運用市場輪廓理論與分群分析之發現低價股票的投資價值

Abstract

綜觀市場上的投資證券學理上，在證券投資的評價上都以基本面及財務面健全與否來建議投資者做為選股依據，即所謂的績優股。常見的評等標準多為財務指標，如：資產負債表的評價；損益表的獲利能力；公司配股狀況分析的股息折現模型．．．等。 但，股票市場的價格為買賣雙方競合的結果，其反應為投資人對公司未來的看法(看漲或看跌)，然而，待公司發佈財務報表時多已為歷史資訊的落後指標，與即時股價的表現往往大相逕庭。 本論文以物理現象角度研究，根據Steidlmayer之市場輪廍理論下，股價變化乃由買賣雙方力量來推使價格的移動方向為向上或向下，直至雙方多數達一致價格時，股價則相對平穩。 市場輪廓理論，市場價格的競價方式有兩種:一為向下競價，即多個參與者想要賣；二為向上競價，多個參與者想要買。價格滯留於某一時間越長，表示眾人越認同，即為價值區間（Value Area）。 本論文即以市場輪廓之理論基礎，試以利用自組織映射圖神經網路的分群技術，祈能發現投資低價股票之報酬率可有高於高價股票的投資價值。

Many investment theories suggest investor to buy blue chip stocks, and suggest buying the stocks based on the fundamentals and financial analysis; such as evaluating company’s balance sheet, profitability of the income statement, and last but not least, share placement situation analysis of the discount model. The price of a stock reflects the buyers and sellers future expectation toward the company. Therefore, it is too late to invest if investors wait for the financial reports of the company because it is the result of both buyers and sellers future expectation. In fact, real-time stock price performance is often quite different than we expected. According to the Market Profile Theory by Steidlmayer, the price fluctuates as a result of the buying and selling force of market participants. The price fluctuation diminished as both sellers and buyers reached a majority consensus price resulting in relatively stable stock price. Market Profile Theory stated that market price auctions in two ways: downward bidding and upward bidding; downward when sellers dominate the market, upward when buyers dominate the market. Value Area is the range of price where participants are more likely to trade in. This thesis implements Market Profile and Self-Organizing Map Neural Network Clustering technique to discover the investment value of low price stock.