the financial statements of the company. However, we don’t think it is sufficient. Stock price performance not only reflects company’s profitability, it is also coefficient with economic cycle. In this research, we will integrate macroeconomic indicators and financial statement data to predict stock price trend.
While there had been a fair number of researches done on predicting stock returns, most of them are using Artificial Neural Networks（ANN）as its prediction technique. Although ANN can perform well in financial data prediction, their algorithms tend to be a black box. In this research, we apply a data mining technique, called CART, to predict stock return. The input variables and dependent variable are comparable to the data set we would use in multiple linear regression or discriminant analysis. Thus, it can increase investor’s confidence on stocks picked.
We had done six cycles of experiment using stocks list in Taiwan Stock Exchange’s electronic sector. The investment strategy based on integrating macroeconomic indicators and financial statement data can give us an average of 8.025% quarterly return. Whereas the average quarterly return for predicting with pure financial statement data, Taiwan Stock Index, and Electronic Sector Index are 2.285%, 6.685%, and 4.954% respectively. Our model evidently outperformed pure financial statement data model and stock index.

並列關鍵字

Financial Statement Data ； Stock Prediction ； Macroeconomic Indicators ； Fundamental Analysis ； Decision Tree

參考文獻

7. 財政部證券暨期貨管理委員會網站 www.sfc.gov.tw

1. [Mohamed L. Hambaba, 1996] Mohamed L. Hambaba, “Intelligent Hybrid System for Data Mining”; Computational Intelligence for Financial Engineering, 1996., Proceedings of the IEEE/IAFE 1996 Conference on , 24-26 March 1996

2. Kurt Thearling www.thearling.com

3. [Ou, J. and Penman H. (1989)], “Financial Statement Analysis and the Prediction of Stock Returns”, Journal of Accounting and Economics, Vol. 11

5. [Sorensen, Keith, Ooi（2000）], “The Decision Tree Approach to Stock Selection – An evolving tree model performs the best”, Journal of Portfolio Management