NASDAQ最大漲幅投機型個股之日內 報酬率-買賣單不對稱關係

Abstract

股票市場中沒有資訊的交易人總是希望能尋找出一個可以反映資訊擁有者所握有的資訊的指標。若能成功找出這個指標，即使並非身為資訊擁有者，也能根據觀察資訊擁有者交易行為中的資訊意涵，據以發展出能使其獲利的交易策略。根據以往的研究，買賣單不對稱極可能是反映隱含私有資訊的一個重要指標。 本文的主旨即在探討NASDAQ 漲幅最大的各股其日內報酬率與買賣單不對稱之間是否真的存在顯著的正向關係。 在進行實證之前，我們先以GARCH (1,1)的模型去測試該模型是否能配適於我們所樣本中的時間序列資料。以了解本文中實證資料的屬性。經過測試後我們發現，對於超過七成的樣本GARCH (1,1)模型可成功描述其隨時間變動的特性。 而本文的主要實證結果為，在所使用的樣本---NASDAQ最大漲幅投機型個股日內資料中, 當期的買賣單不對稱的確與該期的股票報酬率有顯著的正向關係存在。然而前期的買賣單不對稱與當期報酬率間關係卻不顯著。 最後，由於公司資本規模一向被用作衡量該個股資訊不對稱程度的代理變數之一。本論文最後，我們將探討公司的資本規模與買賣單不對稱之間，是否有小型股效果存在，以作為日後選股的參考依據。然而，我們的實證結果僅發現，取過對數的資本市值與買賣單不對稱的係數之間確有負向關係存在，但是此負向關係並不顯著，因此無法就此結果對於小型股效果是否存在加以評論。

In this article, our main purpose is to explore the relationship between order imbalances and stock returns. Considering the liquidity and the efficiency of the market we research, we selected our samples from NASDAQ daily top gainers. We are curious about what factor makes these attractive top gainers win such remarkable positive returns in such a short time period. According to former studies, we decide that order imbalance of the individual stocks might be a proxy that reveals the insiders’ private information and thus it might be a critical factor in influencing stock returns. Before testing whether there are some special relationships exist between stock returns and order imbalances, we have to make judgments on the properties of our data. First we apply GARCH (1,1) model to see whether it can fit our time series data. In our research, we found that for over 70% of our sample stocks, the GARCH (1,1) model can successfully captured their time-variant proprieties. We also found that either by using the FORTRAN program or by using regression models, contemporaneous order imbalances both have a significant influence on the contemporaneous stock returns. However, the influence weakened when lagged order imbalances are employed. As for the test of small firm effect, we see a negative relationship between a firm’s logged market capitalization and its current order imbalance coefficients. However, the relationship is not significant.