臺灣博碩士論文加值系統

　　期貨市場從民國86年開始至今已超過10年，市場上也充斥著各種投資方法與指標，目的就是要讓投資人可以打敗效率市場而獲得超額報酬。期貨擁有高槓桿的特性,若能找出合適交易方法,便能以小搏大獲取極高的利潤，但期貨的價格波動比起股票來的快速，往往另投資人難以捉摸。
　　過去的方法多利用技術分析的指標，再配合類神經網路、基因演算法或統計方法等，進行參數最佳化的演算，此法雖然可行，但可能會陷入過度最佳化的陷阱，或者經過一段時間後，最佳化參數需因時調整等問題。本研究試著用型態探勘(Pattern Mining)方法，針對台指期貨的價格走勢做型態上的拆解，以N字型態為基礎，跳脫傳統M頭、W底、旗型等慣用技術分析型態，改由更基本的N字型態做為分析主軸，藉以找出可獲利的操作模式。本文係運用技術分析裡的K線高低點相連產生的N字型態，以N字結尾端點突破或跌破N字的轉折點做為買入及賣出訊號，再搭配不同週期、停損與資金管控等不同策略，針對2011~2014年台指期貨歷史資料實施回溯測試，以最佳累積報酬率當作擇優條件，作出台股指數期貨當沖交易決策。
　　本研究提出的方法除了不須隨調整最佳化參數，在實證結果也顯示：運用N字型態搭配其它交易策略，應用於台指期貨當沖交易4個年度，其所產生的累計損益均為正報酬。顯示以N字型態為基礎的方法，產生的交易規則，作為執行期貨日內交易的買賣策略是可行的。

Taiwan’s futures market has been running for more than 10 years since 1997. So far, numerous investment methods and indices have been developed, all for the purpose of helping investors beat the efficient market to gain excess returns. Futures trading is characterized by high leverage, meaning that with a good trading strategy, one is likely to gain high returns with a small investment. However, it is also characterized by faster price fluctuation and lower predictability compared to stocks trading.
Previous research has proposed methods involving use of neural network, genetic algorithm or other statistical analyses along with indices of technical analysis to compute optimal parameters for investment. These methods are feasible but may lead to over-fitting or require time-varying adjustments. In this study, a pattern mining method was proposed. Instead of the traditional M-head, W-bottom or flag patterns, this method analyzes N patterns, created by linking the candlesticks, in the trends of Taiwan Futures Exchange (TAIFEX) prices to find profitable ways of futures trading. This method views the breakthrough beyond the ending tip of an N pattern as a signal for buying and the turning point after a slump outside the N pattern as a signal for selling. The effectiveness of this method combined with different investment strategies was also tested using historic data of TAIFEX during 2011~2014. In this test, the best cumulative return was used as the optimal condition for day-trading.
Empirical evidence showed that the proposed N-pattern method combined with other investment strategies can generate positive cumulative returns from day-trading across all the four years. In other words, the trading principles provided by the proposed N-pattern method can be feasible strategies for day-trading in TAIFEX.

摘要 i
Abstact ii
目錄 iii
表目錄 v
圖目錄 vi
第１章 緒論 1
1.1 研究背景 1
1.2 研究動機與目的 1
1.3 主要貢獻 3
1.4 論文架構 4
第２章 文獻探討 5
2.1 台灣股價指數期貨簡介 5
2.1.1 契約規格與交易量 5
2.1.2 交易操作模式 7
2.2 技術分析 7
2.2.1 技術分析概述 7
2.2.2 技術分析方法 8
2.2.3 型態分析法 9
2.3 相關文獻比較 13
第３章 系統設計與演算法 15
3.1 系統架構 15
3.1.1 資料預處裡 15
3.1.3 模擬交易流程 17
3.1.3 績效分析 18
3.2 N字處理流程 19
3.2.1 K線高低點線段連結 19
3.2.2 連接點合併及N字型態截取 23
3.3 交易策略與操作方式 25
3.3.1 N字分類方法 25
3.3.2 交易策略 26
3.3.3 交易操作方式 30
3.3.4 原始交易策略之演化 31
第４章 實驗結果分析 32
4.1 資料集 32
4.2 原始N字策略之績效分析 33
4.3 長短K線週期之績效分析 35
4.4 停損策略績效分析 39
4.5 使用資金管控之績效分析 40
第５章 結論與建議 42
5.1 結論 42
5.2 後續研究建議 42
參考文獻 45

[1]陳蕾如,”K線與型態辨識函數之開發及技術分析之應用”,義守大學財務金融學系,碩士論文,2003.
[2]高偉展,”開盤八法應用於台灣期貨市場之實證研究”,朝陽科技大學財務金融系,碩士論文,2004.
[3]張曉青,”建構智慧型線段切割法於股價買賣點之預測”, 元智大學工業工程與管理研究所, 碩士論文,2007
[4]陳佳榮,”型態辨識應用於技術分析指標之研究”,中山大學財務管理學系研究所,碩士論文,2011.
[5]王瑞東,”以技術圖樣辨識系統建構股票買賣策略之研究”,臺中技術學院資訊科技與應用研究所,碩士論文,2008.
[6]潘靜怡,”以技術分析W底型態建構台指期交易策略之實證研究”,朝陽科技大學財務金融系,碩士論文,2011.
[7]周俊志,”自動交易系統與策略評價之研究”,臺灣大學電機資訊學院資訊工程學系,碩士論文,2007.
[8]莊政文,”以獲勝率為導向的最佳化技術交易規則之研究—以台股期貨為交易標的”,朝陽科技大學資訊管理系,碩士論文,2007.
[9]李建和,”RSI技術指標於台灣指數期貨市場價差交易之實證研究”,高雄應用科技大學金融資訊研究所,碩士論文,2007.
[10]謝正一,”應用技術指標與市場輪廓理論於臺指期貨轉折點之行為發現”,交通大學管理學院資訊管理系,碩士論文,2014.
[11]林慶茂,”順勢交易策略應用於台灣加權指數期貨之實證研究”,中興大學高階經理人碩士在職專班,碩士論文,2005.
[12]施景章,”臺股期貨均線突破策略高頻交易之實證研究”,高雄應用科技大學金融資訊研究所,碩士論文,2011.
[13]黃培琳,”程式交易策略、期貨交易和法人交易之實證研究”,碩士論文,2011.
[14]陳柏淦,”應用三層濾網交易策略於台指期貨市場之研究”,世新大學資訊管理學研究所,碩士論文,2014.
[15]A.Cowles,”Can stock market forcasters forcast?,” ECONOMETRICA, vol. 1,Issue 3, Jul. 1933
[16] H.I Bishara, “Establishing a modified Dow Theory for Canada and testing its buy and sell signals against a buy and hold strategy,” Akron Business and Economic Review, pp.43-51, Winter 1977.
[17] D.A. Glickstein and R. E. Wubbels , “Dow Theory is alive and well,” Journal of Portfolio Management, vol. 9, No. 3: pp. 28-32, Spring 1983.
[18] Saitta, Alex (1995), “Using Dow Theory to catch trends,” Futures: News, Analysis & Strategies for Futures, Options & Derivatives Traders, vol. 24 Issue 11, p46, Oct. 1995.
[19] S.J. Brown, W.N. Goetzmann and A. Kumar , “The Dow Theory: William Peter Hamilton’s track record reconsidered,” The Journal of Finance, vol. 53, Issue 4, pages 1311–1333, Aug. 1998.
[20] G. Caginalp, H. Laurent, “The predictive power of price patterns,” Applied Mathematical Finance, vol. 5, Issue 3-4, 1998.
[21]Jar-Long Wang, Shu-Hui Chan, “Stock market trading rule discovery using pattern recognition and technical analysis,” Expert Systems with Applications , vol. 33, Issue2, p304–315, Aug. 2007.
[22]Chun-Hung Hsieh, Rei-Tong Wang, “Using bull flag template match to gain a higher profit from Taiwan stock exchange market,” The Sixth International Conference on Information and Management Sciences, 2007.
[23]Tsung-Hsun Lu, ”The profitability of candlestick charting in the Taiwan stock market,” Pacific-Basin Finance Journal ,vol. 26, pp. 65–78, Jan. 2014.
[24]Tai-Liang Chen,”Forecasting the Taiwan stock market with a stock trend recognition model based on the characteristic matrix of a bull market,” African Journal of Business Management , vol. 5(23), pp. 9947-9960, Oct. 2011.
[25]M. Dong, X.S. Zhou, “Exploring the fuzzy nature of technical patterns of US stock market,” in Proc. of the Fuzzy System and Knowledge Discovery, Singapore, pp. 324–328, 2002.
[26]R. D.Edwards、J. Magee、W.H.C.Bassetti, Technical Analysis of Stock Trends,9thed, AMACOM, 2007
[27] “艾略特波浪理論”,[Online]. Available:
http://wiki.mbalib.com/zh-tw/%E8%89%BE%E7%95%A5%E7%89%B9%E6%B3%A2%E6%B5%AA%E7%90%86%E8%AE%BA[Accessed 5 May 2015].
[28] “道氏理論”,[Online] Available:
http://wiki.mbalib.com/zh-tw/%E9%81%93%E6%B0%8F%E7%90%86%E8%AE%BA [Accessed 5 May 2015]