金融工程學行為實證之研究-以艾略特波浪理論隱含波浪行為知識的發現為例

Abstract

台灣的股票市場是個淺碟型市場，股市的短期走勢易大幅波動，不容易掌握中長期趨勢。Fama 提出股市是沒有行為的效率市場假說；過去很多學者研究以歷史成交量、成交價為主的技術指標分析未來的走勢，證明股市是有行為的，但是卻沒有指出為何會有這些行為。 本研究嘗試以物理力量行為分析股市多空趨勢消長的情形，探討艾略特波浪圖形力量的存在。以艾略特波浪理論為基礎，提出一個N型波浪形狀的技術指標，利用N型波浪的斜率、位能、夾角、長度、時間的物理力量建構符合泰勒級數精神的倒傳遞類神經網路對台灣股票市場中期趨勢加以探討。 由實證結果得知，以本研究所提出的N型波浪的物理力量結合倒傳遞類神經網路來預測漲股市走勢，經過實驗與統計分析，顯著的大於隨機預測的準確度和投資績效。換句話說，驗證了股市物理力量的存在，而且和股市的走勢有相當程度的關聯，同時也驗證艾略特的波浪行為是有意義的。所提出的N型波浪指標也可以提供給投資人及專業人員一個中長期的參考技術指標。

Taiwan stock market is a shallow-plate market with high volatility, therefore it is difficult to forecast the market trend for a long period of time. While Fama believes behavioral finance is not to replace the efficient market theory, supporters of behavioral finance say that evidence of using technical analysis based on trade volume and price signals an inefficient market. This study investigates whether there is a physical strength of the Elliott wave principle. Based on the Elliott wave principle, we propose an N-shaped wave technical index, and use the physical strength of slope, potential energy, angle, length, and time to construct back-propagation neural networks, and forecast Taiwan stock market trend. The result shows that the proposed N-shaped wave technical index for back-propagation neural networks model is remarkable than random forecasting. Therefore reveal that the physical strength of stock market exists, and Elliott wave principle can be applied on market trend forecasting.