以技術指標量化市場輪廓理論的行為研究

Abstract

J.Peter Steidlmayer的市場輪廓理論(Market Profile)能充分掌握交易者的心理行為，很明確地解析出投資者在市場上的心理行為變化，經過多位專家實證其理論預測市場漲跌趨勢的有效性，但這些定性的市場行為知識以往需要藉由多年交易經驗的投資者才能解釋，況且近年來程式交易普遍化的過程中，如何把能夠掌握投資者心理行為知識的市場輪廓理論充分應用於電腦上，藉由資訊技術的輔助進行交易，不需藉助人為主觀的判定，以達到獲取最大報酬的目的，量化行為知識的方法是一個重要議題。 故本研究嘗試以技術指標的數學公式量化市場輪廓理論之行為知識，以倒傳遞類神經網路架構展示市場輪廓理論量化行為知識之可行性。經由本研究實證市場輪廓量化指標的類神經網路模型，較技術指標更能有效提升預測市場漲跌趨勢的準確度，證實量化指標之可行性

The "Market Profile" theory developed by J. Peter Steidlmayer has been used by investors to understand the psychological and behavioral responses of stock market traders in reaction to changes in the market. Previous studies have proved that the theory is able to effectively predict market trends. Moreover, trader behavior is qualitative market knowledge and requires many years of trading experience by investors in order to explain it. In recent years, trading software has become commonly available, so the ability to integrate market investor psychology theories with the full knowledge of trader behavior, assisted by information technology transactions, removes the need for subjective judgment. In order for investors to achieve the maximum profit, the method of quantifying trader behavior is an important issue.

This study attempts to use the mathematical formulas of technical indicators to quantify the theoretical knowledge of market behavior, based on a back-propagation neural network architecture that shows the theoretical feasibility of quantifying market behavior. This study illustrates that analyzing the empirical market through quantitative indicators based on a neural network model is more effective in enhancing the accuracy of forecasting market fluctuations than by using technical indicators, and, therefore, the feasibility of quantifying the behavior is confirmed.