標題: 具艾略特波浪特徵之股市趨勢研究
The Research on Taiwan Stock Trends Based on Elliott Wave Theory
作者: 陳儒琦
陳安斌
資訊管理研究所
關鍵字: 技術分析;艾略特波浪理論;倒傳遞類神經網路;Technical Analysis;Elliott Wave Theory;Back-Propagation Neural Network
公開日期: 2010
摘要: 本研究嘗試以波浪理論之特徵規則為依據,發現股價波動所隱含的變動規律。由於綜觀過去技術分析之研究,發現過去研究大多著重於技術指標之探討,對於以股價型態為基礎進行趨勢變動分析之研究則較少,且波段與趨勢皆建立於型態上。艾略特波浪理論即以型態為基礎,強調股價的波動具有一定的變動規律性,由五個上升波和三個下跌波之八波組合構成股市的完整週期,因此若能有效研判型態以掌握大趨勢,投資人可獲得較高的報酬並減低風險。 洪若信於2009年提出N型波浪技術指標,本研究將以其研究為基礎,改善N型波浪抓取方式使預測結果達到更高的準確率,同時修正模型參數,以期運用更少的變數提升模型整體之效率,使變數解釋能力更高。 本研究以1987年2月7日至2009年2月6日之台灣加權股價指數作為研究標的。本研究分為兩組實驗,實驗一為比較是否運用波浪特徵規則抓取N型波浪,結果顯示,應用艾略特波浪特徵規則於股價走勢之預測,無論是準確率或是獲利能力上均明顯擁有較佳的結果。實驗二為比較本研究與洪若信之研究方法的差異,結果顯示本研究僅以26個參數(洪之研究使用58個參數)即可有效描述N型波浪,提升模型整體之效率,使變數解釋能力更高。
This study attempt to find the stock price trend based on the rule-based wave theory. As we review technical analysis of the recent study, found that most research emphasize the technical indicators, but few used stock price patterns as the basis for trend analysis. Elliott Wave Theory is one of the most important theories which stresses the stock price variation with certain regularity and consists eight-wave combination to complete stock market cycle. Therefore, if the entire stock market trend could be effectively expressed by those waves combination, investors could get higher returns and lower risks as long as can catch the trend pattern. This study is based on Hung’s N-wave technical indicator(2009), and further tries to improve the accuracy by verifying the method of extracting N-shaped wave. In addition, we redefine the variables to improve the model’s efficiency, and then use fewer variables to achieve the better results. In this study, the data are collected from daily TAIEX between 7th Feb.1987 to 6th Feb. 2009. This study performed two experiments: the first experiment is to compare the method of extracting N-type waves between our method and Hung’s. The result shows that our method which employed Elliott Wave characteristic rules are better, both on accuracy and profitability. The second experiment is to compare the result of our methodology and Hung’s. We conclude that we only use 26 parameters (Hung used 58 parameters) can be describe the N-type wave, and accuracy and profitability in both of which have shown superior.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079734519
http://hdl.handle.net/11536/45484
顯示於類別:畢業論文


文件中的檔案:

  1. 451901.pdf