Title: 模糊集合理論在台灣股價趨勢分析的應用
The Application of Fuzzy Set Theory to the Trend Analysis of Taiwan Stock Price
Authors: 郭張成
Chang Cheng Kuo
An Pin Chen
Keywords: 聚類分析;型態識別;模糊集合;隸屬度函數;cluster analysis;pattern recognition;fuzzy set; membership function
Issue Date: 1992
Abstract: 基本上,股票市場上的股價趨勢可概略分成看漲行情與看跌行情兩種現象 ,但它們之間的區別界線在劃分上存有模糊性及不確定性。本研究嘗試利 用各種技術指標以表現台灣股價趨勢的特徵,並藉由各種聚類方法以說明 股價趨勢型態的存在,再透過統計上的解釋以辨識異常交易現象。然後再 以模糊聚類分析方法所得到隸屬度函數來表現此種帶有模糊性質的股價趨 勢。最後,藉由模糊型態識別方法以賦予電腦模糊思維的判斷能力,並使 電腦能依此模糊概念來識別股價趨勢。本研究的結果顯示,應用統計方法 可辨識出異常交易現象,並解釋股價趨勢型態的存在,不過其間的差異並 非十分明確。這也說明應用模糊理論表達此種界線不清的股價趨勢較為合 理。最後,本研究亦提出一整合性的股市分析系統模型架構,再運用電腦 以此架構將股市的資料轉換成資訊、知識以分析股價趨勢。這將可減輕主 觀意識上的判斷,使分析的結果更為客觀。經由測試的結果顯示,系統在 股價趨勢的識別能力方面有不錯的效果。 This study tries to point out the trend characteristics of Taiwan stock price in the following aspects, such as various Technical Indexes, the existence of trend patterns of stock price by various cluster analysis, and the abnormal transaction phenomenon by means of statistical explanation. Thereafter, the stock price trend being analyzed is redefined by the membership function resulted from fuzzy cluster analysis. Finally, a fuzzy- based forecasting system is implemented to predict the stock price trend. In this paper, some conclusions and discussions are shown as following, applying the statistical methods can identify the abnormal transaction phenomenon, and can explain the existence of trend patterns of stock price. But these trend patterns can not be distinquished very significantly. For this reason, applying the fuzzy set theory is much better to represent the fuzzy trend patterns of stock price. Finally, this study proposes an integrated prototype system which can sequentially transfer stock market data into information and analyze the stock price trend. In consequence, this approach will reduce the subjective judgement in consciousness, and result the analysis more objec- tive. Besides, the performance test of this system is shown having a good recognition ability on stock price trend.
Appears in Collections:Thesis