個別股票的隨機指標(KD)參數制定與買賣策略的選擇

Abstract

在股票市場上，投資人常以淨報酬率來做為評量投資績效的依據，而技術指標通常是作為買賣進出的輔助工具，以期獲得最大報酬率。但是股票投資是一種高報酬高風險的投資行為，所以顯然的單一以報酬率的高低做為買賣操作方法選擇是無法規避高度風險的。 在本文中，我們採用KD指標來模擬運算做為買賣進出的依據指標，其KD周期則採用不同天數(5天、6天、及9天)、週數(4週、6週、及9週)來製作，共有6種。另外在執行買賣的策略上, 我們也提出α、β與γ，3種買賣策略，經交叉運用共產生十八種買賣操作方法。 為判別各買賣操作方法之優劣良窳，我們在台北股市中挑選5檔各產業的龍頭股，分別是台積電、聯電、台塑、中鋼、及國泰金控，以各組買賣操作方法套入歷史資料(2001/1/2~2004/12/31) 來模擬計算並各取得3項績效指標，分別是交易成本、報酬率之變異數、及淨報酬率。至於應如何決定3項績效指標之權重來產生最佳之買賣操作方法呢? 在本研究中採用資料包絡分析法(Data Envelopment Analysis)做為評比各組買賣操作方法績效的理論與工具，資料包絡分析法可以求得各組買賣操作方法在十八組中之相對績效，亦即決定各組最佳績效指標之權重。再以各組最佳績效值來評比各組之績效，排列出十八組買賣操作方法之優劣名次。

It is popular for people to invest in stocks even though it may have high return and high risk respect to bonds and deposit. Individuals often lose money in stock investment since they tend to focus on high return, but ignore the high risk and cost behind the investment. In the stock market, individuals and portfolio managers employ variety analysis techniques to decide buying/selling for short-term investment. The Stochastic Oscillator (KD) is one of the popular analysis techniques. We use the past 991 days’ (or 206 weeks’) opening, closing, the highest, and the lowest prices of five major stocks in Taiwan to generate six types KD curves that are based upon different moving average time intervals. In this research, we consider three selling/buying strategies. We assess the performance of the eighteen operation methods (OMs), the compositions of six KD curves and three strategies, by Data Envelopment Analysis (DEA). Three indices are used: transaction cost, return rate variance, and total return rate. We observed one of the proposed strategies outperforms the others. Depends on the market characteristics of a stock, a particular OM may outperform the others. One would observe the interaction effects between time intervals for generating the KD curves and the selling/buying strategies.