關聯分析運用於股票投資建議之實證研究

The empirical study of correlation analysis applied to the stock investment recommendation

鍾國誠(Guo-Cheng Chung)

指導教授 ： 吳建臺

國立虎尾科技大學/管理學院/經營管理研究所/碩士(2011年)

摘要

摘要
現在許多財經網站都會將專業投資機構所發表的投資建議公告在相關網站上，近期的研究發現廣泛地買進分析一致推薦的股票可帶來超額報酬，但是大部分投資者並沒有足夠的閒置資金能廣泛的買進所有看好的股票，若是他們選錯投資建議的股票就可能造成重大損失，如何分辨股票投資建議的準確度與可靠性也是眾多投資者期盼得知的問題。
本研究希望透過資料探勘之關聯分析將知名財經網站鉅亨網內的投資建議透過資訊擷取與分析的方式，運用SQL Server 2008在完整研究年限、牛市、熊市三方面進行分析並且產生各別的高關聯性投資評等，且再檢驗由此所形成的投資組合的報酬率能否帶給投資者超額報酬，最後與大盤績效表現和其它重要指數做比較看能否勝過它們的績效。
研究結果得知聽從各市場中高關聯性評等所形成的投資組合報酬率依序為在牛市2.78%最高、熊市0.63%次之、至於完整研究期間-0.12%則是最低，在牛市中參考投資建議的確有較佳的獲利績效，但與大盤表現和其它重要指數互相比較後的結果顯示，不能明顯勝過大盤和其它重要指數的績效表現。如不聽從投資建議轉而進行指數化投資方式最高可以分別在3、7、9、12月份獲取明顯高於聽從的最高獲利2.78%，但僅在1、5、8月份明顯低於聽從的最低獲利-0.12%，經過聽從與不聽從的報酬率比較後可以得知，對於長期投資者而言進行指數化投資的方式獲利表現會較好。

關鍵字

投資建議 ； 資料探勘 ； 關聯分析 ； 投資組合 ； 超額報酬

並列摘要

Abstract
　　Nowaday,Many financial website show the investment recommendations expressed by professional investment institutions in to the relevant website,Recent studies have found that it’s easy to appear excess returns if purchasing analysis recommended extensive’s stock unanimously,But most of the investors don’t have enough idle funds to buy a wide range of all favorite stocks,The bad investment advice stock choosing by them may cause significant losses,The stock investment advice accuracy and reliability are the questions which many investors hope to know.
This study will get the well-known investment advice on financial website by through data mining’s association analysis information retrieval and analysis methods. Three fields in study - study period、Bull market、Bear market, was used SQL Server 2008 to conduct analysis and generate the other high relevance investment rating. After that, testing the effect of Remuneration levels on investor’s excess Remuneration. Finally,comparing between the market Efficiency performance and other important indices to know whether their Efficiency is better.
Research results suggest that the Remuneration levels maximum 2.78% in a bull market,The second is in bear market with 0.63%,Full of period study is the lowest -0.12%, invesment recommendation referenced in bull market has good profit Efficiency. But the results after comparing it with the broader market and other key indicators show that: it can not significantly higher than the broader market indices and Efficiency performance of other important indicators performance, this table showed that in March、July、September and December, the profit which investor would got if they followed investment recommendation to conduct indexing investment model was higher than profit they would got if they didn’t follow 2.78%,but in January、May、August, this case was opposite -0.12%,from these situation, long-term investor can get profit better when they conduct indexing investment model.

並列關鍵字

Investment advice ； Data Mining ； Correlation Analysis ； Portfolio ； Excess return

參考文獻

[14] 黃旭輝、許惠婷，2004，”分析師的推薦股票可以實際獲利嗎？”，財金論文叢刊，1期，101~127頁。

[6] 邱麗如，2005，”以布林表格為基礎的資料挖掘方法及其應用”，靜宜大學資訊管理研究所碩士論文。

[16] 謝邦昌，”Microsoft SQL Server 2005 Data Mining 演算法–Association”學術會議PPT報告，中華資料採礦協會出版，2~88頁。

[23] Agrawal, R., Imielinski, T. and Swami, A. N., 1993, “Mining Association Rules between Sets of Items in Large Databases”, Proceeding of ACM SIGMOD International Conference on Management of Data, pp.207-216.

[24] Azzi, S., Bird, R., Ghiringhelli, P. and Rossi, E., 2005, “Biases and Information in Analysts’ Recommendations: The European Experience”, Journal of Asset Management, 6(5), pp.345-380.