標題: 智慧型決策系統運用於台灣股票市場技術面分析之研究
The Study of the Application of Intelligent-Oriented Decision Systems in the Taiwan Stock Market Technical-Analysis
作者: 岑英勤
Ing Chin Tesrn
陳安斌先生
Dr. An-Pin Chen
資訊管理研究所
關鍵字: 智慧型決策系統;技術分析;類神經網路;模糊邏輯;intelligent-oriented decision system; technical analysis;neural network;fuzzy logic
公開日期: 1992
摘要: 本研究針對股市投資的需求,透過電腦有智慧地處理各種技術分析的方法 ,對明日的股價及未來的趨勢作分析與預測,以提供投資決策之用。本研 究所建立之智慧型決策系統依實務界所常使用的一些技術分析方法的特性 ,分別建立了三個主要的模組。第一為K線預測模組,K線為最普遍、基 本的技術分析法。將股市的交易資料轉換成K線的形式後,再配合各類技 術指標,予以類神經網路學習,以對明日的股價作預測;第二為型態辨識 模組,連續K線所形成的某些特殊的股價型態,可提供股價未的走勢的參 考,為一重要的買賣點依據。本模組對各種型態的整理期擷取其特徵屬性 ,予以類神經網路辨認之;第三為模糊邏輯決策模組,為了對各種型態的 反轉期趨勢作研判,整合各種模糊性的市場資訊,透過模糊類神經網路的 運作,可歸納出一些模糊邏輯的規則,作為判斷趨勢的依據。本研究之智 慧型決策系統,透過類神經網路的學習,不需倚賴專家的知識,可自行至 歷史資料中學習各類技術分析的知識。最後並整合了模糊邏輯決策推論, 仿專家思維的方式將知識作高層的推論。目前各模組的測試已獲初步的成 效,未來將可對各種投資之智慧型決策系統的設計提供一研究的方向。 The purpose of this study is to develop an intelligent- oriented decision system(IODS) to predict future stock price and the trend of the price movement. The IODS can support investors to analyze stock market data from the technical analysis's point of view. IODS contains three main modules. The first is bar chart prediction module. It can translate stock trading data into bar chart formation with several technical indices, then it can predict next day bar chart by using neural network package tool. The second is stock price pattern recognition module. It can abstract the features of a price pattern and recognize the consolidation period by neural network tool again. The third is a fuzzy logic decision module, a fuzzy neural network. It can determine the stock price trend by recognizing the reversal period from a period of stock price pattern. Furthermore, in this study the IODS can learn the knowledge from the historical trading data and integrate fuzzy logic to do the knowledge inference processes. Finally, the study provides a guideline for the coming researchers who may develope such an intelligent integration system to help investors to make better decisions.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT810396021
http://hdl.handle.net/11536/56838
顯示於類別:畢業論文