近年來隨著銀行公告利率愈來愈低且通貨膨脹率高攀不下，導致國內投資風氣愈來愈盛行，但投資人並沒有太多時間專研於股票或期貨市場，共同基金是將資金交由專業基金經理人代為管理及操作，已成為國內最盛行之投資工具之ㄧ。
平滑支撐向量迴歸屬於較新穎之類神經網路工具，其學習理論能廣泛被應用於預測及分類模型方面。本研究主要探討以投資人角度利用平滑支撐向量迴歸及倒傳遞類神經網路來預測共同基金之報酬率，且依據預測值挑選出表現較佳之基金，建構出基金報酬率之預測模型，進而提供投資人於作投資決策時之參考依據。經實證研究發現:
1.經由運用平滑支撐向量迴歸或是倒傳遞類神經網路預測模型之選擇基金策略後，其平均績效皆優於大盤表現。另外，若未來ㄧ年為多頭市場，則可獲得較高之超額報酬﹔若為空頭市場，則預測模型之績效表現則略優於大盤績效。
2.於預測漲跌幅方向方面，平滑支撐向量迴歸模型與倒傳遞類神經網路之預測結果皆高達75%以上，其作為投資人一參考依據。
3.本研究之管理意涵在於，平滑支撐向量迴歸預測模型適合於風險承受度較低之投資人﹔另外，倒傳遞類神經網路則較適用於愛好風險之投資人。投資人可依自己風險承受度選擇適合自己之預測模型，以創造最大之超額報酬。

In recent years, with the decline of interest rate and the inflation, people begin to invest their property in many financial instruments instead of deposit in the bank. However, most investors do not have too much time to take care their underlying; they start to invest in mutual fund that provides fund manager to manage investor’s position. Mutual fund has become one of the popular financial instruments in Taiwan.
The smooth support vector regression is the latest developed neural network. Its origin is Machine Learning theory which is a very powerful and effective method to classify and predict the unknown samples. The goal of the study is to use smooth support vector regression and back propagation network, the two latest neural network approaches to predict the annual return of mutual fund and picks 10 funds which rank in top 10 among the prediction. By the implement of smooth support vector regression and back propagation network, right investment is accessible. The conclusions of this research are presented as following:
1.By applying smooth support vector regression and back propagation network in the prediction model can beat the market. In addition, if it is bull market in the following year, we can earn more return than if it is bear market.
2.With the aspect of prediction of highs and lows, both smooth support vector regression and back propagation network are above 75% accuracy.
3.With the aspect of management, smooth support vector regression can earn smaller premium return and take lower risk, it is suitable for the conservative investors; back propagation network earns more premium return but take higher risk, so it is suitable for the investors who prefer higher risk. Investors should evaluate their own attribute and choose the suitable tools as their predict model to earn maximum profit.

[1]張文信(1995)，「以類神經網路預測股價指數漲跌」，國立台灣大學財務金融學研究所碩士論文。
[2]于鴻潔(1997)，「台灣共同基金淨資產價值的預測—類神經網路之應用」，國立政治大學統計研究所碩士論文。
[3]林曉雯(1997)，「類神經網路在台灣股市投資之應用—指標選取與回饋式網路架構之建立」，國立台灣大學資訊管理學研究所碩士論文。
[4]葉怡成(1997)，應用類神經網路，儒林圖書有限公司。
[5]陳育聖(1998)，「利用類神經網路預測股票型共同基金之淨值與績效」，國立台灣大學資訊管理所碩士論文。
[6]楊政麟(1998)，「運用類神經網路於股價指數套利之研究」，國立台灣科技大學資訊管理研究所碩士論文。
[7]葉怡成(2000)，類神經網路模式應用與實作，儒林圖書有限公司。
[8]劉濠葦(2000)，「運用類神經網路於共同基金績效之研究」，國立東華大學企業管理學系碩士論文。
[9]鄭淑娥(2001)，「台灣開放型共同基金績效分類之研究」，國立成功大學統計學系碩士論文。
[10]杜金龍(2002)，技術指標—在台灣股市應用的訣竅，財訊出版社。

[11]許意鈴(2002)，「共同基金淨值之預測—灰色理論、類神經網路及適應性類神經模糊理論系統之應用」，國立台灣科技大學資訊管理研究所碩士論文。
[12]黃綺年(2003)，「統計方法與類神經網路應用於國內開放式股票型基金投資績效分類及投資報酬率預測之研究」，國立成功大學統計學系碩士論文。
[13]李凌均(2004)，「基於支持向量機的機械設備狀態趨勢預測研究」，西安交通大學學報，第38卷第3期，230-234頁。
[14]簡憶如(2005) ，「倒傳遞類神經網路於單一國家股票型基金淨值預測之應用－以亞洲四小龍為例」，大葉大學國際企業管理學系碩士論文。
[15]周聖鈞(2005) ，「應用類神經網路於組合型基金之研究」，國立台北科技大學商業自動化與管理研究所。
[16]陳寬裕(2006) ，「結合遺傳演算法與支援向量迴歸於台灣股票加權指數之預測」，計量管理期刊，第3卷第1期，1-18頁。
[17]賴佳君(2006) ，「應用平滑支撐向量基於台指期貨套利之研究」，國立台灣科技大學資訊管理研究所碩士論文。
[18]周俊宏(2006) ，「運用類神經網路與支撐向量機於個人信用卡授信決策之研究」，國立台灣科技大學資訊管理研究所碩士論文。
[19]陳志龍(2006) ，「運用類神經網路與技術指標預測股票型基金漲跌及交易時機之研究－以臺灣50指數股票型基金為例」，朝陽科技大學資訊管理學系碩士論文。
[20]Chiang, W.-C and Urban T.L. (1995), ”A Neural Network Approach to Mutual Fund Net Asset Value Forecasting”, International Journal of Management Science, Vol.24, No.2 , pp.205-215.
[21]Cybenko, G.. (1989), ”Approximation by Superpositions of A Sigmoidal Function”, Mathematics of Controls, Signals and Systems, Vol.2, No.4, pp.303-314.
[22]Fama, E.F. (1972), ”Components of Investment Performance”, Journal of Finance, Vol.27, No.3, pp.551-567.
[23]Fan, A. and Palaniswami, M. (2001), ”Stock Selection Using Support Vector Machines”, IEEE Computer Society, Vol.3, pp.1793-1798.
[24]Henriksson, R.D. and Merton, R.C (1981), ”On Market Timing and Investment Performance II: Statistical Procedures for Evaluating Forecasting Skills”, Journal of Business, Vol.54, pp.513-534.
[25]Hornik, K.M. Stinchcombe and White, H. (1989), ”Multilayer Feedforward Networks are Universal Approximators”, Neural Networks, Vol.2, No.5, pp.359-366.
[26]Hunag, W., Nakanori Y. and Wang S.-Y. (2004), ”Forecasting Stock Market Movement Direction with Support Vector Machine”, Computer and Operation Research Vol.32, pp.2513-2522.
[27]Indro, D.C. , Jiang C.-X. , Patuwo B.E. and Zhang G..-P. (1999), ”Predicting Mutual Fund Performance Using Artificial Neural Networks”, International Journal of Management Science. Vol.27, pp.378-380.
[28]Jensen, M.C. (1968), ”The Performance of Mutual Funds in the Period 1945-1964”, Journal of Finance, Vol.23, pp389-416.
[29]Lee, Y.-J., Hsieh W.-F., and Huang, C.-M. (2005), ”ε-SSVR: A Smooth Support Vector Machine for ε-insensitive Regression”, IEEE Transactions on Knowledge and Data Engineering ,Vol .17, pp.678-685.
[30]Moses, E.A., Cheyney, J.M. and Veit, E.T. (1987), ”A New and More Complete Performance Measure”, Journal of Portfolio Management, Vol.13, No.2, pp24-33.
[31]Sharpe, W.F. (1966), ”Mutual Funds Performance”, Journal of Business, Vol.39, No.1, pp.119-138.
[32]Trafalis, T.B. and Ince, H. (2000), ”Support Vector Machine for Regression and Applications to Financial Forecasting”, IEEE Computer Society, Vol.6, pp.348-353.
[33]Treynor, J.L. and Mazuy, K.K. (1966), ”Can Mutual Fund Outguess the Market?”, Harvard Business Review, Vol.14, pp.131-136.
[34]Treynor, J.L. (1965), ”How to Rate Management of Investment Funds”, Harvard Business Review, Vol.43,pp.63-75.
[35]Vapnik, V.N. (1995), ”The Nature of Statistical Learning Theory”, Springer-Verlag, New York.
[36]Vellido, A., Lisboa, P.J.G. (1999), ”Neural Networks in Business：A Survey of Applications (1992-1998)”, Expert System with Applications, Vol.51, pp.51-70.