期貨的價格發現功能來自於市場參與者透過買賣來表達對於未來商品價格的看法，故價格發現功能應受參與者的數量與屬性影響，本研究以基差與未平倉量來建構方向性變數與隱含能量變數，探討其與未來現貨價格行進方向和速度之關聯性。並以摩根史坦利資本國際公司(Morgan Stanley Capital International Indices, MSCI)所編製的摩根史坦利台灣加權指數期貨作為研究對象。樣本期間為西元1997年1月10日至西元2013年11月17日，共4139筆日資料。 本研究以多種天數、不同速度建立多個羅吉斯迴歸模型，其實證結果如下：(1)模型預測發生且實際發生之比率皆高於五成，故支持價格發現功能存在之論述。(2)未來五日模型預測發生且實際發生之比率皆優於其他的天數。(3)當方向性變數單獨存在時，會出現不顯著或反指標的情況，但若與隱含能量變數交乘，就會出現顯著正相關的影響。 Price discovery of futures comes from future spot prices which were proposed by market participants through trading. So, the function of price discovery is affected by the quantities and attributes of participants. This research constructs directional variables and implicit energy variables by basis and open interest to discuss the associations among the variables, the direction and the speed of future spot prices. So this research takes Taiwan index futures, which is established by Morgan Stanley Capital International Indices (MSCI), for the study. The period is taken from 10 January 1997 to 17 November 2013, a total of 4139 daily data. This research constructs multiple logistic regression models by a wide variety of days and different speed. The empirical results are as follows:(1) The actual occurrences over forecast occurrences ratio of models are larger than 50%. Therefore, the results support the existence of the price discovery of futures.(2) The actual occurrences over forecast occurrences ratio of the models forecasting for the next 5 days is the best.(3) The coefficients of directional variables, Frequency Basis (FB) and Relative Basis (RB), are, insignificant. However, if they merge with implicit energy variables, there will be a significant positive correlation.
