隔夜效應之發現於台灣股價指數期貨之研究

Abstract

股市是數以千計的市場投資者買賣供需行為的結果，而每位投資者所做的買賣決策都受到各種因素的影響，例如基本面、總體經濟面及消息面等，即市場所有的資訊都會反映在價格上。這些資訊的影響都是連續不斷的，然而交易時間的限制使得股價在時間軸上為不連續的區間。非交易時間的資訊累積無法立即反映在股價的變動上，投資者只能在隔日開盤後針對非交易時間所獲得的資訊進行持股部位的調整，這是為何開盤時的交易活動較為積極的原因之一。然而開盤期間的股價的變化並沒有受到應有的重視，開盤後前十分鐘內的交易活動往往不在被許多日內交易的研究樣本中。本研究針對此隔夜效應現象進行分析，研究結果證實隔夜效應對日內交易的顯著性。 透過股價指數期貨的實驗，證實了隔夜效應的顯著性之後，本研究嘗試利用類神經網路對日內交易進行模擬，模擬結果再度肯定了隔夜效應的重要性，也因此兩個交易日之間所累積的資訊同時反映在開盤時的現象被肯定為對日內交易具價值性。

The term, overnight effect, originates from the restriction on trading hours. The influences to stock prices such as fundamental factors and macroeconomic factors are all continuous. However, due to restriction on trading hours, the generation function of stock prices is discontinuous in time-axis. Various information arrived during overnight period, but could not be responded immediately. Short-term investors could adjust their holding stocks in terms of the overnight information only after opening of the next trading day. Therefore, the influences during non-trading period would only reflect on stock prices at the opening period of the next trading day. This is also the reason why stock returns at beginning trading period are usually with high volatilities and volumes. By experiments on stock index futures, this paper firstly investigates the phenomenon between two trading hours. After the significance of overnight effect is confirmed, the observed overnight effect would be used to analyze the following intraday trading activities by neural network. This paper concludes that information during non-cash period is valuable and should be studied.