| 研究生: |
李惠妍 Li, Hui-Yen |
|---|---|
| 論文名稱: |
類神經網路與迴歸模式在台股指數期貨預測之研究 The Application of Artificial Neural Network and Regression Model for Studying the Taiwan Stock Index Future |
| 指導教授: |
溫敏杰
Wen, Miin-Jye 吳宗正 Wu, Chung-Cheng |
| 學位類別: |
碩士 Master |
| 系所名稱: |
管理學院 - 高階管理碩士在職專班(EMBA) Executive Master of Business Administration (EMBA) |
| 論文出版年: | 2003 |
| 畢業學年度: | 91 |
| 語文別: | 中文 |
| 論文頁數: | 56 |
| 中文關鍵詞: | 迴歸分析 、臺灣加權股價指數期貨 、類神經網路 |
| 外文關鍵詞: | Taiwan stock index future, Regression analysis, Artificial neural network |
| 相關次數: | 點閱:206 下載:4 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
國內證券金融市場於民國87年7月21日正式推出台灣發行量加權股價指數期貨契約,為台灣的金融業自由化及國際化,建立一新里程碑,提供了投資者新投資商品和避險工具,也提供投機客及套利者,以少量資金賺取較大利潤的機會。尤其近幾年來期貨交易熱度愈來愈沸騰,期貨市場的發展潛力愈來愈不可忽視。
有鑑於類神經網路是近年來快速竄起的資訊處理技術,尤其是運用在金融財務方面,都績效卓著。所以本研究嘗試運用類神經網路及統計方法中的迴歸分析,來預測台股指數期貨的隔日收盤指數,以尋求出最適宜的預測模式。本研究結果發現:
1.在倒傳遞類神經網路方面,可發現有隱藏層的模式較無隱藏層的模式預測績效佳,而隱藏層處理單元的數目,以總和法所求出的最為適合,且經計算求得測試期的MAE 為72.866 與MSE 為10231.49。
2.在迴歸分析方面,以去除三筆偏離值,再經由逐步迴歸分析篩選後之當日收盤指數、基差、漲跌、10 日威廉氏指標(W%R)、5 日乖離率(BIAS)等5 種變數的模式最為適合,且經計算求得測試期的MAE 為72.878 與MSE 為9709.518。
3.在改良式類神經網路方面,以逐步迴歸分析篩選後之5 種變數,做為輸入變數,可發現無隱藏層的模式似乎較有隱藏層的模式預測績效稍佳,但差異不大,最後以總和法所求出的最為適合,且經計算求得測試期的MAE 為73.441 與MSE 為10075.31。
4.在三種不同模式的預測績效比較方面,發現迴歸分析績效最好,其次是改良式類神經網路,而預測績效最差的是倒傳遞類神經網路。因此可以發現,變數較多並不一定預測效果較好,可能反而造成一些雜訊,減少了部分變數的影響效果。改良式類神經網路就是將變數精簡,以減少變數間的抵銷效果,而達到較佳的預測效果。
Taiwan stock market formally established Taiwan Stock Index Future Contract on July 21,1998, it could be said a milestone for Taiwan finance career`s liberalization and internationalization.It supplies investors the new investment product and hedging risk tools,and also supplies speculators and arbitragers the opportunity of making more profit at less fund. Especially, the more higher futures transactions,the more potential of futures market in these few years.
Artificial Neural Network is a tool of information technique that rapidly rises in these few years, especially using in finance area, the performance is very outstanding. So the study tries to use Artificial Neural Network and Regression Analysis of Statistical methods in order to predict the next day closing index of FITX,and then find the most suitable prediction model to create the most large rate of gaining profit. The results of the study are :
1. At the aspect of the Back-Propagation Network :
It could be found that the model with hidden layer is better than the model without hidden layer for the prediction performance. About the quantity of the processing element of the hidden layer, the amount method is the most suitable method. After calculation,the MAE of test period is 72.866 and the MSE of test period is 10231.49.
2. At the aspect of Regression Analysis :
By roguing three outliers and selecting through step by step Regression Analysis, the five variables of the closing index,basis difference,up and down,10 days W%R and 5 days BIAS are most suitable.After calculation,the MAE of test period is 72.878 and the MSE of test period is 9709.518.
3. At the aspect of improving Artificial Neural Network :
The five variables selected from using step by step Regression Analysis method are regard as input variables. It could be found that the model without hidden layer is better than the model with hidden layer for the prediction performance, but the difference is small. In a short word, the model of amount method is the most suitable model. After calculation,the MAE of test period is 73.441 and the MSE of test period is 10075.31.
4. At the prediction performance of three models comparison aspect :
It could be found the performance of improving model of Artificial Neural Network is good, Regression Analysis is much better, but Artificial Neural Network is the worst. So it is not certainly better when variables are more but create too much complex, on the contrary, decrease some effects between some variables. Improving model of Artificial Neural Network is to simplify variables in order to decrease effects and get better prediction performance.
中文部分
1. 王淑芬(1995),投資學,華泰書局。
2. 李曉隆(2002),出租公寓之租金價格預測-複迴歸分析與類神經網路的
比較,台灣科技大學企業管理研究所碩士論文。
3. 杜金龍(2002),技術分析在台灣股市應用的訣竅,金錢文化出版公司。
4. 何宜鍵(1997),上市股價報酬率及波動性之預測-類神經網路及多元迴
歸模型配適性之研究,中正大學企業管理研究所碩士論文。
5. 吳宗正、溫敏杰、侯惠月(2001),類神經網路及統計方法在台股指數期
貨預測研究之比較,成功大學學報,第36 卷,人文社會篇,頁91-109。
6. 吳宗正(2002),投資技術分析,華泰文化事業公司。
7. 吳宗正(2003),迴歸分析,華泰書局。
8. 周嚴(1990),期貨投資學,華泰書局。
9. 故國瑜(1996),類神經網路產業盈餘預測及其投資策略之研究-以電子
電機及紡織業為例,政治大學資訊管理研究所碩士論文。
10. 陳松男(1997),現代投資學,新陸書局。
11. 陳順宇(1997),迴歸分析,華泰書局。
12. 張道行(1998),使用類神經網路以短期技術指標作次日股價趨勢預測之
研究,論文研究。
13. 游淑禎(1998),類神經網路應用於臺灣股市預測,臺灣銀行季刊,第
49 卷,第3 期,頁27-60。
14. 葉怡成(2002),應用類神經網路,儒林書局。
15. 葉怡成(2002),類神經網路模式應用與實作,儒林書局。
16. 詹錦宏、袁澤峻(1999),景氣循環與股價指數之預測--類神經網路之應
用,證交資料,450 卷,頁1-15。
17. 楊雅媛(2002),迴歸分析與類神經網路預測能力之比較,政治大學統計
研究所碩士論文。
18. 蔡依玲(2001),台灣股票市場報酬率之研究,成功大學統計研究所碩士
論文。
19. 蔡瑞煌(1995),類神經網路概論,三民書局。
20. 蔡榮裕(1999),現貨盤後期貨交易資訊內涵之研究,輔仁大學金融研究
所碩士論文。
21. 蔡璧如(1997),股票指數期貨交易實務,寰宇出版社。
22. 鄭超文(1989),投資技術分析,貞德書局。
23. 劉明德(1993),期貨與選擇權-理論、實務與策略,故鄉出版社。
24. 臺灣期貨交易所網站http://www.taifex.com.tw/
英文部分
1. Bergerson,K. and Wunsch,D.C.(1991), A Commodity Trading Model Based on a Neural Network-expert System Hybrid ,IJCNN-91,I,pp.289-293.
2. Chiang,W.C., Urban,T.L.and Baldridge,G.W.(1995), A Neural Network Approach to Mutual Fund Net Aasset Value Forecasting , Omega, Int. J. Mgmt. Sci.Vol.24, No.2,pp.205-210.
3. Grudnitski,G.and Osburn,L.(1993), Forecasting S&P and Gold Future Prices : An Application of Neural Networks, Journal of Futures Markets, Vol.13, No.6,pp.631-643.
4. Huang,Yu Chuan(2002), Trading Activity in Stock Index Futures Markets: The evidence of emerging markets, The Journal of Futures Markets, New York,Vol. 22, Iss. 10,pp.983-1003.
5. Kimoto,T. and Asakawa,K.(1990), Stock Market Predication System with Modular Networks, Proc.IEEE int.Conf.Neural Networks,1, pp.1-6.
6. Kimoto,T. and Yoda,M.(1993), Buying and Selling Timing Predication System for Stocks Based on Modular Neural Networks, Fujitsu Sci.Tech.J.,29(3),pp.257-264.
7. Nag ,Ashok K.(2002), Forecasting Daily Foreign Exchange Rates Using Genetically Optimized Neural Networks,Journal of Forecasting, Chichester, Vol. 21, Iss. 7,pp.501-511.
8. Nelson,M.M.(1991), A Practical Guide to Neural Nets,Addison-Wesley.
9. Rumelhart,D.E.,Hinton,G.E.and Williams,R.J.(1986), Learning Internal Representation by Error Propagation, Rumelhart, D.E.and McClelland,J.L.〔Eds.〕,Parallel Distributed Processing : Explorations in the Microstructure of Cognition,vol.1,pp.318-362,MIT Press,Cambridge MA..
10.Skapura,D.M.(1995), Building Neural Networks,pp.137-140.
11.Warner,B.and Misra,M.(1996), Understanding Neural Networks as Statistical Tools, The American Statistician, Vol. 50, No. 4, pp.284-293.
12.Zhang, G. and Hu, M.Y.(1998), Neural Network Forecasting of the British Pound/US Dollar Exchange Rate, Omega, Int. J. Mgmt. Sci. Vol.26, No.4,pp.495-506.