應用市場輪廓理論於台灣期貨指數極短線走勢預測之研究

Abstract

本研究主要以市場輪廓(Market Profile)理論為基礎，將傳統TPO建構而成的市場輪廓圖(TPO chart)應用在比一般傳統更短的時間區間內，並且搭配倒傳遞類神經網路，以技術指標及市場輪廓指標作為輸入變數，期望透過其物理力量的總和判斷，從中找出市場邏輯以及市場結構變化的行為知識。

由實驗結果得知，將市場輪廓應用在更小的時間區間(15秒鐘K線)時，其整體獲利能力優於技術分析指標，而與應用在較長的時間區間(1分鐘K線)上相比時，每口獲利也獲得了明顯的改善，但準確率卻有所下降，證明了市場輪廓在極短線市場上，仍然對於市場的趨勢預測上擁有相當程度的能力。

The research is based on theory of Market Profile, applying TPO chart which is composed of TPO to a shorter time period. The research adopts Back Propagation Neural Network, using technical analysis index and Market Profile index as input data. We attempt to extract the knowledge of variation of market logic and market structure by judgement of their physical strength.

According to the results, we find that applying Market Profile to a shorter time period (15 seconds candle line) can have a stronger profitability than using technical analysis index. Moreover, compared to applying Market Profile to a longer time period (1minute candle line), we have a better profit each contract but worse accuracy using a shorter time period. From our experiment, it shows that when we apply Market Profile to a short-term time period, it still has a certain degree of ability to predict trend of market.