呂長霖

Chang-LinLu

碩士論文（2012）

基本分析為基之股票預測方法研發

Development of a Fundamental Analysis-based Method for Stock Market Forecasting

關鍵字 Keywords

基本分析、多目標決策分析、文字探勘、機器學習

Fundamental Analysis、Multiple Criteria Decision Making、Text Mining、Machine Learning

摘要

股票投資是一高風險之活動，若股票投資者對影響股價變動因素考慮不周及無良好的專業知識與經驗而導致決策錯誤，將會造成血本無歸的情形，因此，為使投資者獲利，市場上有許多方法以預測股票未來走勢，但因為股票市場的不確定性，預測結果常常與實際走勢有落差，因此如何準確的預測個股股價未來走勢一直是股票市場中研究的議題之一。

本研究主要目的為研發一套以基本分析為基之股票預測方法，將個股財務狀況、產業環境、總體經濟與財經新聞做為考量因素，從股票市場中選出體質較佳之個股並預測其股價未來趨勢，最後將預測結果提供給投資者以做為投資決策時之參考。針對上述目的本研究主要研究項目包括：(i) 基本分析為基之股票預測流程設計，(ii) 基本分析為基之股票預測方法發展以及(iii) 基本分析為基之股票預測方法驗證與比較。其中，基本分析為基之股票預測方法包括個股財務指標權重計算方法、個股投資評估方法、財經新聞特徵選取方法、財經新聞預測方法與個股股價預測方法。

Abstract

Stock investment is a high-risk activity, it will have large losses by wrong decisions if investors ill-considered the impact of stock or haven’t professional knowledge and experience. Therefore, it’s many methods to forecast the stock price trend in order to have more profit for investors in stock market. However, it’s often have gap with actual stock price trend. For this reason it has been one of the topics of research that how to accurately forecast stock price trend in the stock market.

Our research propose is development of a fundamental analysis based method for stock market forecasting, considering the influencing factors that including stock financial indicators, environments, macroeconomics and financial news to select more betters’ stock and prediction the stock price trend. Our research topic including: (1) Design of a fundamental analysis-based stock prediction method procedures; (2) Development of a fundamental analysis-based stock prediction method procedures; (3) Measuring and comparison of a fundamental analysis-based stock prediction method procedures.